diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10025 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 12.74274937560528,
+  "global_step": 500000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 0.8907,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 0.7216,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 0.9141247272491455,
+      "eval_runtime": 1.2377,
+      "eval_samples_per_second": 807.918,
+      "eval_steps_per_second": 12.927,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 0.7122,
+      "step": 1500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 0.7114,
+      "step": 2000
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 0.9156034588813782,
+      "eval_runtime": 1.1631,
+      "eval_samples_per_second": 859.777,
+      "eval_steps_per_second": 13.756,
+      "step": 2000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 0.7111,
+      "step": 2500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 0.711,
+      "step": 3000
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 0.915696918964386,
+      "eval_runtime": 1.149,
+      "eval_samples_per_second": 870.294,
+      "eval_steps_per_second": 13.925,
+      "step": 3000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.2e-05,
+      "loss": 0.7107,
+      "step": 3500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 0.7102,
+      "step": 4000
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 0.8820520043373108,
+      "eval_runtime": 1.2331,
+      "eval_samples_per_second": 810.943,
+      "eval_steps_per_second": 12.975,
+      "step": 4000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 0.7096,
+      "step": 4500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 0.709,
+      "step": 5000
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 0.8773962259292603,
+      "eval_runtime": 1.22,
+      "eval_samples_per_second": 819.664,
+      "eval_steps_per_second": 13.115,
+      "step": 5000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 0.7087,
+      "step": 5500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 0.7083,
+      "step": 6000
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 0.8944988250732422,
+      "eval_runtime": 1.1878,
+      "eval_samples_per_second": 841.893,
+      "eval_steps_per_second": 13.47,
+      "step": 6000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.8e-05,
+      "loss": 0.6868,
+      "step": 6500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.4e-05,
+      "loss": 0.6516,
+      "step": 7000
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 0.8944807648658752,
+      "eval_runtime": 1.2213,
+      "eval_samples_per_second": 818.801,
+      "eval_steps_per_second": 13.101,
+      "step": 7000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 0.6267,
+      "step": 7500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 0.6042,
+      "step": 8000
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 0.8855597972869873,
+      "eval_runtime": 1.1662,
+      "eval_samples_per_second": 857.462,
+      "eval_steps_per_second": 13.719,
+      "step": 8000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000102,
+      "loss": 0.5883,
+      "step": 8500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 0.5732,
+      "step": 9000
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 0.8907042741775513,
+      "eval_runtime": 1.2023,
+      "eval_samples_per_second": 831.767,
+      "eval_steps_per_second": 13.308,
+      "step": 9000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 0.559,
+      "step": 9500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 0.5506,
+      "step": 10000
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 0.8921875953674316,
+      "eval_runtime": 1.197,
+      "eval_samples_per_second": 835.455,
+      "eval_steps_per_second": 13.367,
+      "step": 10000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 0.5444,
+      "step": 10500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 0.5385,
+      "step": 11000
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 0.8903993964195251,
+      "eval_runtime": 1.1996,
+      "eval_samples_per_second": 833.63,
+      "eval_steps_per_second": 13.338,
+      "step": 11000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000138,
+      "loss": 0.532,
+      "step": 11500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 0.5276,
+      "step": 12000
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 0.8864779472351074,
+      "eval_runtime": 1.1846,
+      "eval_samples_per_second": 844.173,
+      "eval_steps_per_second": 13.507,
+      "step": 12000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015,
+      "loss": 0.5218,
+      "step": 12500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000156,
+      "loss": 0.517,
+      "step": 13000
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 0.88374263048172,
+      "eval_runtime": 1.1943,
+      "eval_samples_per_second": 837.343,
+      "eval_steps_per_second": 13.397,
+      "step": 13000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000162,
+      "loss": 0.5124,
+      "step": 13500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000168,
+      "loss": 0.5077,
+      "step": 14000
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 0.8863275647163391,
+      "eval_runtime": 1.1844,
+      "eval_samples_per_second": 844.316,
+      "eval_steps_per_second": 13.509,
+      "step": 14000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 0.5027,
+      "step": 14500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 0.498,
+      "step": 15000
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 0.8769687414169312,
+      "eval_runtime": 1.1484,
+      "eval_samples_per_second": 870.806,
+      "eval_steps_per_second": 13.933,
+      "step": 15000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000186,
+      "loss": 0.4938,
+      "step": 15500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 0.4897,
+      "step": 16000
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 0.8793530464172363,
+      "eval_runtime": 1.2247,
+      "eval_samples_per_second": 816.516,
+      "eval_steps_per_second": 13.064,
+      "step": 16000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000198,
+      "loss": 0.4849,
+      "step": 16500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000204,
+      "loss": 0.4791,
+      "step": 17000
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 0.879580557346344,
+      "eval_runtime": 1.1672,
+      "eval_samples_per_second": 856.75,
+      "eval_steps_per_second": 13.708,
+      "step": 17000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 0.4744,
+      "step": 17500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 0.4698,
+      "step": 18000
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 0.8753651976585388,
+      "eval_runtime": 1.1516,
+      "eval_samples_per_second": 868.392,
+      "eval_steps_per_second": 13.894,
+      "step": 18000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 0.4644,
+      "step": 18500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 0.4592,
+      "step": 19000
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 0.8825145363807678,
+      "eval_runtime": 1.2668,
+      "eval_samples_per_second": 789.404,
+      "eval_steps_per_second": 12.63,
+      "step": 19000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000234,
+      "loss": 0.4542,
+      "step": 19500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 0.4489,
+      "step": 20000
+    },
+    {
+      "epoch": 0.51,
+      "eval_loss": 0.878689169883728,
+      "eval_runtime": 1.1589,
+      "eval_samples_per_second": 862.885,
+      "eval_steps_per_second": 13.806,
+      "step": 20000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 0.4437,
+      "step": 20500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 0.439,
+      "step": 21000
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 0.8742354512214661,
+      "eval_runtime": 1.1831,
+      "eval_samples_per_second": 845.245,
+      "eval_steps_per_second": 13.524,
+      "step": 21000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000258,
+      "loss": 0.4339,
+      "step": 21500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 0.4292,
+      "step": 22000
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 0.8848758935928345,
+      "eval_runtime": 1.1886,
+      "eval_samples_per_second": 841.313,
+      "eval_steps_per_second": 13.461,
+      "step": 22000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00027,
+      "loss": 0.4251,
+      "step": 22500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000276,
+      "loss": 0.4212,
+      "step": 23000
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 0.8841528296470642,
+      "eval_runtime": 1.2211,
+      "eval_samples_per_second": 818.95,
+      "eval_steps_per_second": 13.103,
+      "step": 23000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 0.4175,
+      "step": 23500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 0.4142,
+      "step": 24000
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 0.8811973929405212,
+      "eval_runtime": 1.275,
+      "eval_samples_per_second": 784.285,
+      "eval_steps_per_second": 12.549,
+      "step": 24000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000294,
+      "loss": 0.4109,
+      "step": 24500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003,
+      "loss": 0.4076,
+      "step": 25000
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 0.8659318089485168,
+      "eval_runtime": 1.2881,
+      "eval_samples_per_second": 776.338,
+      "eval_steps_per_second": 12.421,
+      "step": 25000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029999920715161553,
+      "loss": 0.4045,
+      "step": 25500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002999968286151326,
+      "loss": 0.4017,
+      "step": 26000
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 0.874371349811554,
+      "eval_runtime": 1.2282,
+      "eval_samples_per_second": 814.215,
+      "eval_steps_per_second": 13.027,
+      "step": 26000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002999928644165624,
+      "loss": 0.3987,
+      "step": 26500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002999873145992569,
+      "loss": 0.3958,
+      "step": 27000
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 0.8821514844894409,
+      "eval_runtime": 1.1414,
+      "eval_samples_per_second": 876.15,
+      "eval_steps_per_second": 14.018,
+      "step": 27000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000299980179223908,
+      "loss": 0.3933,
+      "step": 27500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002999714583685469,
+      "loss": 0.3907,
+      "step": 28000
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 0.8762433528900146,
+      "eval_runtime": 1.2113,
+      "eval_samples_per_second": 825.526,
+      "eval_steps_per_second": 13.208,
+      "step": 28000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00029996115212854366,
+      "loss": 0.3886,
+      "step": 28500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029994926061660554,
+      "loss": 0.3863,
+      "step": 29000
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 0.8757530450820923,
+      "eval_runtime": 1.1818,
+      "eval_samples_per_second": 846.134,
+      "eval_steps_per_second": 13.538,
+      "step": 29000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002999357839627762,
+      "loss": 0.3841,
+      "step": 29500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029992072231443425,
+      "loss": 0.382,
+      "step": 30000
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 0.8755055069923401,
+      "eval_runtime": 1.2044,
+      "eval_samples_per_second": 830.296,
+      "eval_steps_per_second": 13.285,
+      "step": 30000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002999040758362914,
+      "loss": 0.3804,
+      "step": 30500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029988584471039094,
+      "loss": 0.378,
+      "step": 31000
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 0.8780514001846313,
+      "eval_runtime": 1.1913,
+      "eval_samples_per_second": 839.425,
+      "eval_steps_per_second": 13.431,
+      "step": 31000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002998660291361054,
+      "loss": 0.3763,
+      "step": 31500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002998446293301349,
+      "loss": 0.3748,
+      "step": 32000
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 0.8814873099327087,
+      "eval_runtime": 1.3032,
+      "eval_samples_per_second": 767.316,
+      "eval_steps_per_second": 12.277,
+      "step": 32000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002998216455265042,
+      "loss": 0.3735,
+      "step": 32500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029979707797656046,
+      "loss": 0.3716,
+      "step": 33000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 0.868879497051239,
+      "eval_runtime": 1.2062,
+      "eval_samples_per_second": 829.022,
+      "eval_steps_per_second": 13.264,
+      "step": 33000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029977092694897053,
+      "loss": 0.3702,
+      "step": 33500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002997431927297178,
+      "loss": 0.3689,
+      "step": 34000
+    },
+    {
+      "epoch": 0.87,
+      "eval_loss": 0.8758594989776611,
+      "eval_runtime": 1.2131,
+      "eval_samples_per_second": 824.364,
+      "eval_steps_per_second": 13.19,
+      "step": 34000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00029971387562209936,
+      "loss": 0.3677,
+      "step": 34500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029968297594672226,
+      "loss": 0.3665,
+      "step": 35000
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 0.8690294623374939,
+      "eval_runtime": 1.1601,
+      "eval_samples_per_second": 861.986,
+      "eval_steps_per_second": 13.792,
+      "step": 35000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002996504940415005,
+      "loss": 0.365,
+      "step": 35500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029961643026165096,
+      "loss": 0.364,
+      "step": 36000
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 0.8695961236953735,
+      "eval_runtime": 1.2184,
+      "eval_samples_per_second": 820.732,
+      "eval_steps_per_second": 13.132,
+      "step": 36000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029958078497968973,
+      "loss": 0.3627,
+      "step": 36500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002995435585854278,
+      "loss": 0.3614,
+      "step": 37000
+    },
+    {
+      "epoch": 0.94,
+      "eval_loss": 0.8684411644935608,
+      "eval_runtime": 1.2551,
+      "eval_samples_per_second": 796.719,
+      "eval_steps_per_second": 12.747,
+      "step": 37000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002995047514859671,
+      "loss": 0.3603,
+      "step": 37500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002994643641056959,
+      "loss": 0.3592,
+      "step": 38000
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 0.8597822189331055,
+      "eval_runtime": 1.198,
+      "eval_samples_per_second": 834.721,
+      "eval_steps_per_second": 13.356,
+      "step": 38000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000299422396886284,
+      "loss": 0.358,
+      "step": 38500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002993788502866783,
+      "loss": 0.3571,
+      "step": 39000
+    },
+    {
+      "epoch": 0.99,
+      "eval_loss": 0.8572097420692444,
+      "eval_runtime": 1.21,
+      "eval_samples_per_second": 826.463,
+      "eval_steps_per_second": 13.223,
+      "step": 39000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00029933372478309746,
+      "loss": 0.3562,
+      "step": 39500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029928702086902664,
+      "loss": 0.3555,
+      "step": 40000
+    },
+    {
+      "epoch": 1.02,
+      "eval_loss": 0.8637193441390991,
+      "eval_runtime": 1.4051,
+      "eval_samples_per_second": 711.674,
+      "eval_steps_per_second": 11.387,
+      "step": 40000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029923873905521244,
+      "loss": 0.3545,
+      "step": 40500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000299188879869657,
+      "loss": 0.3535,
+      "step": 41000
+    },
+    {
+      "epoch": 1.04,
+      "eval_loss": 0.8638287782669067,
+      "eval_runtime": 1.2182,
+      "eval_samples_per_second": 820.876,
+      "eval_steps_per_second": 13.134,
+      "step": 41000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029913744385761244,
+      "loss": 0.3524,
+      "step": 41500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029908443158157465,
+      "loss": 0.3518,
+      "step": 42000
+    },
+    {
+      "epoch": 1.07,
+      "eval_loss": 0.8664878606796265,
+      "eval_runtime": 1.2547,
+      "eval_samples_per_second": 796.987,
+      "eval_steps_per_second": 12.752,
+      "step": 42000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002990298436212775,
+      "loss": 0.3511,
+      "step": 42500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002989736805736861,
+      "loss": 0.3502,
+      "step": 43000
+    },
+    {
+      "epoch": 1.1,
+      "eval_loss": 0.8559480905532837,
+      "eval_runtime": 1.2272,
+      "eval_samples_per_second": 814.861,
+      "eval_steps_per_second": 13.038,
+      "step": 43000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00029891594305299065,
+      "loss": 0.3494,
+      "step": 43500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00029885663169059926,
+      "loss": 0.3488,
+      "step": 44000
+    },
+    {
+      "epoch": 1.12,
+      "eval_loss": 0.859957754611969,
+      "eval_runtime": 1.2944,
+      "eval_samples_per_second": 772.587,
+      "eval_steps_per_second": 12.361,
+      "step": 44000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002987957471351316,
+      "loss": 0.3478,
+      "step": 44500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00029873329005241137,
+      "loss": 0.3469,
+      "step": 45000
+    },
+    {
+      "epoch": 1.15,
+      "eval_loss": 0.852756917476654,
+      "eval_runtime": 1.2478,
+      "eval_samples_per_second": 801.439,
+      "eval_steps_per_second": 12.823,
+      "step": 45000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00029866926112545925,
+      "loss": 0.3464,
+      "step": 45500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00029860366105448534,
+      "loss": 0.3459,
+      "step": 46000
+    },
+    {
+      "epoch": 1.17,
+      "eval_loss": 0.8597527146339417,
+      "eval_runtime": 1.1814,
+      "eval_samples_per_second": 846.435,
+      "eval_steps_per_second": 13.543,
+      "step": 46000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00029853649055688143,
+      "loss": 0.3451,
+      "step": 46500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00029846775036721337,
+      "loss": 0.3444,
+      "step": 47000
+    },
+    {
+      "epoch": 1.2,
+      "eval_loss": 0.860701322555542,
+      "eval_runtime": 1.2083,
+      "eval_samples_per_second": 827.633,
+      "eval_steps_per_second": 13.242,
+      "step": 47000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002983974412372129,
+      "loss": 0.3438,
+      "step": 47500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00029832556393576934,
+      "loss": 0.3428,
+      "step": 48000
+    },
+    {
+      "epoch": 1.22,
+      "eval_loss": 0.865045964717865,
+      "eval_runtime": 1.2493,
+      "eval_samples_per_second": 800.445,
+      "eval_steps_per_second": 12.807,
+      "step": 48000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002982521192489214,
+      "loss": 0.3425,
+      "step": 48500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002981771079798483,
+      "loss": 0.342,
+      "step": 49000
+    },
+    {
+      "epoch": 1.25,
+      "eval_loss": 0.8640099167823792,
+      "eval_runtime": 1.2369,
+      "eval_samples_per_second": 808.489,
+      "eval_steps_per_second": 12.936,
+      "step": 49000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00029810053094886136,
+      "loss": 0.3417,
+      "step": 49500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00029802238899339473,
+      "loss": 0.3408,
+      "step": 50000
+    },
+    {
+      "epoch": 1.27,
+      "eval_loss": 0.8549481630325317,
+      "eval_runtime": 1.3108,
+      "eval_samples_per_second": 762.871,
+      "eval_steps_per_second": 12.206,
+      "step": 50000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002979426829679962,
+      "loss": 0.3403,
+      "step": 50500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002978614137443183,
+      "loss": 0.3398,
+      "step": 51000
+    },
+    {
+      "epoch": 1.3,
+      "eval_loss": 0.8630042672157288,
+      "eval_runtime": 1.1917,
+      "eval_samples_per_second": 839.116,
+      "eval_steps_per_second": 13.426,
+      "step": 51000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000297778582211108,
+      "loss": 0.3391,
+      "step": 51500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00029769418927419786,
+      "loss": 0.3387,
+      "step": 52000
+    },
+    {
+      "epoch": 1.33,
+      "eval_loss": 0.8540878891944885,
+      "eval_runtime": 1.2825,
+      "eval_samples_per_second": 779.707,
+      "eval_steps_per_second": 12.475,
+      "step": 52000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002976082358564954,
+      "loss": 0.338,
+      "step": 52500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00029752072289797353,
+      "loss": 0.3373,
+      "step": 53000
+    },
+    {
+      "epoch": 1.35,
+      "eval_loss": 0.8587937355041504,
+      "eval_runtime": 1.3038,
+      "eval_samples_per_second": 766.964,
+      "eval_steps_per_second": 12.271,
+      "step": 53000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00029743165135565986,
+      "loss": 0.3373,
+      "step": 53500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00029734102220362654,
+      "loss": 0.3368,
+      "step": 54000
+    },
+    {
+      "epoch": 1.38,
+      "eval_loss": 0.8638780117034912,
+      "eval_runtime": 1.2508,
+      "eval_samples_per_second": 799.458,
+      "eval_steps_per_second": 12.791,
+      "step": 54000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00029724883643297937,
+      "loss": 0.3362,
+      "step": 54500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002971550950518473,
+      "loss": 0.3357,
+      "step": 55000
+    },
+    {
+      "epoch": 1.4,
+      "eval_loss": 0.8546432256698608,
+      "eval_runtime": 1.2572,
+      "eval_samples_per_second": 795.415,
+      "eval_steps_per_second": 12.727,
+      "step": 55000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000297059799085371,
+      "loss": 0.3351,
+      "step": 55500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00029696294957569196,
+      "loss": 0.335,
+      "step": 56000
+    },
+    {
+      "epoch": 1.43,
+      "eval_loss": 0.8535034656524658,
+      "eval_runtime": 1.2455,
+      "eval_samples_per_second": 802.916,
+      "eval_steps_per_second": 12.847,
+      "step": 56000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00029686454758194076,
+      "loss": 0.3343,
+      "step": 56500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00029676459418022594,
+      "loss": 0.334,
+      "step": 57000
+    },
+    {
+      "epoch": 1.45,
+      "eval_loss": 0.8510599136352539,
+      "eval_runtime": 1.2497,
+      "eval_samples_per_second": 800.205,
+      "eval_steps_per_second": 12.803,
+      "step": 57000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002966630904636219,
+      "loss": 0.3335,
+      "step": 57500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002965600375421569,
+      "loss": 0.333,
+      "step": 58000
+    },
+    {
+      "epoch": 1.48,
+      "eval_loss": 0.8525308966636658,
+      "eval_runtime": 1.285,
+      "eval_samples_per_second": 778.223,
+      "eval_steps_per_second": 12.452,
+      "step": 58000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002964554365428013,
+      "loss": 0.3328,
+      "step": 58500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00029634928860945486,
+      "loss": 0.3322,
+      "step": 59000
+    },
+    {
+      "epoch": 1.5,
+      "eval_loss": 0.8536081910133362,
+      "eval_runtime": 1.2803,
+      "eval_samples_per_second": 781.08,
+      "eval_steps_per_second": 12.497,
+      "step": 59000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002962415949029343,
+      "loss": 0.3319,
+      "step": 59500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00029613235660096084,
+      "loss": 0.3314,
+      "step": 60000
+    },
+    {
+      "epoch": 1.53,
+      "eval_loss": 0.8421887159347534,
+      "eval_runtime": 1.3286,
+      "eval_samples_per_second": 752.666,
+      "eval_steps_per_second": 12.043,
+      "step": 60000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00029602157489814693,
+      "loss": 0.3311,
+      "step": 60500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00029590925100598365,
+      "loss": 0.3307,
+      "step": 61000
+    },
+    {
+      "epoch": 1.55,
+      "eval_loss": 0.8626954555511475,
+      "eval_runtime": 1.268,
+      "eval_samples_per_second": 788.647,
+      "eval_steps_per_second": 12.618,
+      "step": 61000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000295795386152827,
+      "loss": 0.3302,
+      "step": 61500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002956799815838848,
+      "loss": 0.3298,
+      "step": 62000
+    },
+    {
+      "epoch": 1.58,
+      "eval_loss": 0.8434558510780334,
+      "eval_runtime": 1.2564,
+      "eval_samples_per_second": 795.92,
+      "eval_steps_per_second": 12.735,
+      "step": 62000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002955630385612029,
+      "loss": 0.3294,
+      "step": 62500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002954445583636515,
+      "loss": 0.3292,
+      "step": 63000
+    },
+    {
+      "epoch": 1.61,
+      "eval_loss": 0.8569065928459167,
+      "eval_runtime": 1.2969,
+      "eval_samples_per_second": 771.061,
+      "eval_steps_per_second": 12.337,
+      "step": 63000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00029532454228691103,
+      "loss": 0.3288,
+      "step": 63500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002952029916434581,
+      "loss": 0.3287,
+      "step": 64000
+    },
+    {
+      "epoch": 1.63,
+      "eval_loss": 0.8517204523086548,
+      "eval_runtime": 1.2275,
+      "eval_samples_per_second": 814.648,
+      "eval_steps_per_second": 13.034,
+      "step": 64000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00029507990776255107,
+      "loss": 0.3282,
+      "step": 64500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00029495529199021555,
+      "loss": 0.3278,
+      "step": 65000
+    },
+    {
+      "epoch": 1.66,
+      "eval_loss": 0.8487727046012878,
+      "eval_runtime": 1.2356,
+      "eval_samples_per_second": 809.327,
+      "eval_steps_per_second": 12.949,
+      "step": 65000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002948291456892296,
+      "loss": 0.3273,
+      "step": 65500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00029470147023910907,
+      "loss": 0.3269,
+      "step": 66000
+    },
+    {
+      "epoch": 1.68,
+      "eval_loss": 0.8469608426094055,
+      "eval_runtime": 1.2425,
+      "eval_samples_per_second": 804.832,
+      "eval_steps_per_second": 12.877,
+      "step": 66000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002945722670360921,
+      "loss": 0.3265,
