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": 11.649852046879005,
+  "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": 0.9072,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 0.6779,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 0.6512336730957031,
+      "eval_runtime": 1.3899,
+      "eval_samples_per_second": 1581.436,
+      "eval_steps_per_second": 25.182,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 0.6516,
+      "step": 1500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 0.6505,
+      "step": 2000
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 0.6479398608207703,
+      "eval_runtime": 1.2819,
+      "eval_samples_per_second": 1714.632,
+      "eval_steps_per_second": 27.303,
+      "step": 2000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 0.6487,
+      "step": 2500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 0.6475,
+      "step": 3000
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 0.6461147665977478,
+      "eval_runtime": 1.2932,
+      "eval_samples_per_second": 1699.633,
+      "eval_steps_per_second": 27.064,
+      "step": 3000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.2e-05,
+      "loss": 0.6469,
+      "step": 3500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 0.6464,
+      "step": 4000
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 0.6451290249824524,
+      "eval_runtime": 1.2355,
+      "eval_samples_per_second": 1779.085,
+      "eval_steps_per_second": 28.329,
+      "step": 4000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 0.6459,
+      "step": 4500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 0.6455,
+      "step": 5000
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 0.6444417238235474,
+      "eval_runtime": 1.2972,
+      "eval_samples_per_second": 1694.435,
+      "eval_steps_per_second": 26.981,
+      "step": 5000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 0.6452,
+      "step": 5500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 0.6448,
+      "step": 6000
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 0.6438854336738586,
+      "eval_runtime": 1.2871,
+      "eval_samples_per_second": 1707.767,
+      "eval_steps_per_second": 27.194,
+      "step": 6000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.8e-05,
+      "loss": 0.64,
+      "step": 6500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.4e-05,
+      "loss": 0.6222,
+      "step": 7000
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 0.6123253703117371,
+      "eval_runtime": 1.2501,
+      "eval_samples_per_second": 1758.21,
+      "eval_steps_per_second": 27.997,
+      "step": 7000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 0.6115,
+      "step": 7500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 0.6019,
+      "step": 8000
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 0.5872684717178345,
+      "eval_runtime": 1.2778,
+      "eval_samples_per_second": 1720.181,
+      "eval_steps_per_second": 27.391,
+      "step": 8000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000102,
+      "loss": 0.5924,
+      "step": 8500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 0.5838,
+      "step": 9000
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 0.5689075589179993,
+      "eval_runtime": 1.2743,
+      "eval_samples_per_second": 1724.838,
+      "eval_steps_per_second": 27.466,
+      "step": 9000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 0.5737,
+      "step": 9500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 0.5612,
+      "step": 10000
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 0.5473007559776306,
+      "eval_runtime": 1.2648,
+      "eval_samples_per_second": 1737.882,
+      "eval_steps_per_second": 27.673,
+      "step": 10000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 0.5516,
+      "step": 10500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 0.5428,
+      "step": 11000
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 0.5300613045692444,
+      "eval_runtime": 1.2658,
+      "eval_samples_per_second": 1736.485,
+      "eval_steps_per_second": 27.651,
+      "step": 11000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000138,
+      "loss": 0.5346,
+      "step": 11500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 0.527,
+      "step": 12000
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 0.5140743255615234,
+      "eval_runtime": 1.2609,
+      "eval_samples_per_second": 1743.192,
+      "eval_steps_per_second": 27.758,
+      "step": 12000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015,
+      "loss": 0.5203,
+      "step": 12500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000156,
+      "loss": 0.5142,
+      "step": 13000
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 0.5022566318511963,
+      "eval_runtime": 1.2481,
+      "eval_samples_per_second": 1761.066,
+      "eval_steps_per_second": 28.042,
+      "step": 13000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000162,