+      "step": 66500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002944415374931243,
+      "loss": 0.3262,
+      "step": 67000
+    },
+    {
+      "epoch": 1.71,
+      "eval_loss": 0.8486921191215515,
+      "eval_runtime": 1.2218,
+      "eval_samples_per_second": 818.455,
+      "eval_steps_per_second": 13.095,
+      "step": 67000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00029430928303984295,
+      "loss": 0.326,
+      "step": 67500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002941755051225616,
+      "loss": 0.3257,
+      "step": 68000
+    },
+    {
+      "epoch": 1.73,
+      "eval_loss": 0.8429991602897644,
+      "eval_runtime": 1.2262,
+      "eval_samples_per_second": 815.556,
+      "eval_steps_per_second": 13.049,
+      "step": 68000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00029404020520425417,
+      "loss": 0.3253,
+      "step": 68500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002939033847645388,
+      "loss": 0.325,
+      "step": 69000
+    },
+    {
+      "epoch": 1.76,
+      "eval_loss": 0.8381994366645813,
+      "eval_runtime": 1.2665,
+      "eval_samples_per_second": 789.582,
+      "eval_steps_per_second": 12.633,
+      "step": 69000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00029376504529966195,
+      "loss": 0.3246,
+      "step": 69500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00029362518832248184,
+      "loss": 0.3248,
+      "step": 70000
+    },
+    {
+      "epoch": 1.78,
+      "eval_loss": 0.848027229309082,
+      "eval_runtime": 1.243,
+      "eval_samples_per_second": 804.488,
+      "eval_steps_per_second": 12.872,
+      "step": 70000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002934838153624519,
+      "loss": 0.3243,
+      "step": 70500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00029334092796560427,
+      "loss": 0.3238,
+      "step": 71000
+    },
+    {
+      "epoch": 1.81,
+      "eval_loss": 0.8432244658470154,
+      "eval_runtime": 1.2842,
+      "eval_samples_per_second": 778.669,
+      "eval_steps_per_second": 12.459,
+      "step": 71000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002931965276945326,
+      "loss": 0.3235,
+      "step": 71500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002930506161283751,
+      "loss": 0.3235,
+      "step": 72000
+    },
+    {
+      "epoch": 1.83,
+      "eval_loss": 0.8572074174880981,
+      "eval_runtime": 1.2536,
+      "eval_samples_per_second": 797.713,
+      "eval_steps_per_second": 12.763,
+      "step": 72000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00029290319486279724,
+      "loss": 0.3229,
+      "step": 72500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002927542655099744,
+      "loss": 0.3227,
+      "step": 73000
+    },
+    {
+      "epoch": 1.86,
+      "eval_loss": 0.8465535044670105,
+      "eval_runtime": 1.184,
+      "eval_samples_per_second": 844.583,
+      "eval_steps_per_second": 13.513,
+      "step": 73000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00029260382969857417,
+      "loss": 0.3223,
+      "step": 73500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00029245188907373845,
+      "loss": 0.3224,
+      "step": 74000
+    },
+    {
+      "epoch": 1.89,
+      "eval_loss": 0.8523721694946289,
+      "eval_runtime": 1.247,
+      "eval_samples_per_second": 801.902,
+      "eval_steps_per_second": 12.83,
+      "step": 74000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002922984452970655,
+      "loss": 0.3219,
+      "step": 74500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000292143500046592,
+      "loss": 0.3217,
+      "step": 75000
+    },
+    {
+      "epoch": 1.91,
+      "eval_loss": 0.8451367616653442,
+      "eval_runtime": 1.266,
+      "eval_samples_per_second": 789.893,
+      "eval_steps_per_second": 12.638,
+      "step": 75000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002919870550167743,
+      "loss": 0.3216,
+      "step": 75500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002918291119184702,
+      "loss": 0.321,
+      "step": 76000
+    },
+    {
+      "epoch": 1.94,
+      "eval_loss": 0.8452543020248413,
+      "eval_runtime": 1.2406,
+      "eval_samples_per_second": 806.062,
+      "eval_steps_per_second": 12.897,
+      "step": 76000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002916696724789201,
+      "loss": 0.321,
+      "step": 76500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00029150873844172823,
+      "loss": 0.3207,
+      "step": 77000
+    },
+    {
+      "epoch": 1.96,
+      "eval_loss": 0.8388876914978027,
+      "eval_runtime": 1.2429,
+      "eval_samples_per_second": 804.592,
+      "eval_steps_per_second": 12.873,
+      "step": 77000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00029134631156684334,
+      "loss": 0.3204,
+      "step": 77500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002911823936305398,
+      "loss": 0.3202,
+      "step": 78000
+    },
+    {
+      "epoch": 1.99,
+      "eval_loss": 0.8390601873397827,
+      "eval_runtime": 1.2051,
+      "eval_samples_per_second": 829.802,
+      "eval_steps_per_second": 13.277,
+      "step": 78000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002910169864253979,
+      "loss": 0.3198,
+      "step": 78500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002908500917602842,
+      "loss": 0.3195,
+      "step": 79000
+    },
+    {
+      "epoch": 2.01,
+      "eval_loss": 0.853469729423523,
+      "eval_runtime": 1.2478,
+      "eval_samples_per_second": 801.439,
+      "eval_steps_per_second": 12.823,
+      "step": 79000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00029068171146033226,
+      "loss": 0.3196,
+      "step": 79500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002905118473669218,
+      "loss": 0.3194,
+      "step": 80000
+    },
+    {
+      "epoch": 2.04,
+      "eval_loss": 0.8577731847763062,
+      "eval_runtime": 1.1943,
+      "eval_samples_per_second": 837.345,
+      "eval_steps_per_second": 13.398,
+      "step": 80000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00029034050133765947,
+      "loss": 0.319,
+      "step": 80500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00029016767524635804,
+      "loss": 0.3194,
+      "step": 81000
+    },
+    {
+      "epoch": 2.06,
+      "eval_loss": 0.851713240146637,
+      "eval_runtime": 1.28,
+      "eval_samples_per_second": 781.247,
+      "eval_steps_per_second": 12.5,
+      "step": 81000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00028999337098301585,
+      "loss": 0.3187,
+      "step": 81500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002898175904537964,
+      "loss": 0.3185,
+      "step": 82000
+    },
+    {
+      "epoch": 2.09,
+      "eval_loss": 0.8537120223045349,
+      "eval_runtime": 1.24,
+      "eval_samples_per_second": 806.425,
+      "eval_steps_per_second": 12.903,
+      "step": 82000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002896403355810075,
+      "loss": 0.3179,
+      "step": 82500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00028946160830307997,
+      "loss": 0.3183,
+      "step": 83000
+    },
+    {
+      "epoch": 2.12,
+      "eval_loss": 0.8353117108345032,
+      "eval_runtime": 1.2534,
+      "eval_samples_per_second": 797.829,
+      "eval_steps_per_second": 12.765,
+      "step": 83000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00028928141057454665,
+      "loss": 0.3182,
+      "step": 83500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002890997443660211,
+      "loss": 0.3177,
+      "step": 84000
+    },
+    {
+      "epoch": 2.14,
+      "eval_loss": 0.845744252204895,
+      "eval_runtime": 1.2558,
+      "eval_samples_per_second": 796.305,
+      "eval_steps_per_second": 12.741,
+      "step": 84000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00028891661166417586,
+      "loss": 0.3174,
+      "step": 84500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00028873201447172074,
+      "loss": 0.3174,
+      "step": 85000
+    },
+    {
+      "epoch": 2.17,
+      "eval_loss": 0.8514609932899475,
+      "eval_runtime": 1.2468,
+      "eval_samples_per_second": 802.073,
+      "eval_steps_per_second": 12.833,
+      "step": 85000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002885459548073812,
+      "loss": 0.3173,
+      "step": 85500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002883584347058758,
+      "loss": 0.3165,
+      "step": 86000
+    },
+    {
+      "epoch": 2.19,
+      "eval_loss": 0.8357125520706177,
+      "eval_runtime": 1.221,
+      "eval_samples_per_second": 818.991,
+      "eval_steps_per_second": 13.104,
+      "step": 86000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00028816945621789437,
+      "loss": 0.3166,
+      "step": 86500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002879790214100753,
+      "loss": 0.3164,
+      "step": 87000
+    },
+    {
+      "epoch": 2.22,
+      "eval_loss": 0.855523943901062,
+      "eval_runtime": 1.2726,
+      "eval_samples_per_second": 785.803,
+      "eval_steps_per_second": 12.573,
+      "step": 87000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002877871323649833,
+      "loss": 0.3161,
+      "step": 87500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002875937911810861,
+      "loss": 0.3159,
+      "step": 88000
+    },
+    {
+      "epoch": 2.24,
+      "eval_loss": 0.842631995677948,
+      "eval_runtime": 1.2074,
+      "eval_samples_per_second": 828.219,
+      "eval_steps_per_second": 13.252,
+      "step": 88000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.000287398999972732,
+      "loss": 0.3159,
+      "step": 88500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00028720276087012636,
+      "loss": 0.3159,
+      "step": 89000
+    },
+    {
+      "epoch": 2.27,
+      "eval_loss": 0.8528951406478882,
+      "eval_runtime": 1.2566,
+      "eval_samples_per_second": 795.823,
+      "eval_steps_per_second": 12.733,
+      "step": 89000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002870050760193086,
+      "loss": 0.3152,
+      "step": 89500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00028680594758212854,
+      "loss": 0.3152,
+      "step": 90000
+    },
+    {
+      "epoch": 2.29,
+      "eval_loss": 0.8297361135482788,
+      "eval_runtime": 1.2298,
+      "eval_samples_per_second": 813.129,
+      "eval_steps_per_second": 13.01,
+      "step": 90000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00028660537773622294,
+      "loss": 0.3158,
+      "step": 90500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00028640336867499143,
+      "loss": 0.3149,
+      "step": 91000
+    },
+    {
+      "epoch": 2.32,
+      "eval_loss": 0.8462135195732117,
+      "eval_runtime": 1.1973,
+      "eval_samples_per_second": 835.246,
+      "eval_steps_per_second": 13.364,
+      "step": 91000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002861999226075728,
+      "loss": 0.3148,
+      "step": 91500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002859950417588206,
+      "loss": 0.3144,
+      "step": 92000
+    },
+    {
+      "epoch": 2.34,
+      "eval_loss": 0.8435172438621521,
+      "eval_runtime": 1.2492,
+      "eval_samples_per_second": 800.501,
+      "eval_steps_per_second": 12.808,
+      "step": 92000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00028578872836927904,
+      "loss": 0.3144,
+      "step": 92500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002855809846951582,
+      "loss": 0.3141,
+      "step": 93000
+    },
+    {
+      "epoch": 2.37,
+      "eval_loss": 0.8363200426101685,
+      "eval_runtime": 1.2168,
+      "eval_samples_per_second": 821.803,
+      "eval_steps_per_second": 13.149,
+      "step": 93000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00028537181300830963,
+      "loss": 0.3138,
+      "step": 93500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002851612155962014,
+      "loss": 0.3139,
+      "step": 94000
+    },
+    {
+      "epoch": 2.4,
+      "eval_loss": 0.8434939384460449,
+      "eval_runtime": 1.2528,
+      "eval_samples_per_second": 798.238,
+      "eval_steps_per_second": 12.772,
+      "step": 94000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002849491947618932,
+      "loss": 0.3139,
+      "step": 94500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002847357528240107,
+      "loss": 0.3139,
+      "step": 95000
+    },
+    {
+      "epoch": 2.42,
+      "eval_loss": 0.8472431898117065,
+      "eval_runtime": 1.2611,
+      "eval_samples_per_second": 792.952,
+      "eval_steps_per_second": 12.687,
+      "step": 95000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002845208921167208,
+      "loss": 0.3137,
+      "step": 95500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00028430461498970584,
+      "loss": 0.3131,
+      "step": 96000
+    },
+    {
+      "epoch": 2.45,
+      "eval_loss": 0.8395859599113464,
+      "eval_runtime": 1.2301,
+      "eval_samples_per_second": 812.96,
+      "eval_steps_per_second": 13.007,
+      "step": 96000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00028408692380813775,
+      "loss": 0.3125,
+      "step": 96500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00028386782095265247,
+      "loss": 0.3125,
+      "step": 97000
+    },
+    {
+      "epoch": 2.47,
+      "eval_loss": 0.8419870734214783,
+      "eval_runtime": 1.245,
+      "eval_samples_per_second": 803.234,
+      "eval_steps_per_second": 12.852,
+      "step": 97000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002836473088193237,
+      "loss": 0.313,
+      "step": 97500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00028342538981963677,
+      "loss": 0.3128,
+      "step": 98000
+    },
+    {
+      "epoch": 2.5,
+      "eval_loss": 0.8410093784332275,
+      "eval_runtime": 1.2759,
+      "eval_samples_per_second": 783.735,
+      "eval_steps_per_second": 12.54,
+      "step": 98000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0002832020663804624,
+      "loss": 0.3124,
+      "step": 98500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00028297734094402986,
+      "loss": 0.3121,
+      "step": 99000
+    },
+    {
+      "epoch": 2.52,
+      "eval_loss": 0.8417559862136841,
+      "eval_runtime": 1.2688,
+      "eval_samples_per_second": 788.14,
+      "eval_steps_per_second": 12.61,
+      "step": 99000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002827512159679005,
+      "loss": 0.3122,
+      "step": 99500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00028252369392494086,
+      "loss": 0.3118,
+      "step": 100000
+    },
+    {
+      "epoch": 2.55,
+      "eval_loss": 0.8400096893310547,
+      "eval_runtime": 1.2486,
+      "eval_samples_per_second": 800.914,
+      "eval_steps_per_second": 12.815,
+      "step": 100000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002822947773032956,
+      "loss": 0.3117,
+      "step": 100500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0002820644686063602,
+      "loss": 0.3112,
+      "step": 101000
+    },
+    {
+      "epoch": 2.57,
+      "eval_loss": 0.8347204327583313,
+      "eval_runtime": 1.2512,
+      "eval_samples_per_second": 799.226,
+      "eval_steps_per_second": 12.788,
+      "step": 101000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00028183277035275363,
+      "loss": 0.3117,
+      "step": 101500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0002815996850762909,
+      "loss": 0.3112,
+      "step": 102000
+    },
+    {
+      "epoch": 2.6,
+      "eval_loss": 0.8288648724555969,
+      "eval_runtime": 1.2077,
+      "eval_samples_per_second": 828.045,
+      "eval_steps_per_second": 13.249,
+      "step": 102000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00028136521532595515,
+      "loss": 0.311,
+      "step": 102500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00028112936366587023,
+      "loss": 0.3112,
+      "step": 103000
+    },
+    {
+      "epoch": 2.63,
+      "eval_loss": 0.8455829620361328,
+      "eval_runtime": 1.2774,
+      "eval_samples_per_second": 782.835,
+      "eval_steps_per_second": 12.525,
+      "step": 103000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00028089213267527184,
+      "loss": 0.311,
+      "step": 103500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002806535249484803,
+      "loss": 0.3107,
+      "step": 104000
+    },
+    {
+      "epoch": 2.65,
+      "eval_loss": 0.8414435982704163,
+      "eval_runtime": 1.2285,
+      "eval_samples_per_second": 813.99,
+      "eval_steps_per_second": 13.024,
+      "step": 104000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00028041354309487135,
+      "loss": 0.3108,
+      "step": 104500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0002801721897388482,
+      "loss": 0.3101,
+      "step": 105000
+    },
+    {
+      "epoch": 2.68,
+      "eval_loss": 0.8326617479324341,
+      "eval_runtime": 1.214,
+      "eval_samples_per_second": 823.726,
+      "eval_steps_per_second": 13.18,
+      "step": 105000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002799294675198124,
+      "loss": 0.3102,
+      "step": 105500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00027968537909213524,
+      "loss": 0.3107,
+      "step": 106000
+    },
+    {
+      "epoch": 2.7,
+      "eval_loss": 0.837422251701355,
+      "eval_runtime": 1.2383,
+      "eval_samples_per_second": 807.572,
+      "eval_steps_per_second": 12.921,
+      "step": 106000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0002794399271251287,
+      "loss": 0.3102,
+      "step": 106500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002791931143030162,
+      "loss": 0.3103,
+      "step": 107000
+    },
+    {
+      "epoch": 2.73,
+      "eval_loss": 0.8471120595932007,
+      "eval_runtime": 1.2736,
+      "eval_samples_per_second": 785.146,
+      "eval_steps_per_second": 12.562,
+      "step": 107000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00027894494332490315,
+      "loss": 0.3104,
+      "step": 107500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002786954169047476,
+      "loss": 0.3095,
+      "step": 108000
+    },
+    {
+      "epoch": 2.75,
+      "eval_loss": 0.845231831073761,
+      "eval_runtime": 1.222,
+      "eval_samples_per_second": 818.338,
+      "eval_steps_per_second": 13.093,
+      "step": 108000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002784445377713306,
+      "loss": 0.3093,
+      "step": 108500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002781923086682261,
+      "loss": 0.3094,
+      "step": 109000
+    },
+    {
+      "epoch": 2.78,
+      "eval_loss": 0.8512564301490784,
+      "eval_runtime": 1.2211,
+      "eval_samples_per_second": 818.908,
+      "eval_steps_per_second": 13.103,
+      "step": 109000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0002779387323537711,
+      "loss": 0.3091,
+      "step": 109500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0002776838116010356,
+      "loss": 0.3094,
+      "step": 110000
+    },
+    {
+      "epoch": 2.8,
+      "eval_loss": 0.8348438739776611,
+      "eval_runtime": 1.2126,
+      "eval_samples_per_second": 824.682,
+      "eval_steps_per_second": 13.195,
+      "step": 110000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002774275491977922,
+      "loss": 0.3111,
+      "step": 110500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0002771699479464853,
+      "loss": 0.3089,
+      "step": 111000
+    },
+    {
+      "epoch": 2.83,
+      "eval_loss": 0.8333858847618103,
+      "eval_runtime": 1.2664,
+      "eval_samples_per_second": 789.664,
+      "eval_steps_per_second": 12.635,
+      "step": 111000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00027691101066420104,
+      "loss": 0.3091,
+      "step": 111500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0002766507401826361,
+      "loss": 0.3089,
+      "step": 112000
+    },
+    {
+      "epoch": 2.85,
+      "eval_loss": 0.8437691330909729,
+      "eval_runtime": 1.2859,
+      "eval_samples_per_second": 777.686,
+      "eval_steps_per_second": 12.443,
+      "step": 112000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0002763891393480666,
+      "loss": 0.3086,
+      "step": 112500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0002761262110213175,
+      "loss": 0.3088,
+      "step": 113000
+    },
+    {
+      "epoch": 2.88,
+      "eval_loss": 0.8328044414520264,
+      "eval_runtime": 1.2617,
+      "eval_samples_per_second": 792.564,
+      "eval_steps_per_second": 12.681,
+      "step": 113000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00027586195807773083,
+      "loss": 0.308,
+      "step": 113500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00027559638340713435,
+      "loss": 0.3085,
+      "step": 114000
+    },
+    {
+      "epoch": 2.91,
+      "eval_loss": 0.831738293170929,
+      "eval_runtime": 1.2398,
+      "eval_samples_per_second": 806.55,
+      "eval_steps_per_second": 12.905,
+      "step": 114000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00027532948991381025,
+      "loss": 0.3079,
+      "step": 114500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00027506128051646287,
+      "loss": 0.3097,
+      "step": 115000
+    },
+    {
+      "epoch": 2.93,
+      "eval_loss": 0.8462249636650085,
+      "eval_runtime": 1.2264,
+      "eval_samples_per_second": 815.428,
+      "eval_steps_per_second": 13.047,
+      "step": 115000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00027479175814818733,
+      "loss": 0.3082,
+      "step": 115500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.000274520925756437,
+      "loss": 0.3082,
+      "step": 116000
+    },
+    {
+      "epoch": 2.96,
+      "eval_loss": 0.8435738682746887,
+      "eval_runtime": 1.3152,
+      "eval_samples_per_second": 760.322,
+      "eval_steps_per_second": 12.165,
+      "step": 116000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00027424878630299157,
+      "loss": 0.3074,
+      "step": 116500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0002739753427639244,
+      "loss": 0.3077,
+      "step": 117000
+    },
+    {
+      "epoch": 2.98,
+      "eval_loss": 0.8435785174369812,
+      "eval_runtime": 1.2302,
+      "eval_samples_per_second": 812.844,
+      "eval_steps_per_second": 13.006,
+      "step": 117000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0002737005981295704,
+      "loss": 0.3078,
+      "step": 117500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0002734245554044927,
+      "loss": 0.3086,
+      "step": 118000
+    },
+    {
+      "epoch": 3.01,
+      "eval_loss": 0.8482502698898315,
+      "eval_runtime": 1.3338,
+      "eval_samples_per_second": 749.745,
+      "eval_steps_per_second": 11.996,
+      "step": 118000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0002731472176074504,
+      "loss": 0.307,
+      "step": 118500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0002728685877713653,
+      "loss": 0.3072,
+      "step": 119000
+    },
+    {
+      "epoch": 3.03,
+      "eval_loss": 0.8354719281196594,
+      "eval_runtime": 1.4002,
+      "eval_samples_per_second": 714.17,
+      "eval_steps_per_second": 11.427,
+      "step": 119000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0002725886689432884,
+      "loss": 0.3075,
+      "step": 119500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0002723074641843674,
+      "loss": 0.3066,
+      "step": 120000
+    },
+    {
+      "epoch": 3.06,
+      "eval_loss": 0.8281124234199524,
+      "eval_runtime": 1.3099,
+      "eval_samples_per_second": 763.433,
+      "eval_steps_per_second": 12.215,
+      "step": 120000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0002720249765698123,
+      "loss": 0.3066,
+      "step": 120500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0002717412091888626,
+      "loss": 0.3072,
+      "step": 121000
+    },
+    {
+      "epoch": 3.08,
+      "eval_loss": 0.8392689824104309,
+      "eval_runtime": 1.2716,
+      "eval_samples_per_second": 786.397,
+      "eval_steps_per_second": 12.582,
+      "step": 121000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00027145616514475274,
+      "loss": 0.3067,
+      "step": 121500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0002711698475546788,
+      "loss": 0.3063,
+      "step": 122000
+    },
+    {
+      "epoch": 3.11,
+      "eval_loss": 0.8436251282691956,
+      "eval_runtime": 1.2483,
+      "eval_samples_per_second": 801.08,
+      "eval_steps_per_second": 12.817,
+      "step": 122000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00027088225954976407,
+      "loss": 0.3067,
+      "step": 122500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0002705934042750249,
+      "loss": 0.3061,
+      "step": 123000
+    },
+    {
+      "epoch": 3.13,
+      "eval_loss": 0.8345765471458435,
+      "eval_runtime": 1.2507,
+      "eval_samples_per_second": 799.522,
+      "eval_steps_per_second": 12.792,
+      "step": 123000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00027030328488933625,
+      "loss": 0.3071,
+      "step": 123500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00027001190456539726,
+      "loss": 0.3059,
+      "step": 124000
+    },
+    {
+      "epoch": 3.16,
+      "eval_loss": 0.8408392071723938,
+      "eval_runtime": 1.3349,
+      "eval_samples_per_second": 749.093,
+      "eval_steps_per_second": 11.985,
+      "step": 124000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0002697192664896965,
+      "loss": 0.3062,
+      "step": 124500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00026942537386247706,
+      "loss": 0.3062,
+      "step": 125000
+    },
+    {
+      "epoch": 3.19,
+      "eval_loss": 0.8384091854095459,
+      "eval_runtime": 1.3159,
+      "eval_samples_per_second": 759.918,
+      "eval_steps_per_second": 12.159,
+      "step": 125000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0002691302298977016,
+      "loss": 0.3058,
+      "step": 125500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0002688338378230173,
+      "loss": 0.307,
+      "step": 126000
+    },
+    {
+      "epoch": 3.21,
+      "eval_loss": 0.837376594543457,
+      "eval_runtime": 1.3811,
+      "eval_samples_per_second": 724.072,
+      "eval_steps_per_second": 11.585,
+      "step": 126000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00026853620087972035,
+      "loss": 0.3067,
+      "step": 126500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00026823732232272065,
+      "loss": 0.3056,
+      "step": 127000
+    },
+    {
+      "epoch": 3.24,
+      "eval_loss": 0.8240262269973755,
+      "eval_runtime": 1.3027,
+      "eval_samples_per_second": 767.62,
+      "eval_steps_per_second": 12.282,
+      "step": 127000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0002679372054205063,
+      "loss": 0.305,
+      "step": 127500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0002676358534551076,
+      "loss": 0.3049,
+      "step": 128000
+    },
+    {
+      "epoch": 3.26,
+      "eval_loss": 0.8262757658958435,
+      "eval_runtime": 1.3606,
+      "eval_samples_per_second": 734.996,
+      "eval_steps_per_second": 11.76,
+      "step": 128000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00026733326972206133,
+      "loss": 0.306,
+      "step": 128500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002670294575303748,
+      "loss": 0.3068,
+      "step": 129000
+    },
+    {
+      "epoch": 3.29,
+      "eval_loss": 0.8301039934158325,
+      "eval_runtime": 1.2964,
+      "eval_samples_per_second": 771.396,
+      "eval_steps_per_second": 12.342,
+      "step": 129000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0002667244202024894,
+      "loss": 0.305,
+      "step": 129500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00026641816107424453,
+      "loss": 0.3053,
+      "step": 130000
+    },
+    {
+      "epoch": 3.31,
+      "eval_loss": 0.8349767923355103,
+      "eval_runtime": 1.3045,
+      "eval_samples_per_second": 766.596,
+      "eval_steps_per_second": 12.266,
+      "step": 130000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0002661106834948409,
+      "loss": 0.3058,
+      "step": 130500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0002658019908268041,
+      "loss": 0.3048,
+      "step": 131000
+    },
+    {
+      "epoch": 3.34,
+      "eval_loss": 0.8294724822044373,
+      "eval_runtime": 1.2682,
+      "eval_samples_per_second": 788.537,
+      "eval_steps_per_second": 12.617,
+      "step": 131000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00026549208644594766,
+      "loss": 0.305,
+      "step": 131500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00026518097374133627,
+      "loss": 0.3051,
+      "step": 132000
+    },
+    {
+      "epoch": 3.36,
+      "eval_loss": 0.8296982049942017,
+      "eval_runtime": 1.3247,
+      "eval_samples_per_second": 754.912,
+      "eval_steps_per_second": 12.079,
+      "step": 132000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00026486865611524853,
+      "loss": 0.3045,
+      "step": 132500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00026455513698314003,
+      "loss": 0.3045,
+      "step": 133000
+    },
+    {
+      "epoch": 3.39,
+      "eval_loss": 0.8295329213142395,
+      "eval_runtime": 1.3528,
+      "eval_samples_per_second": 739.189,
+      "eval_steps_per_second": 11.827,
+      "step": 133000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0002642404197736058,
+      "loss": 0.3045,
+      "step": 133500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0002639245079283428,
+      "loss": 0.3043,
+      "step": 134000
+    },
+    {
+      "epoch": 3.42,
+      "eval_loss": 0.8244909644126892,
+      "eval_runtime": 1.3196,
+      "eval_samples_per_second": 757.824,
+      "eval_steps_per_second": 12.125,
+      "step": 134000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00026360740490211234,
+      "loss": 0.3038,
+      "step": 134500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0002632891141627023,
+      "loss": 0.3037,
+      "step": 135000
+    },
+    {
+      "epoch": 3.44,
+      "eval_loss": 0.8188607096672058,
+      "eval_runtime": 1.2963,
+      "eval_samples_per_second": 771.439,
+      "eval_steps_per_second": 12.343,
+      "step": 135000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00026296963919088923,
+      "loss": 0.304,
+      "step": 135500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00026264898348040024,
+      "loss": 0.3042,
+      "step": 136000
+    },
+    {
+      "epoch": 3.47,
+      "eval_loss": 0.8286353349685669,
+      "eval_runtime": 1.3187,
+      "eval_samples_per_second": 758.324,
+      "eval_steps_per_second": 12.133,
+      "step": 136000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0002623271505378748,
+      "loss": 0.3038,
+      "step": 136500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00026200414388282637,
+      "loss": 0.3038,
+      "step": 137000
+    },
+    {
+      "epoch": 3.49,
+      "eval_loss": 0.8326404690742493,
+      "eval_runtime": 1.3251,
+      "eval_samples_per_second": 754.645,
+      "eval_steps_per_second": 12.074,
+      "step": 137000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00026167996704760406,
+      "loss": 0.3058,
+      "step": 137500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00026135462357735375,
+      "loss": 0.3033,
+      "step": 138000
+    },
+    {
+      "epoch": 3.52,
+      "eval_loss": 0.8183851838111877,
+      "eval_runtime": 1.3581,
+      "eval_samples_per_second": 736.344,
+      "eval_steps_per_second": 11.782,
+      "step": 138000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0002610281170299795,
+      "loss": 0.3035,
+      "step": 138500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00026070045097610465,
+      "loss": 0.3035,
+      "step": 139000
+    },
+    {
+      "epoch": 3.54,
+      "eval_loss": 0.8135998249053955,
+      "eval_runtime": 1.309,
+      "eval_samples_per_second": 763.92,
+      "eval_steps_per_second": 12.223,
+      "step": 139000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0002603716289990326,
+      "loss": 0.3026,
+      "step": 139500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00026004165469470787,
+      "loss": 0.3027,
+      "step": 140000
+    },
+    {
+      "epoch": 3.57,
+      "eval_loss": 0.8287385106086731,
+      "eval_runtime": 1.2779,
+      "eval_samples_per_second": 782.532,
+      "eval_steps_per_second": 12.521,
+      "step": 140000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0002597105316716766,
+      "loss": 0.3038,
+      "step": 140500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.000259378263551047,
+      "loss": 0.303,
+      "step": 141000
+    },
+    {
+      "epoch": 3.59,
+      "eval_loss": 0.8184419870376587,
+      "eval_runtime": 1.2761,
+      "eval_samples_per_second": 783.611,
+      "eval_steps_per_second": 12.538,
+      "step": 141000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0002590448539664501,
+      "loss": 0.3026,
+      "step": 141500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00025871030656399966,
+      "loss": 0.3027,
+      "step": 142000
+    },
+    {
+      "epoch": 3.62,
+      "eval_loss": 0.8444017171859741,
+      "eval_runtime": 1.2977,
+      "eval_samples_per_second": 770.565,
+      "eval_steps_per_second": 12.329,
+      "step": 142000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00025837462500225255,
+      "loss": 0.3026,
+      "step": 142500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0002580378129521685,
+      "loss": 0.3024,
+      "step": 143000
+    },
+    {
+      "epoch": 3.64,
+      "eval_loss": 0.8401538133621216,
+      "eval_runtime": 1.3317,
+      "eval_samples_per_second": 750.939,
+      "eval_steps_per_second": 12.015,
+      "step": 143000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0002576998740970701,
+      "loss": 0.3031,
+      "step": 143500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00025736081213260253,
+      "loss": 0.3027,
+      "step": 144000
+    },
+    {
+      "epoch": 3.67,
+      "eval_loss": 0.8279548287391663,
+      "eval_runtime": 1.3186,
+      "eval_samples_per_second": 758.406,
+      "eval_steps_per_second": 12.134,
+      "step": 144000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0002570206307666931,
+      "loss": 0.3038,
+      "step": 144500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0002566793337195108,
+      "loss": 0.3029,
+      "step": 145000
+    },
+    {
+      "epoch": 3.7,
+      "eval_loss": 0.8254837393760681,
+      "eval_runtime": 1.2997,
+      "eval_samples_per_second": 769.407,
+      "eval_steps_per_second": 12.311,
+      "step": 145000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0002563369247234254,
+      "loss": 0.3038,
+      "step": 145500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0002559934075229669,
+      "loss": 0.3023,
+      "step": 146000
+    },
+    {
+      "epoch": 3.72,
+      "eval_loss": 0.8286687135696411,
+      "eval_runtime": 1.2798,
+      "eval_samples_per_second": 781.358,
+      "eval_steps_per_second": 12.502,
+      "step": 146000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0002556487858747843,
+      "loss": 0.3023,
+      "step": 146500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00025530306354760464,
+      "loss": 0.3024,
+      "step": 147000
+    },
+    {
+      "epoch": 3.75,
+      "eval_loss": 0.8175866007804871,
+      "eval_runtime": 1.2628,
+      "eval_samples_per_second": 791.915,
+      "eval_steps_per_second": 12.671,
+      "step": 147000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.000254956244322192,
+      "loss": 0.3022,
+      "step": 147500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00025460833199130595,
+      "loss": 0.302,
+      "step": 148000
+    },
+    {
+      "epoch": 3.77,
+      "eval_loss": 0.8372282385826111,
+      "eval_runtime": 1.3637,
+      "eval_samples_per_second": 733.286,
+      "eval_steps_per_second": 11.733,
+      "step": 148000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00025425933035965983,
+      "loss": 0.3031,
+      "step": 148500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00025390924324387965,
+      "loss": 0.3019,
+      "step": 149000
+    },
+    {
+      "epoch": 3.8,
+      "eval_loss": 0.8220604658126831,
+      "eval_runtime": 1.363,
+      "eval_samples_per_second": 733.669,
+      "eval_steps_per_second": 11.739,
+      "step": 149000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002535580744724621,
+      "loss": 0.3021,
+      "step": 149500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00025320582788573246,
+      "loss": 0.3016,
+      "step": 150000
+    },
+    {
+      "epoch": 3.82,
+      "eval_loss": 0.8250831365585327,
+      "eval_runtime": 1.3782,
+      "eval_samples_per_second": 725.588,
+      "eval_steps_per_second": 11.609,
+      "step": 150000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.000252852507335803,
+      "loss": 0.3008,
+      "step": 150500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002524981166865307,
+      "loss": 0.3014,
+      "step": 151000
+    },
+    {
+      "epoch": 3.85,
+      "eval_loss": 0.8369771242141724,
+      "eval_runtime": 1.2701,
+      "eval_samples_per_second": 787.344,
+      "eval_steps_per_second": 12.598,
+      "step": 151000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00025214265981347487,
+      "loss": 0.3007,
+      "step": 151500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.000251786140603855,
+      "loss": 0.3012,
+      "step": 152000
+    },
+    {
+      "epoch": 3.87,
+      "eval_loss": 0.8284549117088318,
+      "eval_runtime": 1.3354,
+      "eval_samples_per_second": 748.855,
+      "eval_steps_per_second": 11.982,
+      "step": 152000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00025142856295650795,
+      "loss": 0.3026,
+      "step": 152500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002510699307818457,
+      "loss": 0.3012,
+      "step": 153000
+    },
+    {
+      "epoch": 3.9,
+      "eval_loss": 0.8453160524368286,
+      "eval_runtime": 1.3185,
+      "eval_samples_per_second": 758.466,
+      "eval_steps_per_second": 12.135,
+      "step": 153000
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00025071024800181214,
+      "loss": 0.3009,
+      "step": 153500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0002503495185498405,
+      "loss": 0.3007,
+      "step": 154000
+    },
+    {
+      "epoch": 3.92,
+      "eval_loss": 0.8194591999053955,
+      "eval_runtime": 1.3313,
+      "eval_samples_per_second": 751.123,
+      "eval_steps_per_second": 12.018,
+      "step": 154000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00024998774637081044,
+      "loss": 0.3008,
+      "step": 154500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00024962493542100443,
+      "loss": 0.3009,
+      "step": 155000
+    },
+    {
+      "epoch": 3.95,
+      "eval_loss": 0.8309118747711182,
+      "eval_runtime": 1.304,
+      "eval_samples_per_second": 766.849,
+      "eval_steps_per_second": 12.27,
+      "step": 155000
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0002492610896680649,
+      "loss": 0.3001,
+      "step": 155500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00024889621309095067,
+      "loss": 0.3007,
+      "step": 156000
+    },
+    {
+      "epoch": 3.98,
+      "eval_loss": 0.8357002139091492,
+      "eval_runtime": 1.337,
+      "eval_samples_per_second": 747.964,
+      "eval_steps_per_second": 11.967,
+      "step": 156000
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00024853030967989366,
+      "loss": 0.3005,
+      "step": 156500
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00024816338343635485,
+      "loss": 0.3003,
+      "step": 157000
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 0.8225012421607971,
+      "eval_runtime": 1.2983,
+      "eval_samples_per_second": 770.222,
+      "eval_steps_per_second": 12.324,
+      "step": 157000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0002477954383729809,
+      "loss": 0.2997,
+      "step": 157500
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00024742647851355997,
+      "loss": 0.3014,
+      "step": 158000
+    },
+    {
+      "epoch": 4.03,
+      "eval_loss": 0.8343092203140259,
+      "eval_runtime": 1.2683,
+      "eval_samples_per_second": 788.447,
+      "eval_steps_per_second": 12.615,
+      "step": 158000
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0002470565078929781,
+      "loss": 0.3001,
+      "step": 158500
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00024668553055717465,
+      "loss": 0.3005,
+      "step": 159000
+    },
+    {
+      "epoch": 4.05,
+      "eval_loss": 0.8266953229904175,
+      "eval_runtime": 1.3214,
+      "eval_samples_per_second": 756.765,
+      "eval_steps_per_second": 12.108,
+      "step": 159000
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0002463135505630984,
+      "loss": 0.299,
+      "step": 159500
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00024594057197866283,
+      "loss": 0.2994,
+      "step": 160000
+    },
+    {
+      "epoch": 4.08,
+      "eval_loss": 0.8257858157157898,
+      "eval_runtime": 1.2672,
+      "eval_samples_per_second": 789.172,
+      "eval_steps_per_second": 12.627,
+      "step": 160000
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0002455665988827021,
+      "loss": 0.2998,
+      "step": 160500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0002451916353649261,
+      "loss": 0.2996,
+      "step": 161000
+    },
+    {
+      "epoch": 4.1,
+      "eval_loss": 0.8267243504524231,
+      "eval_runtime": 1.2949,
+      "eval_samples_per_second": 772.28,
+      "eval_steps_per_second": 12.356,
+      "step": 161000
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00024481568552587566,
+      "loss": 0.2999,
+      "step": 161500
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0002444387534768781,
+      "loss": 0.301,
+      "step": 162000
+    },
+    {
+      "epoch": 4.13,
+      "eval_loss": 0.8215921521186829,
+      "eval_runtime": 1.2633,
+      "eval_samples_per_second": 791.589,
+      "eval_steps_per_second": 12.665,
+      "step": 162000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0002440608433400018,
+      "loss": 0.2995,
+      "step": 162500
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00024368195924801158,
+      "loss": 0.2987,
+      "step": 163000
+    },
+    {
+      "epoch": 4.15,
+      "eval_loss": 0.8304322957992554,
+      "eval_runtime": 1.2627,
+      "eval_samples_per_second": 791.952,
+      "eval_steps_per_second": 12.671,
+      "step": 163000
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00024330210534432314,
+      "loss": 0.2989,
+      "step": 163500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0002429212857829579,
+      "loss": 0.2989,
+      "step": 164000
+    },
+    {
+      "epoch": 4.18,
+      "eval_loss": 0.8384572863578796,
+      "eval_runtime": 1.2927,
+      "eval_samples_per_second": 773.554,
+      "eval_steps_per_second": 12.377,
+      "step": 164000
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00024253950472849758,
+      "loss": 0.2995,
+      "step": 164500
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0002421567663560386,
+      "loss": 0.2995,
+      "step": 165000
+    },
+    {
+      "epoch": 4.21,
+      "eval_loss": 0.8305343985557556,
+      "eval_runtime": 1.3381,
+      "eval_samples_per_second": 747.307,
+      "eval_steps_per_second": 11.957,
+      "step": 165000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00024177307485114653,
+      "loss": 0.2996,
+      "step": 165500
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0002413884344098101,
+      "loss": 0.2998,
+      "step": 166000
+    },
+    {
+      "epoch": 4.23,
+      "eval_loss": 0.8391353487968445,
+      "eval_runtime": 1.2414,
+      "eval_samples_per_second": 805.533,
+      "eval_steps_per_second": 12.889,
+      "step": 166000
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00024100284923839568,
+      "loss": 0.2996,
+      "step": 166500
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0002406163235536008,
+      "loss": 0.2991,
+      "step": 167000
+    },
+    {
+      "epoch": 4.26,
+      "eval_loss": 0.8364063501358032,
+      "eval_runtime": 1.2744,
+      "eval_samples_per_second": 784.671,
+      "eval_steps_per_second": 12.555,
+      "step": 167000
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00024022886158240857,
+      "loss": 0.2985,
+      "step": 167500
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0002398404675620409,
+      "loss": 0.2994,
+      "step": 168000
+    },
+    {
+      "epoch": 4.28,
+      "eval_loss": 0.8259013891220093,
+      "eval_runtime": 1.2581,
+      "eval_samples_per_second": 794.863,
+      "eval_steps_per_second": 12.718,
+      "step": 168000
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0002394511457399126,
+      "loss": 0.2983,
+      "step": 168500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00023906090037358478,
+      "loss": 0.2977,
+      "step": 169000
+    },
+    {
+      "epoch": 4.31,
+      "eval_loss": 0.8347362875938416,
+      "eval_runtime": 1.2843,
+      "eval_samples_per_second": 778.639,
+      "eval_steps_per_second": 12.458,
+      "step": 169000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0002386697357307182,
+      "loss": 0.2979,
+      "step": 169500
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00023827765608902676,
+      "loss": 0.2989,
+      "step": 170000
+    },
+    {
+      "epoch": 4.33,
+      "eval_loss": 0.8345726728439331,
+      "eval_runtime": 1.268,
+      "eval_samples_per_second": 788.617,
+      "eval_steps_per_second": 12.618,
+      "step": 170000
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0002378846657362306,
+      "loss": 0.2997,
+      "step": 170500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00023749076897000928,
+      "loss": 0.2997,
+      "step": 171000
+    },
+    {
+      "epoch": 4.36,
+      "eval_loss": 0.8417762517929077,
+      "eval_runtime": 1.3079,
+      "eval_samples_per_second": 764.585,
+      "eval_steps_per_second": 12.233,
+      "step": 171000
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00023709597009795465,
+      "loss": 0.2982,
+      "step": 171500
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.000236700273437524,
+      "loss": 0.2975,
+      "step": 172000
+    },
+    {
+      "epoch": 4.38,
+      "eval_loss": 0.8321195244789124,
+      "eval_runtime": 1.2352,
+      "eval_samples_per_second": 809.602,
+      "eval_steps_per_second": 12.954,
+      "step": 172000
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0002363036833159925,
+      "loss": 0.2973,
+      "step": 172500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00023590620407040633,
+      "loss": 0.2988,
+      "step": 173000
+    },
+    {
+      "epoch": 4.41,
+      "eval_loss": 0.8193117380142212,
+      "eval_runtime": 1.3026,
+      "eval_samples_per_second": 767.685,
+      "eval_steps_per_second": 12.283,
+      "step": 173000
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00023550784004753471,
+      "loss": 0.2981,
+      "step": 173500
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0002351085956038229,
+      "loss": 0.2979,
+      "step": 174000
+    },
+    {
+      "epoch": 4.43,
+      "eval_loss": 0.8212565779685974,
+      "eval_runtime": 1.3056,
+      "eval_samples_per_second": 765.951,
+      "eval_steps_per_second": 12.255,
+      "step": 174000
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00023470847510534407,
+      "loss": 0.2979,
+      "step": 174500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00023430748292775188,
+      "loss": 0.2973,
+      "step": 175000
+    },
+    {
+      "epoch": 4.46,
+      "eval_loss": 0.8219558000564575,
+      "eval_runtime": 1.2557,
+      "eval_samples_per_second": 796.365,
+      "eval_steps_per_second": 12.742,
+      "step": 175000
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0002339056234562326,
+      "loss": 0.2971,
+      "step": 175500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00023350290108545694,
+      "loss": 0.2967,
+      "step": 176000
+    },
+    {
+      "epoch": 4.49,
+      "eval_loss": 0.8285974264144897,
+      "eval_runtime": 1.2416,
+      "eval_samples_per_second": 805.395,
+      "eval_steps_per_second": 12.886,
+      "step": 176000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00023309932021953238,
+      "loss": 0.2971,
+      "step": 176500
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00023269488527195446,
+      "loss": 0.2969,
+      "step": 177000
+    },
+    {
+      "epoch": 4.51,
+      "eval_loss": 0.8219101428985596,
+      "eval_runtime": 1.2379,
+      "eval_samples_per_second": 807.823,
+      "eval_steps_per_second": 12.925,
+      "step": 177000
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00023228960066555907,
+      "loss": 0.297,
+      "step": 177500
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00023188347083247365,
+      "loss": 0.2966,
+      "step": 178000
+    },
+    {
+      "epoch": 4.54,
+      "eval_loss": 0.8279162645339966,
+      "eval_runtime": 1.2407,
+      "eval_samples_per_second": 805.969,
+      "eval_steps_per_second": 12.895,
+      "step": 178000
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00023147650021406905,
+      "loss": 0.2975,
+      "step": 178500
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00023106869326091075,
+      "loss": 0.2966,
+      "step": 179000
+    },
+    {
+      "epoch": 4.56,
+      "eval_loss": 0.8254020810127258,
+      "eval_runtime": 1.2293,
+      "eval_samples_per_second": 813.461,
+      "eval_steps_per_second": 13.015,
+      "step": 179000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00023066005443271017,
+      "loss": 0.2976,
+      "step": 179500
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00023025058819827618,
+      "loss": 0.2968,
+      "step": 180000
+    },
+    {
+      "epoch": 4.59,
+      "eval_loss": 0.8309011459350586,
+      "eval_runtime": 1.2373,
+      "eval_samples_per_second": 808.182,
+      "eval_steps_per_second": 12.931,
+      "step": 180000
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00022984029903546578,
+      "loss": 0.2967,
+      "step": 180500
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00022942919143113572,
+      "loss": 0.2962,
+      "step": 181000
+    },
+    {
+      "epoch": 4.61,
+      "eval_loss": 0.8312578201293945,
+      "eval_runtime": 1.2492,
+      "eval_samples_per_second": 800.5,
+      "eval_steps_per_second": 12.808,
+      "step": 181000
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0002290172698810927,
+      "loss": 0.2971,
+      "step": 181500
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00022860453889004493,
+      "loss": 0.2968,
+      "step": 182000
+    },
+    {
+      "epoch": 4.64,
+      "eval_loss": 0.8232109546661377,
+      "eval_runtime": 1.3244,
+      "eval_samples_per_second": 755.053,
+      "eval_steps_per_second": 12.081,
+      "step": 182000
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00022819100297155235,
+      "loss": 0.2974,
+      "step": 182500
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0002277766666479774,
+      "loss": 0.2967,
+      "step": 183000
+    },
+    {
+      "epoch": 4.66,
+      "eval_loss": 0.8215346932411194,
+      "eval_runtime": 1.2987,
+      "eval_samples_per_second": 769.992,
+      "eval_steps_per_second": 12.32,
+      "step": 183000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00022736153445043595,
+      "loss": 0.2962,
+      "step": 183500
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00022694561091874706,
+      "loss": 0.2958,
+      "step": 184000
+    },
+    {
+      "epoch": 4.69,
+      "eval_loss": 0.8170773983001709,
+      "eval_runtime": 1.2678,
+      "eval_samples_per_second": 788.789,
+      "eval_steps_per_second": 12.621,
+      "step": 184000
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00022652890060138387,
+      "loss": 0.296,
+      "step": 184500
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00022611140805542366,
+      "loss": 0.2958,
+      "step": 185000
+    },
+    {
+      "epoch": 4.71,
+      "eval_loss": 0.8280390501022339,
+      "eval_runtime": 1.2658,
+      "eval_samples_per_second": 789.997,
+      "eval_steps_per_second": 12.64,
+      "step": 185000
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00022569313784649798,
+      "loss": 0.2974,
+      "step": 185500
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0002252740945487429,
+      "loss": 0.2958,
+      "step": 186000
+    },
+    {
+      "epoch": 4.74,
+      "eval_loss": 0.8221661448478699,
+      "eval_runtime": 1.2826,
+      "eval_samples_per_second": 779.639,
+      "eval_steps_per_second": 12.474,
+      "step": 186000
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00022485428274474867,
+      "loss": 0.2961,
+      "step": 186500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00022443370702551,
+      "loss": 0.2958,
+      "step": 187000
+    },
+    {
+      "epoch": 4.77,
+      "eval_loss": 0.8302921056747437,
+      "eval_runtime": 1.3594,
+      "eval_samples_per_second": 735.615,
+      "eval_steps_per_second": 11.77,
+      "step": 187000
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00022401237199037565,
+      "loss": 0.2964,
+      "step": 187500
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0002235902822469979,
+      "loss": 0.2965,
+      "step": 188000
+    },
+    {
+      "epoch": 4.79,
+      "eval_loss": 0.8213106393814087,
+      "eval_runtime": 1.2353,
+      "eval_samples_per_second": 809.533,
+      "eval_steps_per_second": 12.953,
+      "step": 188000
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00022316744241128268,
+      "loss": 0.2957,
+      "step": 188500
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00022274385710733855,
+      "loss": 0.2958,
+      "step": 189000
+    },
+    {
+      "epoch": 4.82,
+      "eval_loss": 0.8167091012001038,
+      "eval_runtime": 1.2465,
+      "eval_samples_per_second": 802.259,
+      "eval_steps_per_second": 12.836,
+      "step": 189000
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00022231953096742672,
+      "loss": 0.2974,
+      "step": 189500
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00022189446863190974,
+      "loss": 0.297,
+      "step": 190000
+    },
+    {
+      "epoch": 4.84,
+      "eval_loss": 0.8272013068199158,
+      "eval_runtime": 1.2813,
+      "eval_samples_per_second": 780.471,
+      "eval_steps_per_second": 12.488,
+      "step": 190000
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00022146867474920118,
+      "loss": 0.2961,
+      "step": 190500
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00022104215397571484,
+      "loss": 0.2959,
+      "step": 191000
+    },
+    {
+      "epoch": 4.87,
+      "eval_loss": 0.8258107900619507,
+      "eval_runtime": 1.248,
+      "eval_samples_per_second": 801.284,
+      "eval_steps_per_second": 12.821,
+      "step": 191000
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0002206149109758135,
+      "loss": 0.2954,
+      "step": 191500
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00022018695042175818,
+      "loss": 0.295,