+      "loss": 0.5079,
+      "step": 13500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000168,
+      "loss": 0.5002,
+      "step": 14000
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 0.4841987192630768,
+      "eval_runtime": 1.3176,
+      "eval_samples_per_second": 1668.178,
+      "eval_steps_per_second": 26.563,
+      "step": 14000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 0.4916,
+      "step": 14500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 0.4852,
+      "step": 15000
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 0.47066375613212585,
+      "eval_runtime": 1.3026,
+      "eval_samples_per_second": 1687.422,
+      "eval_steps_per_second": 26.87,
+      "step": 15000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000186,
+      "loss": 0.4778,
+      "step": 15500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 0.4697,
+      "step": 16000
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 0.4562186300754547,
+      "eval_runtime": 1.2895,
+      "eval_samples_per_second": 1704.503,
+      "eval_steps_per_second": 27.142,
+      "step": 16000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000198,
+      "loss": 0.4636,
+      "step": 16500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000204,
+      "loss": 0.459,
+      "step": 17000
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 0.44295796751976013,
+      "eval_runtime": 1.2516,
+      "eval_samples_per_second": 1756.093,
+      "eval_steps_per_second": 27.963,
+      "step": 17000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 0.4544,
+      "step": 17500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 0.45,
+      "step": 18000
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 0.4343169629573822,
+      "eval_runtime": 1.327,
+      "eval_samples_per_second": 1656.351,
+      "eval_steps_per_second": 26.375,
+      "step": 18000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 0.4458,
+      "step": 18500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 0.4411,
+      "step": 19000
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 0.42496559023857117,
+      "eval_runtime": 1.3285,
+      "eval_samples_per_second": 1654.492,
+      "eval_steps_per_second": 26.345,
+      "step": 19000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000234,
+      "loss": 0.4368,
+      "step": 19500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 0.4322,
+      "step": 20000
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 0.4132905900478363,
+      "eval_runtime": 1.3158,
+      "eval_samples_per_second": 1670.425,
+      "eval_steps_per_second": 26.599,
+      "step": 20000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 0.4278,
+      "step": 20500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 0.4233,
+      "step": 21000
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 0.4032529294490814,
+      "eval_runtime": 1.2842,
+      "eval_samples_per_second": 1711.577,
+      "eval_steps_per_second": 27.254,
+      "step": 21000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000258,
+      "loss": 0.4189,
+      "step": 21500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 0.4146,
+      "step": 22000
+    },
+    {
+      "epoch": 0.51,
+      "eval_loss": 0.39436739683151245,
+      "eval_runtime": 1.2633,
+      "eval_samples_per_second": 1739.901,
+      "eval_steps_per_second": 27.705,
+      "step": 22000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027,
+      "loss": 0.4109,
+      "step": 22500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000276,
+      "loss": 0.4077,
+      "step": 23000
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 0.38647642731666565,
+      "eval_runtime": 1.2503,
+      "eval_samples_per_second": 1758.021,
+      "eval_steps_per_second": 27.994,
+      "step": 23000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 0.4047,
+      "step": 23500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 0.4019,
+      "step": 24000
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 0.381785124540329,
+      "eval_runtime": 1.2485,
+      "eval_samples_per_second": 1760.518,
+      "eval_steps_per_second": 28.034,
+      "step": 24000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000294,
+      "loss": 0.3994,
+      "step": 24500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003,
+      "loss": 0.3968,
+      "step": 25000
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 0.3785216510295868,
+      "eval_runtime": 1.2913,
+      "eval_samples_per_second": 1702.153,
+      "eval_steps_per_second": 27.104,
+      "step": 25000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029999920715161553,
+      "loss": 0.3945,
+      "step": 25500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002999968286151326,
+      "loss": 0.392,
+      "step": 26000
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 0.37318962812423706,
+      "eval_runtime": 1.2758,