+      "step": 192000
+    },
+    {
+      "epoch": 4.89,
+      "eval_loss": 0.8216790556907654,
+      "eval_runtime": 1.2627,
+      "eval_samples_per_second": 791.937,
+      "eval_steps_per_second": 12.671,
+      "step": 192000
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00021975827699365693,
+      "loss": 0.2949,
+      "step": 192500
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00021932889537941365,
+      "loss": 0.295,
+      "step": 193000
+    },
+    {
+      "epoch": 4.92,
+      "eval_loss": 0.8129534721374512,
+      "eval_runtime": 1.2745,
+      "eval_samples_per_second": 784.628,
+      "eval_steps_per_second": 12.554,
+      "step": 193000
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0002188988102746769,
+      "loss": 0.2949,
+      "step": 193500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0002184680263827885,
+      "loss": 0.2968,
+      "step": 194000
+    },
+    {
+      "epoch": 4.94,
+      "eval_loss": 0.8097214698791504,
+      "eval_runtime": 1.2465,
+      "eval_samples_per_second": 802.226,
+      "eval_steps_per_second": 12.836,
+      "step": 194000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00021803654841473204,
+      "loss": 0.2953,
+      "step": 194500
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00021760438108908142,
+      "loss": 0.2947,
+      "step": 195000
+    },
+    {
+      "epoch": 4.97,
+      "eval_loss": 0.806962788105011,
+      "eval_runtime": 1.2893,
+      "eval_samples_per_second": 775.619,
+      "eval_steps_per_second": 12.41,
+      "step": 195000
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0002171715291319494,
+      "loss": 0.2944,
+      "step": 195500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0002167379972769355,
+      "loss": 0.2941,
+      "step": 196000
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 0.8226540684700012,
+      "eval_runtime": 1.2611,
+      "eval_samples_per_second": 792.99,
+      "eval_steps_per_second": 12.688,
+      "step": 196000
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0002163037902650747,
+      "loss": 0.2948,
+      "step": 196500
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0002158689128447853,
+      "loss": 0.294,
+      "step": 197000
+    },
+    {
+      "epoch": 5.02,
+      "eval_loss": 0.813295841217041,
+      "eval_runtime": 1.2515,
+      "eval_samples_per_second": 799.026,
+      "eval_steps_per_second": 12.784,
+      "step": 197000
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00021543336977181704,
+      "loss": 0.2948,
+      "step": 197500
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00021499716580919933,
+      "loss": 0.2947,
+      "step": 198000
+    },
+    {
+      "epoch": 5.05,
+      "eval_loss": 0.8141849637031555,
+      "eval_runtime": 1.3465,
+      "eval_samples_per_second": 742.661,
+      "eval_steps_per_second": 11.883,
+      "step": 198000
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00021456030572718866,
+      "loss": 0.2944,
+      "step": 198500
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.000214122794303217,
+      "loss": 0.2941,
+      "step": 199000
+    },
+    {
+      "epoch": 5.07,
+      "eval_loss": 0.8159200549125671,
+      "eval_runtime": 1.3775,
+      "eval_samples_per_second": 725.952,
+      "eval_steps_per_second": 11.615,
+      "step": 199000
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00021368463632183912,
+      "loss": 0.2948,
+      "step": 199500
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00021324583657468055,
+      "loss": 0.294,
+      "step": 200000
+    },
+    {
+      "epoch": 5.1,
+      "eval_loss": 0.827436625957489,
+      "eval_runtime": 1.3211,
+      "eval_samples_per_second": 756.949,
+      "eval_steps_per_second": 12.111,
+      "step": 200000
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.000212806399860385,
+      "loss": 0.2949,
+      "step": 200500
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00021236633098456196,
+      "loss": 0.2941,
+      "step": 201000
+    },
+    {
+      "epoch": 5.12,
+      "eval_loss": 0.81947922706604,
+      "eval_runtime": 1.576,
+      "eval_samples_per_second": 634.516,
+      "eval_steps_per_second": 10.152,
+      "step": 201000
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002119256347597342,
+      "loss": 0.295,
+      "step": 201500
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.000211484316005285,
+      "loss": 0.2938,
+      "step": 202000
+    },
+    {
+      "epoch": 5.15,
+      "eval_loss": 0.8285008668899536,
+      "eval_runtime": 1.3278,
+      "eval_samples_per_second": 753.153,
+      "eval_steps_per_second": 12.05,
+      "step": 202000
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00021104237954740554,
+      "loss": 0.2936,
+      "step": 202500
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00021059983021904215,
+      "loss": 0.2934,
+      "step": 203000
+    },
+    {
+      "epoch": 5.17,
+      "eval_loss": 0.8159323930740356,
+      "eval_runtime": 1.3067,
+      "eval_samples_per_second": 765.291,
+      "eval_steps_per_second": 12.245,
+      "step": 203000
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00021015667285984336,
+      "loss": 0.2934,
+      "step": 203500
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00020971291231610707,
+      "loss": 0.2932,
+      "step": 204000
+    },
+    {
+      "epoch": 5.2,
+      "eval_loss": 0.8072566986083984,
+      "eval_runtime": 1.4047,
+      "eval_samples_per_second": 711.906,
+      "eval_steps_per_second": 11.39,
+      "step": 204000
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002092685534407274,
+      "loss": 0.2938,
+      "step": 204500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00020882360109314197,
+      "loss": 0.2946,
+      "step": 205000
+    },
+    {
+      "epoch": 5.22,
+      "eval_loss": 0.8255318403244019,
+      "eval_runtime": 1.3095,
+      "eval_samples_per_second": 763.663,
+      "eval_steps_per_second": 12.219,
+      "step": 205000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0002083780601392783,
+      "loss": 0.293,
+      "step": 205500
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0002079319354515008,
+      "loss": 0.2939,
+      "step": 206000
+    },
+    {
+      "epoch": 5.25,
+      "eval_loss": 0.8249548077583313,
+      "eval_runtime": 1.2456,
+      "eval_samples_per_second": 802.804,
+      "eval_steps_per_second": 12.845,
+      "step": 206000
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00020748523190855772,
+      "loss": 0.2929,
+      "step": 206500
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002070379543955273,
+      "loss": 0.2933,
+      "step": 207000
+    },
+    {
+      "epoch": 5.28,
+      "eval_loss": 0.8215068578720093,
+      "eval_runtime": 1.3095,
+      "eval_samples_per_second": 763.673,
+      "eval_steps_per_second": 12.219,
+      "step": 207000
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00020659010780376487,
+      "loss": 0.2928,
+      "step": 207500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00020614169703084896,
+      "loss": 0.2927,
+      "step": 208000
+    },
+    {
+      "epoch": 5.3,
+      "eval_loss": 0.8152709007263184,
+      "eval_runtime": 1.2719,
+      "eval_samples_per_second": 786.201,
+      "eval_steps_per_second": 12.579,
+      "step": 208000
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0002056927269805279,
+      "loss": 0.2928,
+      "step": 208500
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00020524320256266635,
+      "loss": 0.2931,
+      "step": 209000
+    },
+    {
+      "epoch": 5.33,
+      "eval_loss": 0.8284364938735962,
+      "eval_runtime": 1.322,
+      "eval_samples_per_second": 756.405,
+      "eval_steps_per_second": 12.102,
+      "step": 209000
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0002047931286931912,
+      "loss": 0.2923,
+      "step": 209500
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00020434251029403824,
+      "loss": 0.2928,
+      "step": 210000
+    },
+    {
+      "epoch": 5.35,
+      "eval_loss": 0.8203576803207397,
+      "eval_runtime": 1.366,
+      "eval_samples_per_second": 732.046,
+      "eval_steps_per_second": 11.713,
+      "step": 210000
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00020389135229309803,
+      "loss": 0.2923,
+      "step": 210500
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00020343965962416229,
+      "loss": 0.2923,
+      "step": 211000
+    },
+    {
+      "epoch": 5.38,
+      "eval_loss": 0.8264955878257751,
+      "eval_runtime": 1.329,
+      "eval_samples_per_second": 752.449,
+      "eval_steps_per_second": 12.039,
+      "step": 211000
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00020298743722686958,
+      "loss": 0.2922,
+      "step": 211500
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002025346900466516,
+      "loss": 0.2925,
+      "step": 212000
+    },
+    {
+      "epoch": 5.4,
+      "eval_loss": 0.826949954032898,
+      "eval_runtime": 1.2977,
+      "eval_samples_per_second": 770.59,
+      "eval_steps_per_second": 12.329,
+      "step": 212000
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002020814230346791,
+      "loss": 0.2932,
+      "step": 212500
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00020162764114780733,
+      "loss": 0.2926,
+      "step": 213000
+    },
+    {
+      "epoch": 5.43,
+      "eval_loss": 0.8337197303771973,
+      "eval_runtime": 1.293,
+      "eval_samples_per_second": 773.389,
+      "eval_steps_per_second": 12.374,
+      "step": 213000
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002011733493485224,
+      "loss": 0.2923,
+      "step": 213500
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00020071855260488664,
+      "loss": 0.292,
+      "step": 214000
+    },
+    {
+      "epoch": 5.45,
+      "eval_loss": 0.8254857063293457,
+      "eval_runtime": 1.2323,
+      "eval_samples_per_second": 811.466,
+      "eval_steps_per_second": 12.983,
+      "step": 214000
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002002632558904843,
+      "loss": 0.2919,
+      "step": 214500
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00019980746418436736,
+      "loss": 0.292,
+      "step": 215000
+    },
+    {
+      "epoch": 5.48,
+      "eval_loss": 0.8224061131477356,
+      "eval_runtime": 1.2655,
+      "eval_samples_per_second": 790.229,
+      "eval_steps_per_second": 12.644,
+      "step": 215000
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00019935118247100088,
+      "loss": 0.2919,
+      "step": 215500
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00019889441574020864,
+      "loss": 0.2915,
+      "step": 216000
+    },
+    {
+      "epoch": 5.5,
+      "eval_loss": 0.8216863870620728,
+      "eval_runtime": 1.2862,
+      "eval_samples_per_second": 777.468,
+      "eval_steps_per_second": 12.439,
+      "step": 216000
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0001984371689871183,
+      "loss": 0.2918,
+      "step": 216500
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00019797944721210725,
+      "loss": 0.2916,
+      "step": 217000
+    },
+    {
+      "epoch": 5.53,
+      "eval_loss": 0.8251473307609558,
+      "eval_runtime": 1.3615,
+      "eval_samples_per_second": 734.51,
+      "eval_steps_per_second": 11.752,
+      "step": 217000
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00019752125542074736,
+      "loss": 0.2913,
+      "step": 217500
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00019706259862375074,
+      "loss": 0.291,
+      "step": 218000
+    },
+    {
+      "epoch": 5.56,
+      "eval_loss": 0.8243631720542908,
+      "eval_runtime": 1.2451,
+      "eval_samples_per_second": 803.116,
+      "eval_steps_per_second": 12.85,
+      "step": 218000
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00019660348183691453,
+      "loss": 0.292,
+      "step": 218500
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0001961439100810664,
+      "loss": 0.2918,
+      "step": 219000
+    },
+    {
+      "epoch": 5.58,
+      "eval_loss": 0.8229220509529114,
+      "eval_runtime": 1.3199,
+      "eval_samples_per_second": 757.608,
+      "eval_steps_per_second": 12.122,
+      "step": 219000
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00019568388838200952,
+      "loss": 0.2916,
+      "step": 219500
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00019522342177046744,
+      "loss": 0.2911,
+      "step": 220000
+    },
+    {
+      "epoch": 5.61,
+      "eval_loss": 0.824487566947937,
+      "eval_runtime": 1.2486,
+      "eval_samples_per_second": 800.922,
+      "eval_steps_per_second": 12.815,
+      "step": 220000
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00019476251528202922,
+      "loss": 0.2921,
+      "step": 220500
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0001943011739570944,
+      "loss": 0.2911,
+      "step": 221000
+    },
+    {
+      "epoch": 5.63,
+      "eval_loss": 0.8201115727424622,
+      "eval_runtime": 1.4717,
+      "eval_samples_per_second": 679.492,
+      "eval_steps_per_second": 10.872,
+      "step": 221000
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00019383940284081774,
+      "loss": 0.2912,
+      "step": 221500
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00019337720698305431,
+      "loss": 0.2913,
+      "step": 222000
+    },
+    {
+      "epoch": 5.66,
+      "eval_loss": 0.8082127571105957,
+      "eval_runtime": 1.2848,
+      "eval_samples_per_second": 778.358,
+      "eval_steps_per_second": 12.454,
+      "step": 222000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0001929145914383038,
+      "loss": 0.2914,
+      "step": 222500
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00019245156126565586,
+      "loss": 0.2912,
+      "step": 223000
+    },
+    {
+      "epoch": 5.68,
+      "eval_loss": 0.8193914890289307,
+      "eval_runtime": 1.3731,
+      "eval_samples_per_second": 728.26,
+      "eval_steps_per_second": 11.652,
+      "step": 223000
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00019198812152873416,
+      "loss": 0.291,
+      "step": 223500
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00019152427729564144,
+      "loss": 0.2908,
+      "step": 224000
+    },
+    {
+      "epoch": 5.71,
+      "eval_loss": 0.8259610533714294,
+      "eval_runtime": 1.2322,
+      "eval_samples_per_second": 811.569,
+      "eval_steps_per_second": 12.985,
+      "step": 224000
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00019106003363890395,
+      "loss": 0.2906,
+      "step": 224500
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00019059539563541584,
+      "loss": 0.291,
+      "step": 225000
+    },
+    {
+      "epoch": 5.73,
+      "eval_loss": 0.8225913643836975,
+      "eval_runtime": 1.2796,
+      "eval_samples_per_second": 781.52,
+      "eval_steps_per_second": 12.504,
+      "step": 225000
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.000190130368366384,
+      "loss": 0.2908,
+      "step": 225500
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00018966495691727207,
+      "loss": 0.2908,
+      "step": 226000
+    },
+    {
+      "epoch": 5.76,
+      "eval_loss": 0.8231199979782104,
+      "eval_runtime": 1.2853,
+      "eval_samples_per_second": 778.049,
+      "eval_steps_per_second": 12.449,
+      "step": 226000
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0001891991663777451,
+      "loss": 0.2902,
+      "step": 226500
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00018873300184161387,
+      "loss": 0.2903,
+      "step": 227000
+    },
+    {
+      "epoch": 5.79,
+      "eval_loss": 0.8100917935371399,
+      "eval_runtime": 1.3376,
+      "eval_samples_per_second": 747.632,
+      "eval_steps_per_second": 11.962,
+      "step": 227000
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00018826646840677894,
+      "loss": 0.2902,
+      "step": 227500
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00018779957117517532,
+      "loss": 0.2917,
+      "step": 228000
+    },
+    {
+      "epoch": 5.81,
+      "eval_loss": 0.8147779107093811,
+      "eval_runtime": 1.2993,
+      "eval_samples_per_second": 769.654,
+      "eval_steps_per_second": 12.314,
+      "step": 228000
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00018733231525271625,
+      "loss": 0.2917,
+      "step": 228500
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00018686470574923766,
+      "loss": 0.2915,
+      "step": 229000
+    },
+    {
+      "epoch": 5.84,
+      "eval_loss": 0.8212478160858154,
+      "eval_runtime": 1.3093,
+      "eval_samples_per_second": 763.781,
+      "eval_steps_per_second": 12.22,
+      "step": 229000
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00018639674777844224,
+      "loss": 0.2903,
+      "step": 229500
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00018592844645784327,
+      "loss": 0.2901,
+      "step": 230000
+    },
+    {
+      "epoch": 5.86,
+      "eval_loss": 0.812626302242279,
+      "eval_runtime": 1.3081,
+      "eval_samples_per_second": 764.442,
+      "eval_steps_per_second": 12.231,
+      "step": 230000
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00018545980690870903,
+      "loss": 0.2899,
+      "step": 230500
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00018499083425600648,
+      "loss": 0.2898,
+      "step": 231000
+    },
+    {
+      "epoch": 5.89,
+      "eval_loss": 0.8181861639022827,
+      "eval_runtime": 1.2638,
+      "eval_samples_per_second": 791.236,
+      "eval_steps_per_second": 12.66,
+      "step": 231000
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00018452153362834552,
+      "loss": 0.2896,
+      "step": 231500
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00018405191015792254,
+      "loss": 0.29,
+      "step": 232000
+    },
+    {
+      "epoch": 5.91,
+      "eval_loss": 0.8149720430374146,
+      "eval_runtime": 1.2755,
+      "eval_samples_per_second": 783.985,
+      "eval_steps_per_second": 12.544,
+      "step": 232000
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0001835819689804646,
+      "loss": 0.2892,
+      "step": 232500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0001831117152351732,
+      "loss": 0.2905,
+      "step": 233000
+    },
+    {
+      "epoch": 5.94,
+      "eval_loss": 0.8126420378684998,
+      "eval_runtime": 1.2766,
+      "eval_samples_per_second": 783.351,
+      "eval_steps_per_second": 12.534,
+      "step": 233000
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00018264115406466778,
+      "loss": 0.2893,
+      "step": 233500
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00018217029061493007,
+      "loss": 0.2894,
+      "step": 234000
+    },
+    {
+      "epoch": 5.96,
+      "eval_loss": 0.82079017162323,
+      "eval_runtime": 1.3447,
+      "eval_samples_per_second": 743.651,
+      "eval_steps_per_second": 11.898,
+      "step": 234000
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00018169913003524717,
+      "loss": 0.2893,
+      "step": 234500
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00018122767747815594,
+      "loss": 0.2894,
+      "step": 235000
+    },
+    {
+      "epoch": 5.99,
+      "eval_loss": 0.8262202739715576,
+      "eval_runtime": 1.2956,
+      "eval_samples_per_second": 771.835,
+      "eval_steps_per_second": 12.349,
+      "step": 235000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00018075593809938574,
+      "loss": 0.2898,
+      "step": 235500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00018028391705780295,
+      "loss": 0.2899,
+      "step": 236000
+    },
+    {
+      "epoch": 6.01,
+      "eval_loss": 0.8133281469345093,
+      "eval_runtime": 1.2572,
+      "eval_samples_per_second": 795.402,
+      "eval_steps_per_second": 12.726,
+      "step": 236000
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0001798116195153541,
+      "loss": 0.2895,
+      "step": 236500
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001793390506370094,
+      "loss": 0.2891,
+      "step": 237000
+    },
+    {
+      "epoch": 6.04,
+      "eval_loss": 0.8038508296012878,
+      "eval_runtime": 1.2713,
+      "eval_samples_per_second": 786.62,
+      "eval_steps_per_second": 12.586,
+      "step": 237000
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00017886621559070638,
+      "loss": 0.289,
+      "step": 237500
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00017839311954729337,
+      "loss": 0.2887,
+      "step": 238000
+    },
+    {
+      "epoch": 6.07,
+      "eval_loss": 0.8181630969047546,
+      "eval_runtime": 1.3231,
+      "eval_samples_per_second": 755.824,
+      "eval_steps_per_second": 12.093,
+      "step": 238000
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00017791976768047292,
+      "loss": 0.2891,
+      "step": 238500
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00017744616516674518,
+      "loss": 0.2889,
+      "step": 239000
+    },
+    {
+      "epoch": 6.09,
+      "eval_loss": 0.806620717048645,
+      "eval_runtime": 1.2977,
+      "eval_samples_per_second": 770.587,
+      "eval_steps_per_second": 12.329,
+      "step": 239000
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00017697231718535132,
+      "loss": 0.2888,
+      "step": 239500
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00017649822891821707,
+      "loss": 0.2889,
+      "step": 240000
+    },
+    {
+      "epoch": 6.12,
+      "eval_loss": 0.8128913640975952,
+      "eval_runtime": 1.2865,
+      "eval_samples_per_second": 777.28,
+      "eval_steps_per_second": 12.436,
+      "step": 240000
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00017602390554989563,
+      "loss": 0.2885,
+      "step": 240500
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001755493522675115,
+      "loss": 0.2899,
+      "step": 241000
+    },
+    {
+      "epoch": 6.14,
+      "eval_loss": 0.8203925490379333,
+      "eval_runtime": 1.3437,
+      "eval_samples_per_second": 744.194,
+      "eval_steps_per_second": 11.907,
+      "step": 241000
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00017507457426070317,
+      "loss": 0.2894,
+      "step": 241500
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00017459957672156704,
+      "loss": 0.2894,
+      "step": 242000
+    },
+    {
+      "epoch": 6.17,
+      "eval_loss": 0.8142299652099609,
+      "eval_runtime": 1.2947,
+      "eval_samples_per_second": 772.387,
+      "eval_steps_per_second": 12.358,
+      "step": 242000
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00017412436484459998,
+      "loss": 0.2896,
+      "step": 242500
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00017364894382664297,
+      "loss": 0.2893,
+      "step": 243000
+    },
+    {
+      "epoch": 6.19,
+      "eval_loss": 0.8166674971580505,
+      "eval_runtime": 1.2594,
+      "eval_samples_per_second": 794.0,
+      "eval_steps_per_second": 12.704,
+      "step": 243000
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00017317331886682408,
+      "loss": 0.2887,
+      "step": 243500
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0001726974951665017,
+      "loss": 0.2883,
+      "step": 244000
+    },
+    {
+      "epoch": 6.22,
+      "eval_loss": 0.8151687979698181,
+      "eval_runtime": 1.3439,
+      "eval_samples_per_second": 744.094,
+      "eval_steps_per_second": 11.905,
+      "step": 244000
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001722214779292076,
+      "loss": 0.2884,
+      "step": 244500
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00017174527236058998,
+      "loss": 0.2882,
+      "step": 245000
+    },
+    {
+      "epoch": 6.24,
+      "eval_loss": 0.8129110336303711,
+      "eval_runtime": 1.2934,
+      "eval_samples_per_second": 773.156,
+      "eval_steps_per_second": 12.371,
+      "step": 245000
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00017126888366835662,
+      "loss": 0.2879,
+      "step": 245500
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.000170792317062218,
+      "loss": 0.2883,
+      "step": 246000
+    },
+    {
+      "epoch": 6.27,
+      "eval_loss": 0.8146284222602844,
+      "eval_runtime": 1.2384,
+      "eval_samples_per_second": 807.487,
+      "eval_steps_per_second": 12.92,
+      "step": 246000
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00017031557775383011,
+      "loss": 0.2883,
+      "step": 246500
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001698386709567377,
+      "loss": 0.2886,
+      "step": 247000
+    },
+    {
+      "epoch": 6.29,
+      "eval_loss": 0.8157382607460022,
+      "eval_runtime": 1.2484,
+      "eval_samples_per_second": 801.025,
+      "eval_steps_per_second": 12.816,
+      "step": 247000
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0001693616018863171,
+      "loss": 0.2891,
+      "step": 247500
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00016888437575971913,
+      "loss": 0.2886,
+      "step": 248000
+    },
+    {
+      "epoch": 6.32,
+      "eval_loss": 0.8171519041061401,
+      "eval_runtime": 1.2819,
+      "eval_samples_per_second": 780.066,
+      "eval_steps_per_second": 12.481,
+      "step": 248000
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00016840699779581238,
+      "loss": 0.2879,
+      "step": 248500
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00016792947321512573,
+      "loss": 0.2885,
+      "step": 249000
+    },
+    {
+      "epoch": 6.35,
+      "eval_loss": 0.82103031873703,
+      "eval_runtime": 1.2758,
+      "eval_samples_per_second": 783.796,
+      "eval_steps_per_second": 12.541,
+      "step": 249000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00016745180723979144,
+      "loss": 0.2879,
+      "step": 249500
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00016697400509348818,
+      "loss": 0.2886,
+      "step": 250000
+    },
+    {
+      "epoch": 6.37,
+      "eval_loss": 0.8212683200836182,
+      "eval_runtime": 1.2677,
+      "eval_samples_per_second": 788.807,
+      "eval_steps_per_second": 12.621,
+      "step": 250000
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00016649607200138356,
+      "loss": 0.2877,
+      "step": 250500
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00016601801319007743,
+      "loss": 0.2877,
+      "step": 251000
+    },
+    {
+      "epoch": 6.4,
+      "eval_loss": 0.810385525226593,
+      "eval_runtime": 1.2931,
+      "eval_samples_per_second": 773.361,
+      "eval_steps_per_second": 12.374,
+      "step": 251000
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00016553983388754428,
+      "loss": 0.2872,
+      "step": 251500
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00016506153932307636,
+      "loss": 0.2872,
+      "step": 252000
+    },
+    {
+      "epoch": 6.42,
+      "eval_loss": 0.8114423155784607,
+      "eval_runtime": 1.2618,
+      "eval_samples_per_second": 792.538,
+      "eval_steps_per_second": 12.681,
+      "step": 252000
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00016458313472722638,
+      "loss": 0.2874,
+      "step": 252500
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00016410462533175045,
+      "loss": 0.2871,
+      "step": 253000
+    },
+    {
+      "epoch": 6.45,
+      "eval_loss": 0.814827024936676,
+      "eval_runtime": 1.3254,
+      "eval_samples_per_second": 754.465,
+      "eval_steps_per_second": 12.071,
+      "step": 253000
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00016362601636955049,
+      "loss": 0.288,
+      "step": 253500
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00016314731307461754,
+      "loss": 0.2875,
+      "step": 254000
+    },
+    {
+      "epoch": 6.47,
+      "eval_loss": 0.812713623046875,
+      "eval_runtime": 1.2619,
+      "eval_samples_per_second": 792.485,
+      "eval_steps_per_second": 12.68,
+      "step": 254000
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001626685206819742,
+      "loss": 0.2874,
+      "step": 254500
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001621896444276172,
+      "loss": 0.287,
+      "step": 255000
+    },
+    {
+      "epoch": 6.5,
+      "eval_loss": 0.820074200630188,
+      "eval_runtime": 1.2181,
+      "eval_samples_per_second": 820.971,
+      "eval_steps_per_second": 13.136,
+      "step": 255000
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00016171068954846067,
+      "loss": 0.2865,
+      "step": 255500
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00016123166128227835,
+      "loss": 0.2869,
+      "step": 256000
+    },
+    {
+      "epoch": 6.52,
+      "eval_loss": 0.8100768327713013,
+      "eval_runtime": 1.2652,
+      "eval_samples_per_second": 790.379,
+      "eval_steps_per_second": 12.646,
+      "step": 256000
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001607525648676467,
+      "loss": 0.2868,
+      "step": 256500
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001602734055438873,
+      "loss": 0.2868,
+      "step": 257000
+    },
+    {
+      "epoch": 6.55,
+      "eval_loss": 0.814249575138092,
+      "eval_runtime": 1.2757,
+      "eval_samples_per_second": 783.912,
+      "eval_steps_per_second": 12.543,
+      "step": 257000
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00015979418855100963,
+      "loss": 0.2869,
+      "step": 257500
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00015931491912965417,
+      "loss": 0.2869,
+      "step": 258000
+    },
+    {
+      "epoch": 6.58,
+      "eval_loss": 0.815778374671936,
+      "eval_runtime": 1.2351,
+      "eval_samples_per_second": 809.649,
+      "eval_steps_per_second": 12.954,
+      "step": 258000
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001588356025210344,
+      "loss": 0.2866,
+      "step": 258500
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00015835624396688,
+      "loss": 0.2868,
+      "step": 259000
+    },
+    {
+      "epoch": 6.6,
+      "eval_loss": 0.81247878074646,
+      "eval_runtime": 1.3012,
+      "eval_samples_per_second": 768.517,
+      "eval_steps_per_second": 12.296,
+      "step": 259000
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00015787684870937924,
+      "loss": 0.2862,
+      "step": 259500
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00015739742199112196,
+      "loss": 0.2865,
+      "step": 260000
+    },
+    {
+      "epoch": 6.63,
+      "eval_loss": 0.816739559173584,
+      "eval_runtime": 1.3361,
+      "eval_samples_per_second": 748.473,
+      "eval_steps_per_second": 11.976,
+      "step": 260000
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00015691796905504187,
+      "loss": 0.2862,
+      "step": 260500
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00015643849514435944,
+      "loss": 0.2871,
+      "step": 261000
+    },
+    {
+      "epoch": 6.65,
+      "eval_loss": 0.8193889260292053,
+      "eval_runtime": 1.2087,
+      "eval_samples_per_second": 827.32,
+      "eval_steps_per_second": 13.237,
+      "step": 261000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00015595900550252463,
+      "loss": 0.2867,
+      "step": 261500
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00015547950537315926,
+      "loss": 0.2863,
+      "step": 262000
+    },
+    {
+      "epoch": 6.68,
+      "eval_loss": 0.8058978319168091,
+      "eval_runtime": 1.2707,
+      "eval_samples_per_second": 786.99,
+      "eval_steps_per_second": 12.592,
+      "step": 262000
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00015499999999999997,
+      "loss": 0.2862,
+      "step": 262500
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00015452049462684068,
+      "loss": 0.2864,
+      "step": 263000
+    },
+    {
+      "epoch": 6.7,
+      "eval_loss": 0.8194664120674133,
+      "eval_runtime": 1.2245,
+      "eval_samples_per_second": 816.647,
+      "eval_steps_per_second": 13.066,
+      "step": 263000
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00015404099449747535,
+      "loss": 0.2861,
+      "step": 263500
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001535615048556405,
+      "loss": 0.2863,
+      "step": 264000
+    },
+    {
+      "epoch": 6.73,
+      "eval_loss": 0.8099100589752197,
+      "eval_runtime": 1.2591,
+      "eval_samples_per_second": 794.217,
+      "eval_steps_per_second": 12.707,
+      "step": 264000
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001530820309449581,
+      "loss": 0.2861,
+      "step": 264500
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00015260257800887798,
+      "loss": 0.2868,
+      "step": 265000
+    },
+    {
+      "epoch": 6.75,
+      "eval_loss": 0.8127309679985046,
+      "eval_runtime": 1.3337,
+      "eval_samples_per_second": 749.803,
+      "eval_steps_per_second": 11.997,
+      "step": 265000
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.0001521231512906207,
+      "loss": 0.2868,
+      "step": 265500
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00015164375603311998,
+      "loss": 0.2863,
+      "step": 266000
+    },
+    {
+      "epoch": 6.78,
+      "eval_loss": 0.806861162185669,
+      "eval_runtime": 1.2726,
+      "eval_samples_per_second": 785.768,
+      "eval_steps_per_second": 12.572,
+      "step": 266000
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00015116439747896553,
+      "loss": 0.2856,
+      "step": 266500
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00015068508087034578,
+      "loss": 0.2854,
+      "step": 267000
+    },
+    {
+      "epoch": 6.8,
+      "eval_loss": 0.8032740354537964,
+      "eval_runtime": 1.28,
+      "eval_samples_per_second": 781.242,
+      "eval_steps_per_second": 12.5,
+      "step": 267000
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00015020581144899027,
+      "loss": 0.2857,
+      "step": 267500
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0001497265944561127,
+      "loss": 0.2855,
+      "step": 268000
+    },
+    {
+      "epoch": 6.83,
+      "eval_loss": 0.8096611499786377,
+      "eval_runtime": 1.2917,
+      "eval_samples_per_second": 774.163,
+      "eval_steps_per_second": 12.387,
+      "step": 268000
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00014924743513235327,
+      "loss": 0.2856,
+      "step": 268500
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0001487683387177216,
+      "loss": 0.2864,
+      "step": 269000
+    },
+    {
+      "epoch": 6.86,
+      "eval_loss": 0.8095938563346863,
+      "eval_runtime": 1.2896,
+      "eval_samples_per_second": 775.42,
+      "eval_steps_per_second": 12.407,
+      "step": 269000
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00014828931045153928,
+      "loss": 0.2857,
+      "step": 269500
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00014781035557238272,
+      "loss": 0.2865,
+      "step": 270000
+    },
+    {
+      "epoch": 6.88,
+      "eval_loss": 0.8193797469139099,
+      "eval_runtime": 1.2225,
+      "eval_samples_per_second": 817.974,
+      "eval_steps_per_second": 13.088,
+      "step": 270000
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00014733147931802578,
+      "loss": 0.2858,
+      "step": 270500
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00014685268692538238,
+      "loss": 0.2852,
+      "step": 271000
+    },
+    {
+      "epoch": 6.91,
+      "eval_loss": 0.8104480504989624,
+      "eval_runtime": 1.2885,
+      "eval_samples_per_second": 776.067,
+      "eval_steps_per_second": 12.417,
+      "step": 271000
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00014637398363044946,
+      "loss": 0.2851,
+      "step": 271500
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00014589537466824955,
+      "loss": 0.2852,
+      "step": 272000
+    },
+    {
+      "epoch": 6.93,
+      "eval_loss": 0.8213782906532288,
+      "eval_runtime": 1.2391,
+      "eval_samples_per_second": 807.026,
+      "eval_steps_per_second": 12.912,
+      "step": 272000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00014541686527277356,
+      "loss": 0.285,
+      "step": 272500
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00014493846067692358,
+      "loss": 0.2848,
+      "step": 273000
+    },
+    {
+      "epoch": 6.96,
+      "eval_loss": 0.8104705810546875,
+      "eval_runtime": 1.1998,
+      "eval_samples_per_second": 833.443,
+      "eval_steps_per_second": 13.335,
+      "step": 273000
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00014446016611245567,
+      "loss": 0.2855,
+      "step": 273500
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00014398198680992252,
+      "loss": 0.2857,
+      "step": 274000
+    },
+    {
+      "epoch": 6.98,
+      "eval_loss": 0.8124169707298279,
+      "eval_runtime": 1.3472,
+      "eval_samples_per_second": 742.261,
+      "eval_steps_per_second": 11.876,
+      "step": 274000
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00014350392799861636,
+      "loss": 0.286,
+      "step": 274500
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.0001430259949065118,
+      "loss": 0.2849,
+      "step": 275000
+    },
+    {
+      "epoch": 7.01,
+      "eval_loss": 0.8163594007492065,
+      "eval_runtime": 1.223,
+      "eval_samples_per_second": 817.633,
+      "eval_steps_per_second": 13.082,
+      "step": 275000
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.0001425481927602085,
+      "loss": 0.2846,
+      "step": 275500
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.0001420705267848743,
+      "loss": 0.2848,
+      "step": 276000
+    },
+    {
+      "epoch": 7.03,
+      "eval_loss": 0.8182600736618042,
+      "eval_runtime": 1.2318,
+      "eval_samples_per_second": 811.808,
+      "eval_steps_per_second": 12.989,
+      "step": 276000
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00014159300220418757,
+      "loss": 0.2852,
+      "step": 276500
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.0001411156242402808,
+      "loss": 0.2847,
+      "step": 277000
+    },
+    {
+      "epoch": 7.06,
+      "eval_loss": 0.8188433647155762,
+      "eval_runtime": 1.2733,
+      "eval_samples_per_second": 785.37,
+      "eval_steps_per_second": 12.566,
+      "step": 277000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.0001406383981136829,
+      "loss": 0.2844,
+      "step": 277500
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00014016132904326226,
+      "loss": 0.2846,
+      "step": 278000
+    },
+    {
+      "epoch": 7.08,
+      "eval_loss": 0.8136109709739685,
+      "eval_runtime": 1.2428,
+      "eval_samples_per_second": 804.644,
+      "eval_steps_per_second": 12.874,
+      "step": 278000
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.00013968442224616989,
+      "loss": 0.2845,
+      "step": 278500
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00013920768293778195,
+      "loss": 0.2847,
+      "step": 279000
+    },
+    {
+      "epoch": 7.11,
+      "eval_loss": 0.8128536343574524,
+      "eval_runtime": 1.3248,
+      "eval_samples_per_second": 754.825,
+      "eval_steps_per_second": 12.077,
+      "step": 279000
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.00013873111633164336,
+      "loss": 0.2843,
+      "step": 279500
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00013825472763941,
+      "loss": 0.2845,
+      "step": 280000
+    },
+    {
+      "epoch": 7.14,
+      "eval_loss": 0.8165821433067322,
+      "eval_runtime": 1.2675,
+      "eval_samples_per_second": 788.963,
+      "eval_steps_per_second": 12.623,
+      "step": 280000
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.00013777852207079235,
+      "loss": 0.2848,
+      "step": 280500
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.00013730250483349825,
+      "loss": 0.2836,
+      "step": 281000
+    },
+    {
+      "epoch": 7.16,
+      "eval_loss": 0.8174564242362976,
+      "eval_runtime": 1.3231,
+      "eval_samples_per_second": 755.81,
+      "eval_steps_per_second": 12.093,
+      "step": 281000
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.00013682668113317584,
+      "loss": 0.2841,
+      "step": 281500
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.00013635105617335703,
+      "loss": 0.2839,
+      "step": 282000
+    },
+    {
+      "epoch": 7.19,
+      "eval_loss": 0.8130354881286621,
+      "eval_runtime": 1.2454,
+      "eval_samples_per_second": 802.941,
+      "eval_steps_per_second": 12.847,
+      "step": 282000
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.00013587563515539996,
+      "loss": 0.284,
+      "step": 282500
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00013540042327843296,
+      "loss": 0.284,
+      "step": 283000
+    },
+    {
+      "epoch": 7.21,
+      "eval_loss": 0.8058344721794128,
+      "eval_runtime": 1.2695,
+      "eval_samples_per_second": 787.692,
+      "eval_steps_per_second": 12.603,
+      "step": 283000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.00013492542573929678,
+      "loss": 0.284,
+      "step": 283500
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.00013445064773248846,
+      "loss": 0.2839,
+      "step": 284000
+    },
+    {
+      "epoch": 7.24,
+      "eval_loss": 0.8160645961761475,
+      "eval_runtime": 1.2481,
+      "eval_samples_per_second": 801.208,
+      "eval_steps_per_second": 12.819,
+      "step": 284000
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.00013397609445010432,
+      "loss": 0.2843,
+      "step": 284500
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.00013350177108178288,
+      "loss": 0.2842,
+      "step": 285000
+    },
+    {
+      "epoch": 7.26,
+      "eval_loss": 0.8231968879699707,
+      "eval_runtime": 1.3325,
+      "eval_samples_per_second": 750.455,
+      "eval_steps_per_second": 12.007,
+      "step": 285000
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.00013302768281464863,
+      "loss": 0.2839,
+      "step": 285500
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.0001325538348332548,
+      "loss": 0.2835,
+      "step": 286000
+    },
+    {
+      "epoch": 7.29,
+      "eval_loss": 0.8185608386993408,
+      "eval_runtime": 1.2436,
+      "eval_samples_per_second": 804.112,
+      "eval_steps_per_second": 12.866,
+      "step": 286000
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.00013208023231952706,
+      "loss": 0.2839,
+      "step": 286500
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.0001316068804527066,
+      "loss": 0.2837,
+      "step": 287000
+    },
+    {
+      "epoch": 7.31,
+      "eval_loss": 0.8180215954780579,
+      "eval_runtime": 1.2676,
+      "eval_samples_per_second": 788.901,
+      "eval_steps_per_second": 12.622,
+      "step": 287000
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.00013113378440929353,
+      "loss": 0.2841,
+      "step": 287500
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.00013066094936299056,
+      "loss": 0.2835,
+      "step": 288000
+    },
+    {
+      "epoch": 7.34,
+      "eval_loss": 0.8164823651313782,
+      "eval_runtime": 1.3849,
+      "eval_samples_per_second": 722.072,
+      "eval_steps_per_second": 11.553,
+      "step": 288000
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.00013018838048464582,
+      "loss": 0.2834,
+      "step": 288500
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.00012971608294219702,
+      "loss": 0.2835,
+      "step": 289000
+    },
+    {
+      "epoch": 7.37,
+      "eval_loss": 0.8121501207351685,
+      "eval_runtime": 1.2558,
+      "eval_samples_per_second": 796.286,
+      "eval_steps_per_second": 12.741,
+      "step": 289000
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.00012924406190061423,
+      "loss": 0.2835,
+      "step": 289500
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.0001287723225218441,
+      "loss": 0.2832,
+      "step": 290000
+    },
+    {
+      "epoch": 7.39,
+      "eval_loss": 0.8192352056503296,
+      "eval_runtime": 1.3002,
+      "eval_samples_per_second": 769.098,
+      "eval_steps_per_second": 12.306,
+      "step": 290000
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.00012830086996475274,
+      "loss": 0.2829,
+      "step": 290500
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00012782970938506988,
+      "loss": 0.2829,
+      "step": 291000
+    },
+    {
+      "epoch": 7.42,
+      "eval_loss": 0.8085091710090637,
+      "eval_runtime": 1.2838,
+      "eval_samples_per_second": 778.958,
+      "eval_steps_per_second": 12.463,
+      "step": 291000
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.00012735884593533222,
+      "loss": 0.2828,
+      "step": 291500
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.0001268882847648268,
+      "loss": 0.2827,
+      "step": 292000
+    },
+    {
+      "epoch": 7.44,
+      "eval_loss": 0.8085604310035706,
+      "eval_runtime": 1.2676,
+      "eval_samples_per_second": 788.867,
+      "eval_steps_per_second": 12.622,
+      "step": 292000
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.00012641803101953535,
+      "loss": 0.2829,
+      "step": 292500
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00012594808984207743,
+      "loss": 0.2829,
+      "step": 293000
+    },
+    {
+      "epoch": 7.47,
+      "eval_loss": 0.8101683855056763,
+      "eval_runtime": 1.2565,
+      "eval_samples_per_second": 795.849,
+      "eval_steps_per_second": 12.734,
+      "step": 293000
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00012547846637165445,
+      "loss": 0.2826,
+      "step": 293500
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.00012500916574399346,
+      "loss": 0.2829,
+      "step": 294000
+    },
+    {
+      "epoch": 7.49,
+      "eval_loss": 0.808244526386261,
+      "eval_runtime": 1.3275,
+      "eval_samples_per_second": 753.27,
+      "eval_steps_per_second": 12.052,
+      "step": 294000
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.00012454019309129095,
+      "loss": 0.2826,
+      "step": 294500
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.0001240715535421567,
+      "loss": 0.2828,
+      "step": 295000
+    },
+    {
+      "epoch": 7.52,
+      "eval_loss": 0.809823751449585,
+      "eval_runtime": 1.297,
+      "eval_samples_per_second": 770.984,
+      "eval_steps_per_second": 12.336,
+      "step": 295000
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.00012360325222155773,
+      "loss": 0.2825,
+      "step": 295500
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.00012313529425076228,
+      "loss": 0.2828,
+      "step": 296000
+    },
+    {
+      "epoch": 7.54,
+      "eval_loss": 0.8033545017242432,
+      "eval_runtime": 1.3012,
+      "eval_samples_per_second": 768.545,
+      "eval_steps_per_second": 12.297,
+      "step": 296000
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.00012266768474728372,
+      "loss": 0.283,
+      "step": 296500
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.00012220042882482468,
+      "loss": 0.2831,
+      "step": 297000
+    },
+    {
+      "epoch": 7.57,
+      "eval_loss": 0.8072009086608887,
+      "eval_runtime": 1.2643,
+      "eval_samples_per_second": 790.967,
+      "eval_steps_per_second": 12.655,
+      "step": 297000
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.00012173353159322102,
+      "loss": 0.2826,
+      "step": 297500
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.00012126699815838609,
+      "loss": 0.2825,
+      "step": 298000
+    },
+    {
+      "epoch": 7.59,
+      "eval_loss": 0.8063223958015442,
+      "eval_runtime": 1.3228,
+      "eval_samples_per_second": 755.96,
+      "eval_steps_per_second": 12.095,
+      "step": 298000
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.00012080083362225484,
+      "loss": 0.2819,
+      "step": 298500
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.00012033504308272786,
+      "loss": 0.282,
+      "step": 299000
+    },
+    {
+      "epoch": 7.62,
+      "eval_loss": 0.812494158744812,
+      "eval_runtime": 1.3695,
+      "eval_samples_per_second": 730.168,
+      "eval_steps_per_second": 11.683,
+      "step": 299000
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.00011986963163361598,
+      "loss": 0.2826,
+      "step": 299500
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.0001194046043645841,
+      "loss": 0.2823,
+      "step": 300000
+    },
+    {
+      "epoch": 7.65,
+      "eval_loss": 0.815354585647583,
+      "eval_runtime": 1.3233,
+      "eval_samples_per_second": 755.698,
+      "eval_steps_per_second": 12.091,
+      "step": 300000
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.00011893996636109606,
+      "loss": 0.2819,
+      "step": 300500
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.00011847572270435852,
+      "loss": 0.2818,
+      "step": 301000
+    },
+    {
+      "epoch": 7.67,
+      "eval_loss": 0.8139034509658813,
+      "eval_runtime": 1.299,
+      "eval_samples_per_second": 769.829,
+      "eval_steps_per_second": 12.317,
+      "step": 301000
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.00011801187847126579,
+      "loss": 0.2821,
+      "step": 301500
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.00011754843873434411,
+      "loss": 0.2818,
+      "step": 302000
+    },
+    {
+      "epoch": 7.7,
+      "eval_loss": 0.8098216652870178,
+      "eval_runtime": 1.2656,
+      "eval_samples_per_second": 790.131,
+      "eval_steps_per_second": 12.642,
+      "step": 302000
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.00011708540856169612,
+      "loss": 0.2816,
+      "step": 302500
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.00011662279301694567,
+      "loss": 0.2826,
+      "step": 303000
+    },
+    {
+      "epoch": 7.72,
+      "eval_loss": 0.8181115984916687,
+      "eval_runtime": 1.2294,
+      "eval_samples_per_second": 813.399,
+      "eval_steps_per_second": 13.014,
+      "step": 303000
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.0001161605971591822,
+      "loss": 0.2817,
+      "step": 303500
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.00011569882604290559,
+      "loss": 0.2825,
+      "step": 304000
+    },
+    {
+      "epoch": 7.75,
+      "eval_loss": 0.8146129250526428,
+      "eval_runtime": 1.2431,
+      "eval_samples_per_second": 804.472,
+      "eval_steps_per_second": 12.872,
+      "step": 304000
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.00011523748471797075,
+      "loss": 0.2816,
+      "step": 304500
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 0.00011477657822953255,
+      "loss": 0.2813,
+      "step": 305000
+    },
+    {
+      "epoch": 7.77,
+      "eval_loss": 0.8215622305870056,
+      "eval_runtime": 1.2616,
+      "eval_samples_per_second": 792.667,
+      "eval_steps_per_second": 12.683,
+      "step": 305000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.00011431611161799043,
+      "loss": 0.2818,
+      "step": 305500
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.0001138560899189335,
+      "loss": 0.2814,
+      "step": 306000
+    },
+    {
+      "epoch": 7.8,
+      "eval_loss": 0.8133662343025208,
+      "eval_runtime": 1.3232,
+      "eval_samples_per_second": 755.725,
+      "eval_steps_per_second": 12.092,
+      "step": 306000
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.00011339651816308543,
+      "loss": 0.2818,
+      "step": 306500
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.00011293740137624925,
+      "loss": 0.2808,
+      "step": 307000
+    },
+    {
+      "epoch": 7.82,
+      "eval_loss": 0.811087429523468,
+      "eval_runtime": 1.3842,
+      "eval_samples_per_second": 722.427,
+      "eval_steps_per_second": 11.559,
+      "step": 307000
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.00011247874457925261,
+      "loss": 0.281,
+      "step": 307500
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.0001120205527878927,
+      "loss": 0.2808,
+      "step": 308000
+    },
+    {
+      "epoch": 7.85,
+      "eval_loss": 0.8111096620559692,
+      "eval_runtime": 1.3207,
+      "eval_samples_per_second": 757.185,
+      "eval_steps_per_second": 12.115,
+      "step": 308000
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.00011156283101288165,
+      "loss": 0.2813,
+      "step": 308500
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.00011110558425979132,
+      "loss": 0.2811,
+      "step": 309000
+    },
+    {
+      "epoch": 7.88,
+      "eval_loss": 0.8077411651611328,
+      "eval_runtime": 1.3568,
+      "eval_samples_per_second": 737.044,
+      "eval_steps_per_second": 11.793,
+      "step": 309000
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.00011064881752899906,
+      "loss": 0.281,
+      "step": 309500
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.00011019253581563262,
+      "loss": 0.2812,
+      "step": 310000
+    },
+    {
+      "epoch": 7.9,
+      "eval_loss": 0.8111229538917542,
+      "eval_runtime": 1.2708,
+      "eval_samples_per_second": 786.914,
+      "eval_steps_per_second": 12.591,
+      "step": 310000
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.00010973674410951567,
+      "loss": 0.281,
+      "step": 310500
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.00010928144739511337,
+      "loss": 0.281,
+      "step": 311000
+    },
+    {
+      "epoch": 7.93,
+      "eval_loss": 0.806954562664032,
+      "eval_runtime": 1.274,
+      "eval_samples_per_second": 784.956,
+      "eval_steps_per_second": 12.559,
+      "step": 311000
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.00010882665065147757,
+      "loss": 0.2805,
+      "step": 311500
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 0.00010837235885219267,
+      "loss": 0.2807,
+      "step": 312000
+    },
+    {
+      "epoch": 7.95,
+      "eval_loss": 0.8041293621063232,
+      "eval_runtime": 1.273,
+      "eval_samples_per_second": 785.527,
+      "eval_steps_per_second": 12.568,
+      "step": 312000
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.00010791857696532089,
+      "loss": 0.2808,
+      "step": 312500
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.00010746530995334832,
+      "loss": 0.2811,
+      "step": 313000
+    },
+    {
+      "epoch": 7.98,
+      "eval_loss": 0.8100136518478394,
+      "eval_runtime": 1.2743,
+      "eval_samples_per_second": 784.749,
+      "eval_steps_per_second": 12.556,
+      "step": 313000
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.0001070125627731304,
+      "loss": 0.2809,
+      "step": 313500
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.0001065603403758377,
+      "loss": 0.2821,
+      "step": 314000
+    },
+    {
+      "epoch": 8.0,
+      "eval_loss": 0.828373908996582,
+      "eval_runtime": 1.2715,
+      "eval_samples_per_second": 786.498,
+      "eval_steps_per_second": 12.584,
+      "step": 314000
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 0.00010610864770690196,
+      "loss": 0.2806,
+      "step": 314500
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 0.00010565748970596172,
+      "loss": 0.2808,
+      "step": 315000
+    },
+    {
+      "epoch": 8.03,
+      "eval_loss": 0.8072633743286133,
+      "eval_runtime": 1.2547,
+      "eval_samples_per_second": 796.999,
+      "eval_steps_per_second": 12.752,
+      "step": 315000
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 0.00010520687130680884,
+      "loss": 0.2809,
+      "step": 315500
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 0.00010475679743733364,
+      "loss": 0.2805,
+      "step": 316000
+    },
+    {
+      "epoch": 8.05,
+      "eval_loss": 0.8141467571258545,
+      "eval_runtime": 1.2856,