+      "eval_samples_per_second": 1722.782,
+      "eval_steps_per_second": 27.433,
+      "step": 26000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002999928644165624,
+      "loss": 0.3897,
+      "step": 26500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002999873145992569,
+      "loss": 0.3869,
+      "step": 27000
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 0.3669659197330475,
+      "eval_runtime": 1.2869,
+      "eval_samples_per_second": 1707.967,
+      "eval_steps_per_second": 27.197,
+      "step": 27000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000299980179223908,
+      "loss": 0.385,
+      "step": 27500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002999714583685469,
+      "loss": 0.3828,
+      "step": 28000
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 0.3615981340408325,
+      "eval_runtime": 1.2961,
+      "eval_samples_per_second": 1695.914,
+      "eval_steps_per_second": 27.005,
+      "step": 28000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00029996115212854366,
+      "loss": 0.3808,
+      "step": 28500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00029994926061660554,
+      "loss": 0.3786,
+      "step": 29000
+    },
+    {
+      "epoch": 0.68,
+      "eval_loss": 0.3613170087337494,
+      "eval_runtime": 1.3013,
+      "eval_samples_per_second": 1689.114,
+      "eval_steps_per_second": 26.897,
+      "step": 29000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002999357839627762,
+      "loss": 0.377,
+      "step": 29500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00029992072231443425,
+      "loss": 0.3751,
+      "step": 30000
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 0.35551005601882935,
+      "eval_runtime": 1.2661,
+      "eval_samples_per_second": 1736.053,
+      "eval_steps_per_second": 27.644,
+      "step": 30000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002999040758362914,
+      "loss": 0.3731,
+      "step": 30500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00029988584471039094,
+      "loss": 0.3713,
+      "step": 31000
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 0.35287660360336304,
+      "eval_runtime": 1.3423,
+      "eval_samples_per_second": 1637.545,
+      "eval_steps_per_second": 26.076,
+      "step": 31000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002998660291361054,
+      "loss": 0.3699,
+      "step": 31500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002998446293301349,
+      "loss": 0.3681,
+      "step": 32000
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.35174888372421265,
+      "eval_runtime": 1.3006,
+      "eval_samples_per_second": 1690.029,
+      "eval_steps_per_second": 26.911,
+      "step": 32000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002998216455265042,
+      "loss": 0.3667,
+      "step": 32500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029979707797656046,
+      "loss": 0.3656,
+      "step": 33000
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 0.34766125679016113,
+      "eval_runtime": 1.3363,
+      "eval_samples_per_second": 1644.872,
+      "eval_steps_per_second": 26.192,
+      "step": 33000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029977092694897053,
+      "loss": 0.3641,
+      "step": 33500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002997431927297178,
+      "loss": 0.3626,
+      "step": 34000
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 0.3423311412334442,
+      "eval_runtime": 1.3208,
+      "eval_samples_per_second": 1664.116,
+      "eval_steps_per_second": 26.499,
+      "step": 34000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029971387562209936,
+      "loss": 0.3616,
+      "step": 34500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029968297594672226,
+      "loss": 0.3601,
+      "step": 35000
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 0.34427037835121155,
+      "eval_runtime": 1.2929,
+      "eval_samples_per_second": 1700.099,
+      "eval_steps_per_second": 27.072,
+      "step": 35000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002996504940415005,
+      "loss": 0.3589,
+      "step": 35500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029961643026165096,
+      "loss": 0.3577,
+      "step": 36000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 0.34002143144607544,
+      "eval_runtime": 1.2974,
+      "eval_samples_per_second": 1694.186,
+      "eval_steps_per_second": 26.977,
+      "step": 36000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029958078497968973,
+      "loss": 0.3568,
+      "step": 36500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002995435585854278,
+      "loss": 0.3555,
+      "step": 37000
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 0.3368704915046692,
+      "eval_runtime": 1.3166,
+      "eval_samples_per_second": 1669.417,
+      "eval_steps_per_second": 26.583,
+      "step": 37000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002995047514859671,
+      "loss": 0.3542,