+      "eval_samples_per_second": 777.823,
+      "eval_steps_per_second": 12.445,
+      "step": 316000
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 0.00010430727301947202,
+      "loss": 0.2804,
+      "step": 316500
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 0.00010385830296915104,
+      "loss": 0.2801,
+      "step": 317000
+    },
+    {
+      "epoch": 8.08,
+      "eval_loss": 0.8066883683204651,
+      "eval_runtime": 1.2502,
+      "eval_samples_per_second": 799.87,
+      "eval_steps_per_second": 12.798,
+      "step": 317000
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 0.00010340989219623508,
+      "loss": 0.2803,
+      "step": 317500
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 0.0001029620456044727,
+      "loss": 0.28,
+      "step": 318000
+    },
+    {
+      "epoch": 8.1,
+      "eval_loss": 0.8122532963752747,
+      "eval_runtime": 1.2711,
+      "eval_samples_per_second": 786.71,
+      "eval_steps_per_second": 12.587,
+      "step": 318000
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.00010251476809144226,
+      "loss": 0.2801,
+      "step": 318500
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 0.00010206806454849917,
+      "loss": 0.2802,
+      "step": 319000
+    },
+    {
+      "epoch": 8.13,
+      "eval_loss": 0.8077669143676758,
+      "eval_runtime": 1.2395,
+      "eval_samples_per_second": 806.793,
+      "eval_steps_per_second": 12.909,
+      "step": 319000
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 0.00010162193986072167,
+      "loss": 0.2805,
+      "step": 319500
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 0.00010117639890685795,
+      "loss": 0.2799,
+      "step": 320000
+    },
+    {
+      "epoch": 8.16,
+      "eval_loss": 0.8211207389831543,
+      "eval_runtime": 1.2456,
+      "eval_samples_per_second": 802.848,
+      "eval_steps_per_second": 12.846,
+      "step": 320000
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 0.00010073144655927253,
+      "loss": 0.2809,
+      "step": 320500
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 0.0001002870876838929,
+      "loss": 0.28,
+      "step": 321000
+    },
+    {
+      "epoch": 8.18,
+      "eval_loss": 0.8134902119636536,
+      "eval_runtime": 1.3191,
+      "eval_samples_per_second": 758.118,
+      "eval_steps_per_second": 12.13,
+      "step": 321000
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.984332714015662e-05,
+      "loss": 0.2799,
+      "step": 321500
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 9.94001697809578e-05,
+      "loss": 0.2796,
+      "step": 322000
+    },
+    {
+      "epoch": 8.21,
+      "eval_loss": 0.8164393305778503,
+      "eval_runtime": 1.2535,
+      "eval_samples_per_second": 797.791,
+      "eval_steps_per_second": 12.765,
+      "step": 322000
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 9.895762045259445e-05,
+      "loss": 0.2797,
+      "step": 322500
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 9.851568399471498e-05,
+      "loss": 0.2793,
+      "step": 323000
+    },
+    {
+      "epoch": 8.23,
+      "eval_loss": 0.8119146823883057,
+      "eval_runtime": 1.2514,
+      "eval_samples_per_second": 799.097,
+      "eval_steps_per_second": 12.786,
+      "step": 323000
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 9.807436524026574e-05,
+      "loss": 0.2797,
+      "step": 323500
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 9.763366901543801e-05,
+      "loss": 0.2791,
+      "step": 324000
+    },
+    {
+      "epoch": 8.26,
+      "eval_loss": 0.8064904808998108,
+      "eval_runtime": 1.2268,
+      "eval_samples_per_second": 815.12,
+      "eval_steps_per_second": 13.042,
+      "step": 324000
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 9.719360013961495e-05,
+      "loss": 0.2793,
+      "step": 324500
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 9.675416342531944e-05,
+      "loss": 0.2793,
+      "step": 325000
+    },
+    {
+      "epoch": 8.28,
+      "eval_loss": 0.8141771554946899,
+      "eval_runtime": 1.2692,
+      "eval_samples_per_second": 787.921,
+      "eval_steps_per_second": 12.607,
+      "step": 325000
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 9.631536367816086e-05,
+      "loss": 0.2798,
+      "step": 325500
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 9.587720569678299e-05,
+      "loss": 0.2794,
+      "step": 326000
+    },
+    {
+      "epoch": 8.31,
+      "eval_loss": 0.803835391998291,
+      "eval_runtime": 1.24,
+      "eval_samples_per_second": 806.471,
+      "eval_steps_per_second": 12.904,
+      "step": 326000
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 9.543969427281131e-05,
+      "loss": 0.2791,
+      "step": 326500
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 9.500283419080062e-05,
+      "loss": 0.2792,
+      "step": 327000
+    },
+    {
+      "epoch": 8.33,
+      "eval_loss": 0.81174635887146,
+      "eval_runtime": 1.301,
+      "eval_samples_per_second": 768.659,
+      "eval_steps_per_second": 12.299,
+      "step": 327000
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 9.45666302281829e-05,
+      "loss": 0.2787,
+      "step": 327500
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 9.413108715521467e-05,
+      "loss": 0.2789,
+      "step": 328000
+    },
+    {
+      "epoch": 8.36,
+      "eval_loss": 0.8118357062339783,
+      "eval_runtime": 1.2706,
+      "eval_samples_per_second": 787.037,
+      "eval_steps_per_second": 12.593,
+      "step": 328000
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 9.369620973492525e-05,
+      "loss": 0.2794,
+      "step": 328500
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 9.326200272306445e-05,
+      "loss": 0.2793,
+      "step": 329000
+    },
+    {
+      "epoch": 8.38,
+      "eval_loss": 0.809190034866333,
+      "eval_runtime": 1.2623,
+      "eval_samples_per_second": 792.184,
+      "eval_steps_per_second": 12.675,
+      "step": 329000
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 9.282847086805059e-05,
+      "loss": 0.2788,
+      "step": 329500
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 9.239561891091853e-05,
+      "loss": 0.279,
+      "step": 330000
+    },
+    {
+      "epoch": 8.41,
+      "eval_loss": 0.8081182837486267,
+      "eval_runtime": 1.2445,
+      "eval_samples_per_second": 803.539,
+      "eval_steps_per_second": 12.857,
+      "step": 330000
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 9.196345158526793e-05,
+      "loss": 0.2786,
+      "step": 330500
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 9.153197361721149e-05,
+      "loss": 0.2792,
+      "step": 331000
+    },
+    {
+      "epoch": 8.44,
+      "eval_loss": 0.8179449439048767,
+      "eval_runtime": 1.254,
+      "eval_samples_per_second": 797.474,
+      "eval_steps_per_second": 12.76,
+      "step": 331000
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 9.110118972532302e-05,
+      "loss": 0.2788,
+      "step": 331500
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 9.067110462058634e-05,
+      "loss": 0.2788,
+      "step": 332000
+    },
+    {
+      "epoch": 8.46,
+      "eval_loss": 0.8141337633132935,
+      "eval_runtime": 1.3186,
+      "eval_samples_per_second": 758.376,
+      "eval_steps_per_second": 12.134,
+      "step": 332000
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 9.024172300634305e-05,
+      "loss": 0.2789,
+      "step": 332500
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 8.981304957824182e-05,
+      "loss": 0.2785,
+      "step": 333000
+    },
+    {
+      "epoch": 8.49,
+      "eval_loss": 0.8111985325813293,
+      "eval_runtime": 1.2328,
+      "eval_samples_per_second": 811.183,
+      "eval_steps_per_second": 12.979,
+      "step": 333000
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 8.938508902418643e-05,
+      "loss": 0.2786,
+      "step": 333500
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 8.89578460242851e-05,
+      "loss": 0.2786,
+      "step": 334000
+    },
+    {
+      "epoch": 8.51,
+      "eval_loss": 0.8080055117607117,
+      "eval_runtime": 1.2558,
+      "eval_samples_per_second": 796.279,
+      "eval_steps_per_second": 12.74,
+      "step": 334000
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 8.85313252507988e-05,
+      "loss": 0.2785,
+      "step": 334500
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 8.810553136809027e-05,
+      "loss": 0.2788,
+      "step": 335000
+    },
+    {
+      "epoch": 8.54,
+      "eval_loss": 0.8105797171592712,
+      "eval_runtime": 1.2508,
+      "eval_samples_per_second": 799.461,
+      "eval_steps_per_second": 12.791,
+      "step": 335000
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 8.76804690325733e-05,
+      "loss": 0.2792,
+      "step": 335500
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 8.725614289266137e-05,
+      "loss": 0.279,
+      "step": 336000
+    },
+    {
+      "epoch": 8.56,
+      "eval_loss": 0.8105652928352356,
+      "eval_runtime": 1.5015,
+      "eval_samples_per_second": 666.004,
+      "eval_steps_per_second": 10.656,
+      "step": 336000
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 8.683255758871734e-05,
+      "loss": 0.2782,
+      "step": 336500
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 8.640971775300207e-05,
+      "loss": 0.2781,
+      "step": 337000
+    },
+    {
+      "epoch": 8.59,
+      "eval_loss": 0.8099997043609619,
+      "eval_runtime": 1.3182,
+      "eval_samples_per_second": 758.589,
+      "eval_steps_per_second": 12.137,
+      "step": 337000
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 8.598762800962431e-05,
+      "loss": 0.2779,
+      "step": 337500
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 8.55662929744899e-05,
+      "loss": 0.278,
+      "step": 338000
+    },
+    {
+      "epoch": 8.61,
+      "eval_loss": 0.8251625895500183,
+      "eval_runtime": 1.3504,
+      "eval_samples_per_second": 740.537,
+      "eval_steps_per_second": 11.849,
+      "step": 338000
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 8.514571725525124e-05,
+      "loss": 0.2778,
+      "step": 338500
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 8.47259054512571e-05,
+      "loss": 0.2777,
+      "step": 339000
+    },
+    {
+      "epoch": 8.64,
+      "eval_loss": 0.8137243390083313,
+      "eval_runtime": 1.2434,
+      "eval_samples_per_second": 804.259,
+      "eval_steps_per_second": 12.868,
+      "step": 339000
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 8.430686215350198e-05,
+      "loss": 0.2777,
+      "step": 339500
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 8.388859194457636e-05,
+      "loss": 0.2778,
+      "step": 340000
+    },
+    {
+      "epoch": 8.67,
+      "eval_loss": 0.8186727166175842,
+      "eval_runtime": 1.2904,
+      "eval_samples_per_second": 774.959,
+      "eval_steps_per_second": 12.399,
+      "step": 340000
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 8.347109939861605e-05,
+      "loss": 0.2778,
+      "step": 340500
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 8.305438908125285e-05,
+      "loss": 0.2773,
+      "step": 341000
+    },
+    {
+      "epoch": 8.69,
+      "eval_loss": 0.8103182315826416,
+      "eval_runtime": 1.3256,
+      "eval_samples_per_second": 754.398,
+      "eval_steps_per_second": 12.07,
+      "step": 341000
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 8.263846554956402e-05,
+      "loss": 0.2776,
+      "step": 341500
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 8.222333335202254e-05,
+      "loss": 0.2779,
+      "step": 342000
+    },
+    {
+      "epoch": 8.72,
+      "eval_loss": 0.8094441294670105,
+      "eval_runtime": 1.28,
+      "eval_samples_per_second": 781.22,
+      "eval_steps_per_second": 12.5,
+      "step": 342000
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 8.18089970284477e-05,
+      "loss": 0.2783,
+      "step": 342500
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 8.1395461109955e-05,
+      "loss": 0.2777,
+      "step": 343000
+    },
+    {
+      "epoch": 8.74,
+      "eval_loss": 0.802370548248291,
+      "eval_runtime": 1.2621,
+      "eval_samples_per_second": 792.35,
+      "eval_steps_per_second": 12.678,
+      "step": 343000
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 8.098273011890726e-05,
+      "loss": 0.2772,
+      "step": 343500
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 8.057080856886426e-05,
+      "loss": 0.277,
+      "step": 344000
+    },
+    {
+      "epoch": 8.77,
+      "eval_loss": 0.8032689094543457,
+      "eval_runtime": 1.2677,
+      "eval_samples_per_second": 788.842,
+      "eval_steps_per_second": 12.621,
+      "step": 344000
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 8.015970096453414e-05,
+      "loss": 0.2771,
+      "step": 344500
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.974941180172382e-05,
+      "loss": 0.2771,
+      "step": 345000
+    },
+    {
+      "epoch": 8.79,
+      "eval_loss": 0.8085340857505798,
+      "eval_runtime": 1.2759,
+      "eval_samples_per_second": 783.767,
+      "eval_steps_per_second": 12.54,
+      "step": 345000
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 7.933994556728976e-05,
+      "loss": 0.2768,
+      "step": 345500
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 7.893130673908927e-05,
+      "loss": 0.2773,
+      "step": 346000
+    },
+    {
+      "epoch": 8.82,
+      "eval_loss": 0.8130091428756714,
+      "eval_runtime": 1.2476,
+      "eval_samples_per_second": 801.539,
+      "eval_steps_per_second": 12.825,
+      "step": 346000
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 7.852349978593091e-05,
+      "loss": 0.2774,
+      "step": 346500
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 7.811652916752633e-05,
+      "loss": 0.2775,
+      "step": 347000
+    },
+    {
+      "epoch": 8.84,
+      "eval_loss": 0.8052127957344055,
+      "eval_runtime": 1.2887,
+      "eval_samples_per_second": 775.993,
+      "eval_steps_per_second": 12.416,
+      "step": 347000
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 7.771039933444092e-05,
+      "loss": 0.2771,
+      "step": 347500
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 7.730511472804544e-05,
+      "loss": 0.2769,
+      "step": 348000
+    },
+    {
+      "epoch": 8.87,
+      "eval_loss": 0.804824948310852,
+      "eval_runtime": 1.2577,
+      "eval_samples_per_second": 795.106,
+      "eval_steps_per_second": 12.722,
+      "step": 348000
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 7.690067978046758e-05,
+      "loss": 0.2767,
+      "step": 348500
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 7.649709891454298e-05,
+      "loss": 0.2769,
+      "step": 349000
+    },
+    {
+      "epoch": 8.89,
+      "eval_loss": 0.8068578839302063,
+      "eval_runtime": 1.3222,
+      "eval_samples_per_second": 756.317,
+      "eval_steps_per_second": 12.101,
+      "step": 349000
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 7.609437654376742e-05,
+      "loss": 0.277,
+      "step": 349500
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 7.569251707224812e-05,
+      "loss": 0.2774,
+      "step": 350000
+    },
+    {
+      "epoch": 8.92,
+      "eval_loss": 0.8125748038291931,
+      "eval_runtime": 1.2634,
+      "eval_samples_per_second": 791.502,
+      "eval_steps_per_second": 12.664,
+      "step": 350000
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 7.529152489465592e-05,
+      "loss": 0.277,
+      "step": 350500
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 7.489140439617708e-05,
+      "loss": 0.2766,
+      "step": 351000
+    },
+    {
+      "epoch": 8.95,
+      "eval_loss": 0.8036056160926819,
+      "eval_runtime": 1.2298,
+      "eval_samples_per_second": 813.125,
+      "eval_steps_per_second": 13.01,
+      "step": 351000
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 7.449215995246522e-05,
+      "loss": 0.2765,
+      "step": 351500
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 7.409379592959367e-05,
+      "loss": 0.2765,
+      "step": 352000
+    },
+    {
+      "epoch": 8.97,
+      "eval_loss": 0.8099916577339172,
+      "eval_runtime": 1.258,
+      "eval_samples_per_second": 794.9,
+      "eval_steps_per_second": 12.718,
+      "step": 352000
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 7.369631668400746e-05,
+      "loss": 0.2763,
+      "step": 352500
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 7.3299726562476e-05,
+      "loss": 0.2762,
+      "step": 353000
+    },
+    {
+      "epoch": 9.0,
+      "eval_loss": 0.8091428279876709,
+      "eval_runtime": 1.2439,
+      "eval_samples_per_second": 803.907,
+      "eval_steps_per_second": 12.863,
+      "step": 353000
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 7.290402990204531e-05,
+      "loss": 0.2763,
+      "step": 353500
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 7.250923102999073e-05,
+      "loss": 0.2765,
+      "step": 354000
+    },
+    {
+      "epoch": 9.02,
+      "eval_loss": 0.8080966472625732,
+      "eval_runtime": 1.3706,
+      "eval_samples_per_second": 729.606,
+      "eval_steps_per_second": 11.674,
+      "step": 354000
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 7.211533426376934e-05,
+      "loss": 0.2762,
+      "step": 354500
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 7.172234391097317e-05,
+      "loss": 0.2763,
+      "step": 355000
+    },
+    {
+      "epoch": 9.05,
+      "eval_loss": 0.8072100877761841,
+      "eval_runtime": 1.3323,
+      "eval_samples_per_second": 750.572,
+      "eval_steps_per_second": 12.009,
+      "step": 355000
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 7.133026426928173e-05,
+      "loss": 0.2764,
+      "step": 355500
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 7.093909962641514e-05,
+      "loss": 0.2763,
+      "step": 356000
+    },
+    {
+      "epoch": 9.07,
+      "eval_loss": 0.8050107359886169,
+      "eval_runtime": 1.3391,
+      "eval_samples_per_second": 746.746,
+      "eval_steps_per_second": 11.948,
+      "step": 356000
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 7.054885426008737e-05,
+      "loss": 0.276,
+      "step": 356500
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 7.015953243795907e-05,
+      "loss": 0.2763,
+      "step": 357000
+    },
+    {
+      "epoch": 9.1,
+      "eval_loss": 0.8131558299064636,
+      "eval_runtime": 1.3479,
+      "eval_samples_per_second": 741.911,
+      "eval_steps_per_second": 11.871,
+      "step": 357000
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 6.97711384175914e-05,
+      "loss": 0.2762,
+      "step": 357500
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 6.938367644639911e-05,
+      "loss": 0.2758,
+      "step": 358000
+    },
+    {
+      "epoch": 9.12,
+      "eval_loss": 0.8091667294502258,
+      "eval_runtime": 1.3351,
+      "eval_samples_per_second": 749.026,
+      "eval_steps_per_second": 11.984,
+      "step": 358000
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 6.899715076160425e-05,
+      "loss": 0.2757,
+      "step": 358500
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 6.861156559018986e-05,
+      "loss": 0.2758,
+      "step": 359000
+    },
+    {
+      "epoch": 9.15,
+      "eval_loss": 0.8032931685447693,
+      "eval_runtime": 1.2963,
+      "eval_samples_per_second": 771.398,
+      "eval_steps_per_second": 12.342,
+      "step": 359000
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 6.822692514885346e-05,
+      "loss": 0.2757,
+      "step": 359500
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 6.784323364396135e-05,
+      "loss": 0.2757,
+      "step": 360000
+    },
+    {
+      "epoch": 9.17,
+      "eval_loss": 0.8121919631958008,
+      "eval_runtime": 1.3627,
+      "eval_samples_per_second": 733.817,
+      "eval_steps_per_second": 11.741,
+      "step": 360000
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 6.746049527150238e-05,
+      "loss": 0.276,
+      "step": 360500
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 6.707871421704209e-05,
+      "loss": 0.2757,
+      "step": 361000
+    },
+    {
+      "epoch": 9.2,
+      "eval_loss": 0.8061186075210571,
+      "eval_runtime": 1.2926,
+      "eval_samples_per_second": 773.629,
+      "eval_steps_per_second": 12.378,
+      "step": 361000
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 6.669789465567683e-05,
+      "loss": 0.2754,
+      "step": 361500
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 6.631804075198838e-05,
+      "loss": 0.2754,
+      "step": 362000
+    },
+    {
+      "epoch": 9.23,
+      "eval_loss": 0.8106433153152466,
+      "eval_runtime": 1.3154,
+      "eval_samples_per_second": 760.201,
+      "eval_steps_per_second": 12.163,
+      "step": 362000
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 6.593915665999816e-05,
+      "loss": 0.275,
+      "step": 362500
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 6.55612465231219e-05,
+      "loss": 0.2755,
+      "step": 363000
+    },
+    {
+      "epoch": 9.25,
+      "eval_loss": 0.8047894835472107,
+      "eval_runtime": 1.3007,
+      "eval_samples_per_second": 768.793,
+      "eval_steps_per_second": 12.301,
+      "step": 363000
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 6.518431447412434e-05,
+      "loss": 0.2754,
+      "step": 363500
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 6.480836463507392e-05,
+      "loss": 0.2753,
+      "step": 364000
+    },
+    {
+      "epoch": 9.28,
+      "eval_loss": 0.8103929758071899,
+      "eval_runtime": 1.3172,
+      "eval_samples_per_second": 759.199,
+      "eval_steps_per_second": 12.147,
+      "step": 364000
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 6.443340111729786e-05,
+      "loss": 0.2753,
+      "step": 364500
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 6.405942802133713e-05,
+      "loss": 0.2753,
+      "step": 365000
+    },
+    {
+      "epoch": 9.3,
+      "eval_loss": 0.8095191717147827,
+      "eval_runtime": 1.3369,
+      "eval_samples_per_second": 747.972,
+      "eval_steps_per_second": 11.968,
+      "step": 365000
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 6.36864494369016e-05,
+      "loss": 0.2754,
+      "step": 365500
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 6.331446944282534e-05,
+      "loss": 0.2753,
+      "step": 366000
+    },
+    {
+      "epoch": 9.33,
+      "eval_loss": 0.8096633553504944,
+      "eval_runtime": 1.2639,
+      "eval_samples_per_second": 791.187,
+      "eval_steps_per_second": 12.659,
+      "step": 366000
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 6.294349210702188e-05,
+      "loss": 0.2748,
+      "step": 366500
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 6.257352148643998e-05,
+      "loss": 0.2752,
+      "step": 367000
+    },
+    {
+      "epoch": 9.35,
+      "eval_loss": 0.808983564376831,
+      "eval_runtime": 1.349,
+      "eval_samples_per_second": 741.265,
+      "eval_steps_per_second": 11.86,
+      "step": 367000
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 6.220456162701908e-05,
+      "loss": 0.2751,
+      "step": 367500
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 6.183661656364515e-05,
+      "loss": 0.2749,
+      "step": 368000
+    },
+    {
+      "epoch": 9.38,
+      "eval_loss": 0.8059231638908386,
+      "eval_runtime": 1.3338,
+      "eval_samples_per_second": 749.736,
+      "eval_steps_per_second": 11.996,
+      "step": 368000
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 6.146969032010631e-05,
+      "loss": 0.2747,
+      "step": 368500
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 6.110378690904928e-05,
+      "loss": 0.2749,
+      "step": 369000
+    },
+    {
+      "epoch": 9.4,
+      "eval_loss": 0.8113678097724915,
+      "eval_runtime": 1.2927,
+      "eval_samples_per_second": 773.561,
+      "eval_steps_per_second": 12.377,
+      "step": 369000
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 6.073891033193507e-05,
+      "loss": 0.2749,
+      "step": 369500
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 6.037506457899553e-05,
+      "loss": 0.2747,
+      "step": 370000
+    },
+    {
+      "epoch": 9.43,
+      "eval_loss": 0.8089262843132019,
+      "eval_runtime": 1.3047,
+      "eval_samples_per_second": 766.44,
+      "eval_steps_per_second": 12.263,
+      "step": 370000
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 6.0012253629189544e-05,
+      "loss": 0.2753,
+      "step": 370500
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.965048145015944e-05,
+      "loss": 0.2745,
+      "step": 371000
+    },
+    {
+      "epoch": 9.46,
+      "eval_loss": 0.807996928691864,
+      "eval_runtime": 1.2961,
+      "eval_samples_per_second": 771.545,
+      "eval_steps_per_second": 12.345,
+      "step": 371000
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.928975199818785e-05,
+      "loss": 0.2744,
+      "step": 371500
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.893006921815428e-05,
+      "loss": 0.2745,
+      "step": 372000
+    },
+    {
+      "epoch": 9.48,
+      "eval_loss": 0.8102282881736755,
+      "eval_runtime": 1.3228,
+      "eval_samples_per_second": 755.978,
+      "eval_steps_per_second": 12.096,
+      "step": 372000
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.857143704349198e-05,
+      "loss": 0.2749,
+      "step": 372500
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.8213859396144986e-05,
+      "loss": 0.2747,
+      "step": 373000
+    },
+    {
+      "epoch": 9.51,
+      "eval_loss": 0.8059311509132385,
+      "eval_runtime": 1.3365,
+      "eval_samples_per_second": 748.226,
+      "eval_steps_per_second": 11.972,
+      "step": 373000
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.785734018652507e-05,
+      "loss": 0.2743,
+      "step": 373500
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.750188331346927e-05,
+      "loss": 0.2742,
+      "step": 374000
+    },
+    {
+      "epoch": 9.53,
+      "eval_loss": 0.8084880113601685,
+      "eval_runtime": 1.3112,
+      "eval_samples_per_second": 762.675,
+      "eval_steps_per_second": 12.203,
+      "step": 374000
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.714749266419695e-05,
+      "loss": 0.2741,
+      "step": 374500
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.6794172114267566e-05,
+      "loss": 0.2742,
+      "step": 375000
+    },
+    {
+      "epoch": 9.56,
+      "eval_loss": 0.8031001091003418,
+      "eval_runtime": 1.2768,
+      "eval_samples_per_second": 783.2,
+      "eval_steps_per_second": 12.531,
+      "step": 375000
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.6441925527537914e-05,
+      "loss": 0.2738,
+      "step": 375500
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.60907567561203e-05,
+      "loss": 0.274,
+      "step": 376000
+    },
+    {
+      "epoch": 9.58,
+      "eval_loss": 0.8067184090614319,
+      "eval_runtime": 1.3299,
+      "eval_samples_per_second": 751.95,
+      "eval_steps_per_second": 12.031,
+      "step": 376000
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.574066964034012e-05,
+      "loss": 0.274,
+      "step": 376500
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.539166800869402e-05,
+      "loss": 0.274,
+      "step": 377000
+    },
+    {
+      "epoch": 9.61,
+      "eval_loss": 0.8057398200035095,
+      "eval_runtime": 1.2808,
+      "eval_samples_per_second": 780.764,
+      "eval_steps_per_second": 12.492,
+      "step": 377000
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.5043755677807955e-05,
+      "loss": 0.2738,
+      "step": 377500
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.4696936452395344e-05,
+      "loss": 0.274,
+      "step": 378000
+    },
+    {
+      "epoch": 9.63,
+      "eval_loss": 0.8031483292579651,
+      "eval_runtime": 1.2827,
+      "eval_samples_per_second": 779.634,
+      "eval_steps_per_second": 12.474,
+      "step": 378000
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.435121412521576e-05,
+      "loss": 0.2738,
+      "step": 378500
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.400659247703307e-05,
+      "loss": 0.2738,
+      "step": 379000
+    },
+    {
+      "epoch": 9.66,
+      "eval_loss": 0.8066932559013367,
+      "eval_runtime": 1.3559,
+      "eval_samples_per_second": 737.531,
+      "eval_steps_per_second": 11.8,
+      "step": 379000
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.36630752765745e-05,
+      "loss": 0.2739,
+      "step": 379500
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.3320666280489146e-05,
+      "loss": 0.2737,
+      "step": 380000
+    },
+    {
+      "epoch": 9.68,
+      "eval_loss": 0.8089754581451416,
+      "eval_runtime": 1.319,
+      "eval_samples_per_second": 758.149,
+      "eval_steps_per_second": 12.13,
+      "step": 380000
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.2979369233306834e-05,
+      "loss": 0.2732,
+      "step": 380500
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.26391878673975e-05,
+      "loss": 0.2736,
+      "step": 381000
+    },
+    {
+      "epoch": 9.71,
+      "eval_loss": 0.8043585419654846,
+      "eval_runtime": 1.3216,
+      "eval_samples_per_second": 756.661,
+      "eval_steps_per_second": 12.107,
+      "step": 381000
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.230012590292987e-05,
+      "loss": 0.2736,
+      "step": 381500
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.1962187047831517e-05,
+      "loss": 0.2739,
+      "step": 382000
+    },
+    {
+      "epoch": 9.74,
+      "eval_loss": 0.8078375458717346,
+      "eval_runtime": 1.2548,
+      "eval_samples_per_second": 796.917,
+      "eval_steps_per_second": 12.751,
+      "step": 382000
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.162537499774743e-05,
+      "loss": 0.2736,
+      "step": 382500
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.128969343600032e-05,
+      "loss": 0.2729,
+      "step": 383000
+    },
+    {
+      "epoch": 9.76,
+      "eval_loss": 0.8074722290039062,
+      "eval_runtime": 1.3011,
+      "eval_samples_per_second": 768.601,
+      "eval_steps_per_second": 12.298,
+      "step": 383000
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.09551460335499e-05,
+      "loss": 0.2732,
+      "step": 383500
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.062173644895296e-05,
+      "loss": 0.2735,
+      "step": 384000
+    },
+    {
+      "epoch": 9.79,
+      "eval_loss": 0.8107377290725708,
+      "eval_runtime": 1.3439,
+      "eval_samples_per_second": 744.113,
+      "eval_steps_per_second": 11.906,
+      "step": 384000
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.0289468328323434e-05,
+      "loss": 0.2728,
+      "step": 384500
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.995834530529208e-05,
+      "loss": 0.2729,
+      "step": 385000
+    },
+    {
+      "epoch": 9.81,
+      "eval_loss": 0.8120101690292358,
+      "eval_runtime": 1.2896,
+      "eval_samples_per_second": 775.434,
+      "eval_steps_per_second": 12.407,
+      "step": 385000
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.9628371000967394e-05,
+      "loss": 0.2731,
+      "step": 385500
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.929954902389534e-05,
+      "loss": 0.2731,
+      "step": 386000
+    },
+    {
+      "epoch": 9.84,
+      "eval_loss": 0.8058800101280212,
+      "eval_runtime": 1.3835,
+      "eval_samples_per_second": 722.807,
+      "eval_steps_per_second": 11.565,
+      "step": 386000
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.897188297002046e-05,
+      "loss": 0.2732,
+      "step": 386500
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.8645376422646226e-05,
+      "loss": 0.2727,
+      "step": 387000
+    },
+    {
+      "epoch": 9.86,
+      "eval_loss": 0.8082349896430969,
+      "eval_runtime": 1.3707,
+      "eval_samples_per_second": 729.563,
+      "eval_steps_per_second": 11.673,
+      "step": 387000
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.832003295239591e-05,
+      "loss": 0.2728,
+      "step": 387500
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.7995856117173624e-05,
+      "loss": 0.2726,
+      "step": 388000
+    },
+    {
+      "epoch": 9.89,
+      "eval_loss": 0.8089985847473145,
+      "eval_runtime": 1.3037,
+      "eval_samples_per_second": 767.033,
+      "eval_steps_per_second": 12.273,
+      "step": 388000
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.767284946212521e-05,
+      "loss": 0.2727,
+      "step": 388500
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.735101651959977e-05,
+      "loss": 0.2727,
+      "step": 389000
+    },
+    {
+      "epoch": 9.91,
+      "eval_loss": 0.8019598126411438,
+      "eval_runtime": 1.3529,
+      "eval_samples_per_second": 739.159,
+      "eval_steps_per_second": 11.827,
+      "step": 389000
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.7030360809110754e-05,
+      "loss": 0.2726,
+      "step": 389500
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.6710885837297726e-05,
+      "loss": 0.273,
+      "step": 390000
+    },
+    {
+      "epoch": 9.94,
+      "eval_loss": 0.8114839792251587,
+      "eval_runtime": 1.3138,
+      "eval_samples_per_second": 761.14,
+      "eval_steps_per_second": 12.178,
+      "step": 390000
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.639259509788768e-05,
+      "loss": 0.2727,
+      "step": 390500
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.60754920716572e-05,
+      "loss": 0.2727,
+      "step": 391000
+    },
+    {
+      "epoch": 9.96,
+      "eval_loss": 0.8076984882354736,
+      "eval_runtime": 1.3566,
+      "eval_samples_per_second": 737.152,
+      "eval_steps_per_second": 11.794,
+      "step": 391000
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.5759580226394167e-05,
+      "loss": 0.2727,
+      "step": 391500
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.544486301685993e-05,
+      "loss": 0.2726,
+      "step": 392000
+    },
+    {
+      "epoch": 9.99,
+      "eval_loss": 0.8174979090690613,
+      "eval_runtime": 1.2981,
+      "eval_samples_per_second": 770.339,
+      "eval_steps_per_second": 12.325,
+      "step": 392000
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.5131343884751484e-05,
+      "loss": 0.2723,
+      "step": 392500
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.4819026258663774e-05,
+      "loss": 0.2722,
+      "step": 393000
+    },
+    {
+      "epoch": 10.02,
+      "eval_loss": 0.8073344826698303,
+      "eval_runtime": 1.3446,
+      "eval_samples_per_second": 743.69,
+      "eval_steps_per_second": 11.899,
+      "step": 393000
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.450791355405234e-05,
+      "loss": 0.2722,
+      "step": 393500
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.419800917319588e-05,
+      "loss": 0.2725,
+      "step": 394000
+    },
+    {
+      "epoch": 10.04,
+      "eval_loss": 0.8089223504066467,
+      "eval_runtime": 1.4579,
+      "eval_samples_per_second": 685.895,
+      "eval_steps_per_second": 10.974,
+      "step": 394000
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.3889316505159056e-05,
+      "loss": 0.2723,
+      "step": 394500
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.3581838925755465e-05,
+      "loss": 0.2721,
+      "step": 395000
+    },
+    {
+      "epoch": 10.07,
+      "eval_loss": 0.8180590867996216,
+      "eval_runtime": 1.4056,
+      "eval_samples_per_second": 711.416,
+      "eval_steps_per_second": 11.383,
+      "step": 395000
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.327557979751057e-05,
+      "loss": 0.2722,
+      "step": 395500
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.297054246962517e-05,
+      "loss": 0.2722,
+      "step": 396000
+    },
+    {
+      "epoch": 10.09,
+      "eval_loss": 0.8067134022712708,
+      "eval_runtime": 1.3816,
+      "eval_samples_per_second": 723.788,
+      "eval_steps_per_second": 11.581,
+      "step": 396000
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.266673027793864e-05,
+      "loss": 0.2717,
+      "step": 396500
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.236414654489242e-05,
+      "loss": 0.2721,
+      "step": 397000
+    },
+    {
+      "epoch": 10.12,
+      "eval_loss": 0.8154900074005127,
+      "eval_runtime": 1.3572,
+      "eval_samples_per_second": 736.809,
+      "eval_steps_per_second": 11.789,
+      "step": 397000
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.206279457949371e-05,
+      "loss": 0.272,
+      "step": 397500
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.1762677677279335e-05,
+      "loss": 0.2718,
+      "step": 398000
+    },
+    {
+      "epoch": 10.14,
+      "eval_loss": 0.814974844455719,
+      "eval_runtime": 1.3197,
+      "eval_samples_per_second": 757.75,
+      "eval_steps_per_second": 12.124,
+      "step": 398000
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.146379912027964e-05,
+      "loss": 0.2715,
+      "step": 398500
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.1166162176982664e-05,
+      "loss": 0.272,
+      "step": 399000
+    },
+    {
+      "epoch": 10.17,
+      "eval_loss": 0.8131051063537598,
+      "eval_runtime": 1.3374,
+      "eval_samples_per_second": 747.697,
+      "eval_steps_per_second": 11.963,
+      "step": 399000
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.086977010229838e-05,
+      "loss": 0.2717,
+      "step": 399500
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.057462613752294e-05,
+      "loss": 0.2721,
+      "step": 400000
+    },
+    {
+      "epoch": 10.19,
+      "eval_loss": 0.8092121481895447,
+      "eval_runtime": 1.3464,
+      "eval_samples_per_second": 742.696,
+      "eval_steps_per_second": 11.883,
+      "step": 400000
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.0280733510303475e-05,
+      "loss": 0.2717,
+      "step": 400500
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.9988095434602716e-05,
+      "loss": 0.2715,
+      "step": 401000
+    },
+    {
+      "epoch": 10.22,
+      "eval_loss": 0.8082920908927917,
+      "eval_runtime": 1.4032,
+      "eval_samples_per_second": 712.676,
+      "eval_steps_per_second": 11.403,
+      "step": 401000
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.9696715110663726e-05,
+      "loss": 0.2715,
+      "step": 401500
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.9406595724975116e-05,
+      "loss": 0.2717,
+      "step": 402000
+    },
+    {
+      "epoch": 10.25,
+      "eval_loss": 0.8099567890167236,
+      "eval_runtime": 1.3773,
+      "eval_samples_per_second": 726.038,
+      "eval_steps_per_second": 11.617,
+      "step": 402000
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.9117740450235914e-05,
+      "loss": 0.2714,
+      "step": 402500
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.8830152445321163e-05,
+      "loss": 0.2715,
+      "step": 403000
+    },
+    {
+      "epoch": 10.27,
+      "eval_loss": 0.8108043670654297,
+      "eval_runtime": 1.3727,
+      "eval_samples_per_second": 728.507,
+      "eval_steps_per_second": 11.656,
+      "step": 403000
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.854383485524724e-05,
+      "loss": 0.2714,
+      "step": 403500
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.8258790811137425e-05,
+      "loss": 0.2715,
+      "step": 404000
+    },
+    {
+      "epoch": 10.3,
+      "eval_loss": 0.8089959621429443,
+      "eval_runtime": 1.3482,
+      "eval_samples_per_second": 741.756,
+      "eval_steps_per_second": 11.868,
+      "step": 404000
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.7975023430187676e-05,
+      "loss": 0.2715,
+      "step": 404500
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.7692535815632624e-05,
+      "loss": 0.2716,
+      "step": 405000
+    },
+    {
+      "epoch": 10.32,
+      "eval_loss": 0.8159785270690918,
+      "eval_runtime": 1.3646,
+      "eval_samples_per_second": 732.809,
+      "eval_steps_per_second": 11.725,
+      "step": 405000
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.741133105671159e-05,
+      "loss": 0.2713,
+      "step": 405500
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.713141222863474e-05,
+      "loss": 0.2712,
+      "step": 406000
+    },
+    {
+      "epoch": 10.35,
+      "eval_loss": 0.8142402172088623,
+      "eval_runtime": 1.3812,
+      "eval_samples_per_second": 724.033,
+      "eval_steps_per_second": 11.585,
+      "step": 406000
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.6852782392549584e-05,
+      "loss": 0.2713,
+      "step": 406500
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.657544459550729e-05,
+      "loss": 0.2712,
+      "step": 407000
+    },
+    {
+      "epoch": 10.37,
+      "eval_loss": 0.8070803284645081,
+      "eval_runtime": 1.5878,
+      "eval_samples_per_second": 629.794,
+      "eval_steps_per_second": 10.077,
+      "step": 407000
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.6299401870429606e-05,
+      "loss": 0.2712,
+      "step": 407500
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.6024657236075546e-05,
+      "loss": 0.2712,
+      "step": 408000
+    },
+    {
+      "epoch": 10.4,
+      "eval_loss": 0.8115477561950684,
+      "eval_runtime": 1.3152,
+      "eval_samples_per_second": 760.367,
+      "eval_steps_per_second": 12.166,
+      "step": 408000
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.575121369700841e-05,
+      "loss": 0.2714,
+      "step": 408500
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.5479074243562995e-05,
+      "loss": 0.2709,
+      "step": 409000
+    },
+    {
+      "epoch": 10.42,
+      "eval_loss": 0.80928635597229,
+      "eval_runtime": 1.3461,
+      "eval_samples_per_second": 742.862,
+      "eval_steps_per_second": 11.886,
+      "step": 409000
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.5208241851812644e-05,
+      "loss": 0.2711,
+      "step": 409500
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.493871948353709e-05,
+      "loss": 0.271,
+      "step": 410000
+    },
+    {
+      "epoch": 10.45,
+      "eval_loss": 0.8108725547790527,
+      "eval_runtime": 1.3475,
+      "eval_samples_per_second": 742.097,
+      "eval_steps_per_second": 11.874,
+      "step": 410000
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.4670510086189736e-05,
+      "loss": 0.271,
+      "step": 410500
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.440361659286563e-05,
+      "loss": 0.2712,
+      "step": 411000
+    },
+    {
+      "epoch": 10.47,
+      "eval_loss": 0.8162291049957275,
+      "eval_runtime": 1.3642,
+      "eval_samples_per_second": 733.056,
+      "eval_steps_per_second": 11.729,
+      "step": 411000
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.413804192226918e-05,
+      "loss": 0.271,
+      "step": 411500
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.387378897868246e-05,
+      "loss": 0.2709,
+      "step": 412000
+    },
+    {
+      "epoch": 10.5,
+      "eval_loss": 0.8157584071159363,
+      "eval_runtime": 1.3922,
+      "eval_samples_per_second": 718.295,
+      "eval_steps_per_second": 11.493,
+      "step": 412000
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.361086065193336e-05,
+      "loss": 0.2705,
+      "step": 412500
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.334925981736389e-05,
+      "loss": 0.2706,
+      "step": 413000
+    },
+    {
+      "epoch": 10.53,
+      "eval_loss": 0.8102970123291016,
+      "eval_runtime": 1.343,
+      "eval_samples_per_second": 744.614,
+      "eval_steps_per_second": 11.914,
+      "step": 413000
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.3088989335798925e-05,
+      "loss": 0.2707,
+      "step": 413500
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.283005205351467e-05,
+      "loss": 0.2709,
+      "step": 414000
+    },
+    {
+      "epoch": 10.55,
+      "eval_loss": 0.8069249987602234,
+      "eval_runtime": 1.3236,
+      "eval_samples_per_second": 755.504,
+      "eval_steps_per_second": 12.088,
+      "step": 414000
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.2572450802207845e-05,
+      "loss": 0.2705,
+      "step": 414500
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.2316188398964344e-05,
+      "loss": 0.2706,
+      "step": 415000
+    },
+    {
+      "epoch": 10.58,
+      "eval_loss": 0.812959611415863,
+      "eval_runtime": 1.3895,
+      "eval_samples_per_second": 719.683,
+      "eval_steps_per_second": 11.515,
+      "step": 415000
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.206126764622888e-05,
+      "loss": 0.2708,
+      "step": 415500
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.180769133177392e-05,
+      "loss": 0.2706,
+      "step": 416000
+    },
+    {
+      "epoch": 10.6,
+      "eval_loss": 0.8125792145729065,
+      "eval_runtime": 1.3667,
+      "eval_samples_per_second": 731.714,
+      "eval_steps_per_second": 11.707,
+      "step": 416000
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.155546222866939e-05,
+      "loss": 0.2704,
+      "step": 416500
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.130458309525239e-05,
+      "loss": 0.2704,
+      "step": 417000
+    },
+    {
+      "epoch": 10.63,
+      "eval_loss": 0.8181024193763733,
+      "eval_runtime": 1.3224,
+      "eval_samples_per_second": 756.173,
+      "eval_steps_per_second": 12.099,
+      "step": 417000
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.1055056675096826e-05,
+      "loss": 0.2702,
+      "step": 417500
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.0806885696983816e-05,
+      "loss": 0.2704,
+      "step": 418000
+    },
+    {
+      "epoch": 10.65,
+      "eval_loss": 0.8099979162216187,
+      "eval_runtime": 1.388,
+      "eval_samples_per_second": 720.469,
+      "eval_steps_per_second": 11.528,
+      "step": 418000
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.056007287487128e-05,
+      "loss": 0.2705,
+      "step": 418500
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.0314620907864744e-05,
+      "loss": 0.2702,
+      "step": 419000
+    },
+    {
+      "epoch": 10.68,
+      "eval_loss": 0.8089344501495361,
+      "eval_runtime": 1.3483,
+      "eval_samples_per_second": 741.661,
+      "eval_steps_per_second": 11.867,
+      "step": 419000
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.0070532480187637e-05,
+      "loss": 0.2701,
+      "step": 419500
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 2.9827810261151784e-05,
+      "loss": 0.2702,
+      "step": 420000
+    },
+    {
+      "epoch": 10.7,
+      "eval_loss": 0.8132815957069397,
+      "eval_runtime": 1.3269,
+      "eval_samples_per_second": 753.638,
+      "eval_steps_per_second": 12.058,
+      "step": 420000
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 2.9586456905128618e-05,
+      "loss": 0.27,
+      "step": 420500
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.9346475051519687e-05,
+      "loss": 0.2699,
+      "step": 421000
+    },
+    {
+      "epoch": 10.73,
+      "eval_loss": 0.8154944777488708,
+      "eval_runtime": 1.3065,
+      "eval_samples_per_second": 765.411,
+      "eval_steps_per_second": 12.247,
+      "step": 421000
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.910786732472815e-05,
+      "loss": 0.27,
+      "step": 421500
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 2.887063633412981e-05,
+      "loss": 0.2701,
+      "step": 422000
+    },
+    {
+      "epoch": 10.75,
+      "eval_loss": 0.8139403462409973,
+      "eval_runtime": 1.2861,
+      "eval_samples_per_second": 777.555,
+      "eval_steps_per_second": 12.441,
+      "step": 422000
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 2.863478467404478e-05,
+      "loss": 0.2699,
+      "step": 422500
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.8400314923709112e-05,
+      "loss": 0.2701,
+      "step": 423000
+    },
+    {
+      "epoch": 10.78,
+      "eval_loss": 0.8132787942886353,
+      "eval_runtime": 1.3439,
+      "eval_samples_per_second": 744.104,
+      "eval_steps_per_second": 11.906,
+      "step": 423000
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 2.816722964724636e-05,
+      "loss": 0.2697,
+      "step": 423500
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 2.793553139363981e-05,
+      "loss": 0.2701,
+      "step": 424000
+    },
+    {
+      "epoch": 10.81,
+      "eval_loss": 0.8099727034568787,
+      "eval_runtime": 1.3264,
+      "eval_samples_per_second": 753.909,
+      "eval_steps_per_second": 12.063,
+      "step": 424000
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.7705222696704366e-05,
+      "loss": 0.27,
+      "step": 424500
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.7476306075059096e-05,
+      "loss": 0.2696,
+      "step": 425000
+    },
+    {
+      "epoch": 10.83,
+      "eval_loss": 0.8077329993247986,
+      "eval_runtime": 1.3315,
+      "eval_samples_per_second": 751.038,
+      "eval_steps_per_second": 12.017,
+      "step": 425000
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 2.7248784032099478e-05,
+      "loss": 0.2696,
+      "step": 425500
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 2.7022659055970144e-05,
+      "loss": 0.2696,
+      "step": 426000
+    },
+    {
+      "epoch": 10.86,
+      "eval_loss": 0.8097303509712219,
+      "eval_runtime": 1.3191,
+      "eval_samples_per_second": 758.11,
+      "eval_steps_per_second": 12.13,
+      "step": 426000
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.6797933619537604e-05,
+      "loss": 0.2694,
+      "step": 426500
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 2.6574610180363166e-05,
+      "loss": 0.2698,
+      "step": 427000
+    },
+    {
+      "epoch": 10.88,
+      "eval_loss": 0.8035640120506287,
+      "eval_runtime": 1.3921,
+      "eval_samples_per_second": 718.328,
+      "eval_steps_per_second": 11.493,
+      "step": 427000
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 2.6352691180676286e-05,
+      "loss": 0.2696,
+      "step": 427500
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.6132179047347505e-05,
+      "loss": 0.2698,
+      "step": 428000
+    },
+    {
+      "epoch": 10.91,
+      "eval_loss": 0.8066699504852295,
+      "eval_runtime": 1.5174,
+      "eval_samples_per_second": 659.02,
+      "eval_steps_per_second": 10.544,
+      "step": 428000
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.5913076191862238e-05,
+      "loss": 0.2696,
+      "step": 428500
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 2.5695385010294165e-05,
+      "loss": 0.2699,
+      "step": 429000
+    },
+    {
+      "epoch": 10.93,
+      "eval_loss": 0.813107430934906,
+      "eval_runtime": 1.3419,
+      "eval_samples_per_second": 745.228,
+      "eval_steps_per_second": 11.924,
+      "step": 429000
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 2.5479107883279144e-05,
+      "loss": 0.2692,
+      "step": 429500
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.5264247175989292e-05,
+      "loss": 0.2695,
+      "step": 430000
+    },
+    {
+      "epoch": 10.96,
+      "eval_loss": 0.8058525323867798,
+      "eval_runtime": 1.3197,
+      "eval_samples_per_second": 757.744,
+      "eval_steps_per_second": 12.124,
+      "step": 430000
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 2.5050805238106804e-05,
+      "loss": 0.2692,
+      "step": 430500
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 2.4838784403798542e-05,
+      "loss": 0.2695,
+      "step": 431000
+    },
+    {
+      "epoch": 10.98,
+      "eval_loss": 0.8141555190086365,
+      "eval_runtime": 1.4076,
+      "eval_samples_per_second": 710.446,
+      "eval_steps_per_second": 11.367,
+      "step": 431000
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.4628186991690346e-05,
+      "loss": 0.2696,
+      "step": 431500
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 2.4419015304841797e-05,
+      "loss": 0.2693,
+      "step": 432000
+    },
+    {
+      "epoch": 11.01,
+      "eval_loss": 0.8079712986946106,
+      "eval_runtime": 1.249,
+      "eval_samples_per_second": 800.662,
+      "eval_steps_per_second": 12.811,
+      "step": 432000
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 2.4211271630720957e-05,
+      "loss": 0.2694,
+      "step": 432500
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 2.4004958241179347e-05,
+      "loss": 0.2695,
+      "step": 433000
+    },
+    {
+      "epoch": 11.04,
+      "eval_loss": 0.8100965619087219,
+      "eval_runtime": 1.3521,
+      "eval_samples_per_second": 739.569,
+      "eval_steps_per_second": 11.833,
+      "step": 433000
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.3800077392427193e-05,
+      "loss": 0.2693,
+      "step": 433500
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 2.3596631325008536e-05,
+      "loss": 0.2692,
+      "step": 434000
+    },
+    {
+      "epoch": 11.06,
+      "eval_loss": 0.8111158609390259,
+      "eval_runtime": 1.2947,
+      "eval_samples_per_second": 772.383,
+      "eval_steps_per_second": 12.358,
+      "step": 434000
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.3394622263777042e-05,
+      "loss": 0.2691,
+      "step": 434500
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.3194052417871433e-05,
+      "loss": 0.2693,
+      "step": 435000
+    },
+    {
+      "epoch": 11.09,
+      "eval_loss": 0.8063649535179138,
+      "eval_runtime": 1.2815,
+      "eval_samples_per_second": 780.354,
+      "eval_steps_per_second": 12.486,
+      "step": 435000
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 2.2994923980691425e-05,
+      "loss": 0.269,
+      "step": 435500
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.279723912987365e-05,
+      "loss": 0.2689,
+      "step": 436000
+    },
+    {
+      "epoch": 11.11,
+      "eval_loss": 0.8065616488456726,
+      "eval_runtime": 1.347,
+      "eval_samples_per_second": 742.376,
+      "eval_steps_per_second": 11.878,
+      "step": 436000
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 2.2601000027268006e-05,
+      "loss": 0.269,
+      "step": 436500
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.2406208818913857e-05,
+      "loss": 0.2688,
+      "step": 437000
+    },
+    {
+      "epoch": 11.14,
+      "eval_loss": 0.8145356178283691,
+      "eval_runtime": 1.3131,
+      "eval_samples_per_second": 761.584,
+      "eval_steps_per_second": 12.185,
+      "step": 437000
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 2.221286763501666e-05,
+      "loss": 0.2687,
+      "step": 437500
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.2020978589924673e-05,
+      "loss": 0.2691,
+      "step": 438000
+    },
+    {
+      "epoch": 11.16,
+      "eval_loss": 0.8088080883026123,
+      "eval_runtime": 1.3003,
+      "eval_samples_per_second": 769.026,
+      "eval_steps_per_second": 12.304,
+      "step": 438000
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.1830543782105647e-05,
+      "loss": 0.2689,
+      "step": 438500
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 2.1641565294124206e-05,
+      "loss": 0.2689,
+      "step": 439000
+    },
+    {
+      "epoch": 11.19,
+      "eval_loss": 0.8115460872650146,
+      "eval_runtime": 1.3182,
+      "eval_samples_per_second": 758.583,
+      "eval_steps_per_second": 12.137,
+      "step": 439000
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.1454045192618794e-05,
+      "loss": 0.2685,
+      "step": 439500
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 2.1267985528279212e-05,
+      "loss": 0.2688,
+      "step": 440000
+    },
+    {
+      "epoch": 11.21,
+      "eval_loss": 0.8065956830978394,
+      "eval_runtime": 1.2152,
+      "eval_samples_per_second": 822.887,
+      "eval_steps_per_second": 13.166,
+      "step": 440000
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.1083388335824145e-05,
+      "loss": 0.2689,
+      "step": 440500
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 2.0900255633978873e-05,
+      "loss": 0.2689,
+      "step": 441000
+    },
+    {
+      "epoch": 11.24,
+      "eval_loss": 0.8037525415420532,
+      "eval_runtime": 1.2548,
+      "eval_samples_per_second": 796.966,
+      "eval_steps_per_second": 12.751,
+      "step": 441000
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.0718589425453314e-05,
+      "loss": 0.2686,
+      "step": 441500
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 2.0538391696920015e-05,
+      "loss": 0.2687,
+      "step": 442000
+    },
+    {
+      "epoch": 11.26,
+      "eval_loss": 0.8066098690032959,
+      "eval_runtime": 1.3012,
+      "eval_samples_per_second": 768.545,
+      "eval_steps_per_second": 12.297,
+      "step": 442000
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 2.035966441899249e-05,
+      "loss": 0.2687,
+      "step": 442500
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.0182409546203555e-05,
+      "loss": 0.2688,
+      "step": 443000
+    },
+    {
+      "epoch": 11.29,
+      "eval_loss": 0.812473714351654,
+      "eval_runtime": 1.3079,
+      "eval_samples_per_second": 764.56,
+      "eval_steps_per_second": 12.233,
+      "step": 443000
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 2.000662901698415e-05,
+      "loss": 0.2689,
+      "step": 443500
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.983232475364195e-05,
+      "loss": 0.2686,
+      "step": 444000
+    },
+    {
+      "epoch": 11.32,
+      "eval_loss": 0.8055201768875122,
+      "eval_runtime": 1.3423,
+      "eval_samples_per_second": 744.998,
+      "eval_steps_per_second": 11.92,
+      "step": 444000
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.9659498662340474e-05,
+      "loss": 0.2686,
+      "step": 444500
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.948815263307819e-05,
+      "loss": 0.2686,
+      "step": 445000
+    },
+    {
+      "epoch": 11.34,
+      "eval_loss": 0.8064922094345093,
+      "eval_runtime": 1.2899,
+      "eval_samples_per_second": 775.27,
+      "eval_steps_per_second": 12.404,
+      "step": 445000
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.9318288539667765e-05,
+      "loss": 0.2686,
+      "step": 445500
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.914990823971574e-05,
+      "loss": 0.2685,
+      "step": 446000
+    },
+    {
+      "epoch": 11.37,
+      "eval_loss": 0.813357412815094,
+      "eval_runtime": 1.286,
+      "eval_samples_per_second": 777.596,
+      "eval_steps_per_second": 12.442,
+      "step": 446000
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.8983013574602096e-05,
+      "loss": 0.2687,
+      "step": 446500
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.8817606369460156e-05,
+      "loss": 0.2684,
+      "step": 447000
+    },
+    {
+      "epoch": 11.39,
+      "eval_loss": 0.806776225566864,
+      "eval_runtime": 1.2826,
+      "eval_samples_per_second": 779.686,
+      "eval_steps_per_second": 12.475,
+      "step": 447000
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.865368843315663e-05,
+      "loss": 0.2684,
+      "step": 447500
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.8491261558271762e-05,
+      "loss": 0.2683,
+      "step": 448000
+    },
+    {
+      "epoch": 11.42,
+      "eval_loss": 0.808623194694519,
+      "eval_runtime": 1.2653,
+      "eval_samples_per_second": 790.357,
+      "eval_steps_per_second": 12.646,
+      "step": 448000
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.833032752107986e-05,
+      "loss": 0.2682,
+      "step": 448500
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.817088808152978e-05,
+      "loss": 0.2684,
+      "step": 449000
+    },
+    {
+      "epoch": 11.44,
+      "eval_loss": 0.8024921417236328,
+      "eval_runtime": 1.2578,
+      "eval_samples_per_second": 795.067,
+      "eval_steps_per_second": 12.721,
+      "step": 449000
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.801294498322569e-05,
+      "loss": 0.2685,
+      "step": 449500
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.7856499953407978e-05,
+      "loss": 0.2682,
+      "step": 450000
+    },
+    {
+      "epoch": 11.47,
+      "eval_loss": 0.8072661757469177,
+      "eval_runtime": 1.2702,
+      "eval_samples_per_second": 787.258,
+      "eval_steps_per_second": 12.596,
+      "step": 450000
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.770155470293445e-05,
+      "loss": 0.2684,
+      "step": 450500
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.7548110926261522e-05,
+      "loss": 0.2682,
+      "step": 451000
+    },
+    {
+      "epoch": 11.49,
+      "eval_loss": 0.8042126297950745,
+      "eval_runtime": 1.3486,
+      "eval_samples_per_second": 741.514,
+      "eval_steps_per_second": 11.864,
+      "step": 451000
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.7396170301425777e-05,
+      "loss": 0.2683,
+      "step": 451500
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.7245734490025544e-05,
+      "loss": 0.2683,
+      "step": 452000
+    },
+    {
+      "epoch": 11.52,
+      "eval_loss": 0.8097262382507324,
+      "eval_runtime": 1.2797,
+      "eval_samples_per_second": 781.424,
+      "eval_steps_per_second": 12.503,
+      "step": 452000
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.7096805137202738e-05,
+      "loss": 0.2682,
+      "step": 452500
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 1.6949383871624917e-05,
+      "loss": 0.2678,
+      "step": 453000
+    },
+    {
+      "epoch": 11.54,
+      "eval_loss": 0.8061630725860596,
+      "eval_runtime": 1.2741,
+      "eval_samples_per_second": 784.845,
+      "eval_steps_per_second": 12.558,
+      "step": 453000
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.6803472305467368e-05,
+      "loss": 0.2683,
+      "step": 453500
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.665907203439568e-05,
+      "loss": 0.2678,
+      "step": 454000
+    },
+    {
+      "epoch": 11.57,
+      "eval_loss": 0.808397114276886,
+      "eval_runtime": 1.2841,
+      "eval_samples_per_second": 778.785,
+      "eval_steps_per_second": 12.461,
+      "step": 454000
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.6516184637548058e-05,
+      "loss": 0.2681,
+      "step": 454500
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.6374811677518142e-05,
+      "loss": 0.2681,
+      "step": 455000
+    },
+    {
+      "epoch": 11.6,
+      "eval_loss": 0.8134533762931824,
+      "eval_runtime": 1.2711,
+      "eval_samples_per_second": 786.7,
+      "eval_steps_per_second": 12.587,
+      "step": 455000
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.6234954700338025e-05,
+      "loss": 0.2679,
+      "step": 455500
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.6096615235461148e-05,
+      "loss": 0.2678,
+      "step": 456000
+    },
+    {
+      "epoch": 11.62,
+      "eval_loss": 0.8097538948059082,
+      "eval_runtime": 1.2712,
+      "eval_samples_per_second": 786.682,
+      "eval_steps_per_second": 12.587,
+      "step": 456000
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.59597947957458e-05,
+      "loss": 0.2678,
+      "step": 456500
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.5824494877438344e-05,
+      "loss": 0.2681,
+      "step": 457000
+    },
+    {
+      "epoch": 11.65,
+      "eval_loss": 0.8078885674476624,
+      "eval_runtime": 1.2534,
+      "eval_samples_per_second": 797.802,
+      "eval_steps_per_second": 12.765,
+      "step": 457000
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.569071696015702e-05,
+      "loss": 0.2679,
+      "step": 457500
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.555846250687569e-05,
+      "loss": 0.2679,
+      "step": 458000
+    },
+    {
+      "epoch": 11.67,
+      "eval_loss": 0.8051915764808655,
+      "eval_runtime": 1.2442,
+      "eval_samples_per_second": 803.754,
+      "eval_steps_per_second": 12.86,
+      "step": 458000
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.542773296390789e-05,
+      "loss": 0.2679,
+      "step": 458500
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.5298529760890945e-05,
+      "loss": 0.268,
+      "step": 459000
+    },
+    {
+      "epoch": 11.7,
+      "eval_loss": 0.8038084506988525,
+      "eval_runtime": 1.3255,
+      "eval_samples_per_second": 754.438,
+      "eval_steps_per_second": 12.071,
+      "step": 459000
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.5170854310770376e-05,
+      "loss": 0.2678,
+      "step": 459500
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.5044708009784457e-05,
+      "loss": 0.268,
+      "step": 460000
+    },
+    {
+      "epoch": 11.72,
+      "eval_loss": 0.8099783062934875,
+      "eval_runtime": 1.3008,
+      "eval_samples_per_second": 768.777,
+      "eval_steps_per_second": 12.3,
+      "step": 460000
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.4920092237448903e-05,
+      "loss": 0.2677,
+      "step": 460500
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.4797008356541874e-05,
+      "loss": 0.2677,
+      "step": 461000
+    },
+    {
+      "epoch": 11.75,
+      "eval_loss": 0.8057255744934082,
+      "eval_runtime": 1.2455,
+      "eval_samples_per_second": 802.869,
+      "eval_steps_per_second": 12.846,
+      "step": 461000
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.4675457713088947e-05,
+      "loss": 0.2676,
+      "step": 461500
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.4555441636348494e-05,
+      "loss": 0.2676,
+      "step": 462000
+    },
+    {
+      "epoch": 11.77,
+      "eval_loss": 0.8142445087432861,
+      "eval_runtime": 1.2424,
+      "eval_samples_per_second": 804.89,
+      "eval_steps_per_second": 12.878,
+      "step": 462000
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.4436961438797095e-05,
+      "loss": 0.2679,
+      "step": 462500
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.4320018416115206e-05,
+      "loss": 0.2679,
+      "step": 463000
+    },
+    {
+      "epoch": 11.8,
+      "eval_loss": 0.8076353073120117,
+      "eval_runtime": 1.3419,
+      "eval_samples_per_second": 745.235,
+      "eval_steps_per_second": 11.924,
+      "step": 463000
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.4204613847173003e-05,
+      "loss": 0.2674,
+      "step": 463500
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.4090748994016354e-05,
+      "loss": 0.2676,
+      "step": 464000
+    },
+    {
+      "epoch": 11.83,
+      "eval_loss": 0.8087449669837952,
+      "eval_runtime": 1.2685,
+      "eval_samples_per_second": 788.361,
+      "eval_steps_per_second": 12.614,
+      "step": 464000
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.3978425101853049e-05,
+      "loss": 0.2681,
+      "step": 464500
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.3867643399039165e-05,
+      "loss": 0.2677,
+      "step": 465000
+    },
+    {
+      "epoch": 11.85,
+      "eval_loss": 0.8065741062164307,
+      "eval_runtime": 1.2179,
+      "eval_samples_per_second": 821.068,
+      "eval_steps_per_second": 13.137,
+      "step": 465000
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.3758405097065648e-05,
+      "loss": 0.2676,
+      "step": 465500
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.3650711390545131e-05,
+      "loss": 0.2673,
+      "step": 466000
+    },
+    {
+      "epoch": 11.88,
+      "eval_loss": 0.8059167861938477,
+      "eval_runtime": 1.3496,
+      "eval_samples_per_second": 740.983,
+      "eval_steps_per_second": 11.856,
+      "step": 466000
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.3544563457198657e-05,
+      "loss": 0.2673,
+      "step": 466500
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.343996245784307e-05,
+      "loss": 0.2676,
+      "step": 467000
+    },
+    {
+      "epoch": 11.9,
+      "eval_loss": 0.8066999316215515,
+      "eval_runtime": 1.2927,
+      "eval_samples_per_second": 773.582,
+      "eval_steps_per_second": 12.377,
+      "step": 467000
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.3336909536378107e-05,
+      "loss": 0.2678,
+      "step": 467500
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.3235405819774022e-05,
+      "loss": 0.2675,
+      "step": 468000
+    },
+    {
+      "epoch": 11.93,
+      "eval_loss": 0.8042522072792053,
+      "eval_runtime": 1.2835,
+      "eval_samples_per_second": 779.135,
+      "eval_steps_per_second": 12.466,
+      "step": 468000
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.3135452418059208e-05,
+      "loss": 0.2675,
+      "step": 468500
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 1.3037050424308027e-05,
+      "loss": 0.2675,
+      "step": 469000
+    },
+    {
+      "epoch": 11.95,
+      "eval_loss": 0.8102657198905945,
+      "eval_runtime": 1.2622,
+      "eval_samples_per_second": 792.261,
+      "eval_steps_per_second": 12.676,
+      "step": 469000
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.2940200914628945e-05,
+      "loss": 0.2676,
+      "step": 469500
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.2844904948152644e-05,
+      "loss": 0.2673,
+      "step": 470000
+    },
+    {
+      "epoch": 11.98,
+      "eval_loss": 0.8092105984687805,
+      "eval_runtime": 1.2931,
+      "eval_samples_per_second": 773.322,
+      "eval_steps_per_second": 12.373,
+      "step": 470000
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.2751163567020592e-05,
+      "loss": 0.2674,
+      "step": 470500
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.2658977796373478e-05,
+      "loss": 0.2676,
+      "step": 471000
+    },
+    {
+      "epoch": 12.0,
+      "eval_loss": 0.8069332838058472,
+      "eval_runtime": 1.4521,
+      "eval_samples_per_second": 688.659,
+      "eval_steps_per_second": 11.019,
+      "step": 471000
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.2568348644340153e-05,
+      "loss": 0.2674,
+      "step": 471500
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.2479277102026465e-05,
+      "loss": 0.2673,
+      "step": 472000
+    },
+    {
+      "epoch": 12.03,
+      "eval_loss": 0.8061708211898804,
+      "eval_runtime": 1.3407,
+      "eval_samples_per_second": 745.886,
+      "eval_steps_per_second": 11.934,
+      "step": 472000
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.2391764143504556e-05,
+      "loss": 0.2672,
+      "step": 472500
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.2305810725802118e-05,
+      "loss": 0.2673,
+      "step": 473000
+    },
+    {
+      "epoch": 12.05,
+      "eval_loss": 0.8025499582290649,
+      "eval_runtime": 1.3181,
+      "eval_samples_per_second": 758.664,
+      "eval_steps_per_second": 12.139,
+      "step": 473000
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.222141778889195e-05,
+      "loss": 0.2674,
+      "step": 473500
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.2138586255681707e-05,
+      "loss": 0.2672,
+      "step": 474000
+    },
+    {
+      "epoch": 12.08,
+      "eval_loss": 0.8044158220291138,
+      "eval_runtime": 1.3667,
+      "eval_samples_per_second": 731.706,
+      "eval_steps_per_second": 11.707,
+      "step": 474000
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.2057317032003731e-05,
+      "loss": 0.2673,
+      "step": 474500
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.1977611006605263e-05,
+      "loss": 0.2671,
+      "step": 475000
+    },
+    {
+      "epoch": 12.11,
+      "eval_loss": 0.8067881464958191,
+      "eval_runtime": 1.3916,
+      "eval_samples_per_second": 718.588,
+      "eval_steps_per_second": 11.497,
+      "step": 475000
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.1899469051138602e-05,
+      "loss": 0.267,
+      "step": 475500
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.1822892020151667e-05,
+      "loss": 0.2672,
+      "step": 476000
+    },
+    {
+      "epoch": 12.13,
+      "eval_loss": 0.8038668632507324,
+      "eval_runtime": 1.2711,
+      "eval_samples_per_second": 786.735,
+      "eval_steps_per_second": 12.588,
+      "step": 476000
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.1747880751078614e-05,
+      "loss": 0.2673,
+      "step": 476500
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.1674436064230637e-05,
+      "loss": 0.2673,
+      "step": 477000
+    },
+    {
+      "epoch": 12.16,
+      "eval_loss": 0.8077553510665894,
+      "eval_runtime": 1.3519,
+      "eval_samples_per_second": 739.677,
+      "eval_steps_per_second": 11.835,
+      "step": 477000
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.1602558762787069e-05,
+      "loss": 0.2672,
+      "step": 477500
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.1532249632786582e-05,
+      "loss": 0.2671,
+      "step": 478000
+    },
+    {
+      "epoch": 12.18,
+      "eval_loss": 0.8061319589614868,
+      "eval_runtime": 1.3348,
+      "eval_samples_per_second": 749.202,
+      "eval_steps_per_second": 11.987,
+      "step": 478000
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.1463509443118552e-05,
+      "loss": 0.2672,
+      "step": 478500
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.1396338945514663e-05,
+      "loss": 0.2673,
+      "step": 479000
+    },
+    {
+      "epoch": 12.21,
+      "eval_loss": 0.802213728427887,
+      "eval_runtime": 1.3304,
+      "eval_samples_per_second": 751.663,
+      "eval_steps_per_second": 12.027,
+      "step": 479000
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.133073887454072e-05,
+      "loss": 0.267,
+      "step": 479500
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.1266709947588599e-05,
+      "loss": 0.267,
+      "step": 480000
+    },
+    {
+      "epoch": 12.23,
+      "eval_loss": 0.8064542412757874,
+      "eval_runtime": 1.3174,
+      "eval_samples_per_second": 759.093,
+      "eval_steps_per_second": 12.145,
+      "step": 480000
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.1204252864868377e-05,
+      "loss": 0.2669,
+      "step": 480500
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.1143368309400725e-05,
+      "loss": 0.2672,
+      "step": 481000
+    },
+    {
+      "epoch": 12.26,
+      "eval_loss": 0.8035285472869873,
+      "eval_runtime": 1.597,
+      "eval_samples_per_second": 626.161,
+      "eval_steps_per_second": 10.019,
+      "step": 481000
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.1084056947009348e-05,
+      "loss": 0.2671,
+      "step": 481500
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.1026319426313837e-05,
+      "loss": 0.267,
+      "step": 482000
+    },
+    {
+      "epoch": 12.28,
+      "eval_loss": 0.8039422035217285,
+      "eval_runtime": 1.2756,
+      "eval_samples_per_second": 783.963,
+      "eval_steps_per_second": 12.543,
+      "step": 482000
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.097015637872247e-05,
+      "loss": 0.2669,
+      "step": 482500
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.0915568418425301e-05,
+      "loss": 0.2669,
+      "step": 483000
+    },
+    {
+      "epoch": 12.31,
+      "eval_loss": 0.8074455857276917,
+      "eval_runtime": 1.3487,
+      "eval_samples_per_second": 741.458,
+      "eval_steps_per_second": 11.863,
+      "step": 483000
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.0862556142387571e-05,
+      "loss": 0.2673,
+      "step": 483500
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.081112013034298e-05,
+      "loss": 0.267,
+      "step": 484000
+    },
+    {
+      "epoch": 12.33,
+      "eval_loss": 0.8040180802345276,
+      "eval_runtime": 1.3584,
+      "eval_samples_per_second": 736.142,
+      "eval_steps_per_second": 11.778,
+      "step": 484000
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.0761260944787561e-05,
+      "loss": 0.267,
+      "step": 484500
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.0712979130973347e-05,
+      "loss": 0.267,
+      "step": 485000
+    },
+    {
+      "epoch": 12.36,
+      "eval_loss": 0.8028098940849304,
+      "eval_runtime": 1.3341,
+      "eval_samples_per_second": 749.592,
+      "eval_steps_per_second": 11.993,
+      "step": 485000
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.0666275216902535e-05,
+      "loss": 0.2668,
+      "step": 485500
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.0621149713321656e-05,
+      "loss": 0.2668,
+      "step": 486000
+    },
+    {
+      "epoch": 12.39,
+      "eval_loss": 0.8054640889167786,
+      "eval_runtime": 1.3289,
+      "eval_samples_per_second": 752.527,
+      "eval_steps_per_second": 12.04,
+      "step": 486000
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.0577603113715964e-05,
+      "loss": 0.2669,
+      "step": 486500
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.0535635894304106e-05,
+      "loss": 0.2669,
+      "step": 487000
+    },
+    {
+      "epoch": 12.41,
+      "eval_loss": 0.8062050342559814,
+      "eval_runtime": 1.3114,
+      "eval_samples_per_second": 762.54,
+      "eval_steps_per_second": 12.201,
+      "step": 487000
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.0495248514032875e-05,
+      "loss": 0.2669,
+      "step": 487500
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.045644141457218e-05,
+      "loss": 0.2669,
+      "step": 488000
+    },
+    {
+      "epoch": 12.44,
+      "eval_loss": 0.8053330779075623,
+      "eval_runtime": 1.3085,
+      "eval_samples_per_second": 764.21,
+      "eval_steps_per_second": 12.227,
+      "step": 488000
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.0419215020310254e-05,
+      "loss": 0.2671,
+      "step": 488500
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.0383569738348988e-05,
+      "loss": 0.267,
+      "step": 489000
+    },
+    {
+      "epoch": 12.46,
+      "eval_loss": 0.8089292049407959,
+      "eval_runtime": 1.3008,
+      "eval_samples_per_second": 768.73,
+      "eval_steps_per_second": 12.3,
+      "step": 489000
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.0349505958499436e-05,
+      "loss": 0.2671,
+      "step": 489500
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.0317024053277693e-05,
+      "loss": 0.267,
+      "step": 490000
+    },
+    {
+      "epoch": 12.49,
+      "eval_loss": 0.8080971837043762,
+      "eval_runtime": 1.3533,
+      "eval_samples_per_second": 738.941,
+      "eval_steps_per_second": 11.823,
+      "step": 490000
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.0286124377900624e-05,
+      "loss": 0.2668,
+      "step": 490500
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.0256807270282153e-05,
+      "loss": 0.267,
+      "step": 491000
+    },
+    {
+      "epoch": 12.51,
+      "eval_loss": 0.8052739500999451,
+      "eval_runtime": 1.2792,
+      "eval_samples_per_second": 781.716,
+      "eval_steps_per_second": 12.507,
+      "step": 491000
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.0229073051029455e-05,
+      "loss": 0.2669,
+      "step": 491500
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.020292202343952e-05,
+      "loss": 0.2668,
+      "step": 492000
+    },
+    {
+      "epoch": 12.54,
+      "eval_loss": 0.8052670955657959,
+      "eval_runtime": 1.3282,
+      "eval_samples_per_second": 752.92,
+      "eval_steps_per_second": 12.047,
+      "step": 492000
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.0178354473495813e-05,
+      "loss": 0.2667,
+      "step": 492500
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.0155370669865077e-05,
+      "loss": 0.2671,
+      "step": 493000
+    },
+    {
+      "epoch": 12.56,
+      "eval_loss": 0.8097056746482849,
+      "eval_runtime": 1.3313,
+      "eval_samples_per_second": 751.148,
+      "eval_steps_per_second": 12.018,
+      "step": 493000
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.0133970863894557e-05,
+      "loss": 0.267,
+      "step": 493500
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.0114155289609061e-05,
+      "loss": 0.267,
+      "step": 494000
+    },
+    {
+      "epoch": 12.59,
+      "eval_loss": 0.8087899684906006,
+      "eval_runtime": 1.3408,
+      "eval_samples_per_second": 745.8,
+      "eval_steps_per_second": 11.933,
+      "step": 494000
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.0095924163708572e-05,
+      "loss": 0.2669,
+      "step": 494500
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.0079277685565724e-05,
+      "loss": 0.2669,
+      "step": 495000
+    },
+    {
+      "epoch": 12.62,
+      "eval_loss": 0.8080546855926514,
+      "eval_runtime": 1.3886,
+      "eval_samples_per_second": 720.171,
+      "eval_steps_per_second": 11.523,
+      "step": 495000
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.0064216037223772e-05,
+      "loss": 0.2665,
+      "step": 495500
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.0050739383394454e-05,
+      "loss": 0.2667,
+      "step": 496000
+    },
+    {
+      "epoch": 12.64,
+      "eval_loss": 0.8046788573265076,
+      "eval_runtime": 1.367,
+      "eval_samples_per_second": 731.525,
+      "eval_steps_per_second": 11.704,
+      "step": 496000
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.003884787145633e-05,
+      "loss": 0.2666,
+      "step": 496500
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.002854163145305e-05,
+      "loss": 0.2667,
+      "step": 497000
+    },
+    {
+      "epoch": 12.67,
+      "eval_loss": 0.804325520992279,
+      "eval_runtime": 1.334,
+      "eval_samples_per_second": 749.623,
+      "eval_steps_per_second": 11.994,
+      "step": 497000
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.0019820776091995e-05,
+      "loss": 0.267,
+      "step": 497500
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.0012685400743077e-05,
+      "loss": 0.2669,
+      "step": 498000
+    },
+    {
+      "epoch": 12.69,
+      "eval_loss": 0.805090069770813,
+      "eval_runtime": 1.2804,
+      "eval_samples_per_second": 781.031,
+      "eval_steps_per_second": 12.496,
+      "step": 498000
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.0007135583437572e-05,
+      "loss": 0.2671,
+      "step": 498500
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.0003171384867436e-05,
+      "loss": 0.2669,
+      "step": 499000
+    },
+    {
+      "epoch": 12.72,
+      "eval_loss": 0.8085483312606812,
+      "eval_runtime": 1.3193,
+      "eval_samples_per_second": 757.977,
+      "eval_steps_per_second": 12.128,
+      "step": 499000
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.0000792848384467e-05,
+      "loss": 0.2669,
+      "step": 499500
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1e-05,
+      "loss": 0.2666,
+      "step": 500000
+    },
+    {
+      "epoch": 12.74,
+      "eval_loss": 0.8054067492485046,
+      "eval_runtime": 1.3574,
+      "eval_samples_per_second": 736.683,
+      "eval_steps_per_second": 11.787,
+      "step": 500000
+    },
+    {
+      "epoch": 12.74,
+      "step": 500000,
+      "total_flos": 1.597419519176846e+22,
+      "train_loss": 0.3053010062255859,
+      "train_runtime": 144737.6086,
+      "train_samples_per_second": 1768.718,
+      "train_steps_per_second": 3.455
+    }
+  ],
+  "max_steps": 500000,
+  "num_train_epochs": 13,
+  "total_flos": 1.597419519176846e+22,
+  "trial_name": null,
+  "trial_params": null
+}