+      "step": 37500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002994643641056959,
+      "loss": 0.3532,
+      "step": 38000
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 0.3357352018356323,
+      "eval_runtime": 1.2847,
+      "eval_samples_per_second": 1710.953,
+      "eval_steps_per_second": 27.244,
+      "step": 38000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000299422396886284,
+      "loss": 0.3523,
+      "step": 38500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002993788502866783,
+      "loss": 0.3514,
+      "step": 39000
+    },
+    {
+      "epoch": 0.91,
+      "eval_loss": 0.33294928073883057,
+      "eval_runtime": 1.272,
+      "eval_samples_per_second": 1728.032,
+      "eval_steps_per_second": 27.516,
+      "step": 39000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029933372478309746,
+      "loss": 0.3503,
+      "step": 39500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029928702086902664,
+      "loss": 0.3493,
+      "step": 40000
+    },
+    {
+      "epoch": 0.93,
+      "eval_loss": 0.332314670085907,
+      "eval_runtime": 1.318,
+      "eval_samples_per_second": 1667.699,
+      "eval_steps_per_second": 26.556,
+      "step": 40000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029923873905521244,
+      "loss": 0.3483,
+      "step": 40500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000299188879869657,
+      "loss": 0.3477,
+      "step": 41000
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 0.33138296008110046,
+      "eval_runtime": 1.2875,
+      "eval_samples_per_second": 1707.216,
+      "eval_steps_per_second": 27.185,
+      "step": 41000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029913744385761244,
+      "loss": 0.3466,
+      "step": 41500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00029908443158157465,
+      "loss": 0.3459,
+      "step": 42000
+    },
+    {
+      "epoch": 0.98,
+      "eval_loss": 0.32563212513923645,
+      "eval_runtime": 1.2983,
+      "eval_samples_per_second": 1692.964,
+      "eval_steps_per_second": 26.958,
+      "step": 42000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002990298436212775,
+      "loss": 0.3448,
+      "step": 42500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002989736805736861,
+      "loss": 0.3441,
+      "step": 43000
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.32700827717781067,
+      "eval_runtime": 1.3967,
+      "eval_samples_per_second": 1573.716,
+      "eval_steps_per_second": 25.059,
+      "step": 43000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00029891594305299065,
+      "loss": 0.3432,
+      "step": 43500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029885663169059926,
+      "loss": 0.3424,
+      "step": 44000
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 0.3216361701488495,
+      "eval_runtime": 1.4298,
+      "eval_samples_per_second": 1537.25,
+      "eval_steps_per_second": 24.479,
+      "step": 44000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002987957471351316,
+      "loss": 0.3415,
+      "step": 44500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029873329005241137,
+      "loss": 0.3408,
+      "step": 45000
+    },
+    {
+      "epoch": 1.05,
+      "eval_loss": 0.324531614780426,
+      "eval_runtime": 1.3974,
+      "eval_samples_per_second": 1572.95,
+      "eval_steps_per_second": 25.047,
+      "step": 45000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029866926112545925,
+      "loss": 0.3399,
+      "step": 45500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029860366105448534,
+      "loss": 0.3394,
+      "step": 46000
+    },
+    {
+      "epoch": 1.07,
+      "eval_loss": 0.3231622278690338,
+      "eval_runtime": 1.3921,
+      "eval_samples_per_second": 1578.889,
+      "eval_steps_per_second": 25.142,
+      "step": 46000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029853649055688143,
+      "loss": 0.3387,
+      "step": 46500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00029846775036721337,
+      "loss": 0.338,
+      "step": 47000
+    },
+    {
+      "epoch": 1.1,
+      "eval_loss": 0.32254886627197266,
+      "eval_runtime": 1.4182,
+      "eval_samples_per_second": 1549.86,
+      "eval_steps_per_second": 24.679,
+      "step": 47000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002983974412372129,
+      "loss": 0.337,
+      "step": 47500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00029832556393576934,
+      "loss": 0.3363,
+      "step": 48000
+    },
+    {
+      "epoch": 1.12,
+      "eval_loss": 0.3195147216320038,
+      "eval_runtime": 1.408,
+      "eval_samples_per_second": 1561.07,
+      "eval_steps_per_second": 24.858,
+      "step": 48000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002982521192489214,
+      "loss": 0.3357,
+      "step": 48500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002981771079798483,
+      "loss": 0.3353,
+      "step": 49000
+    },
+    {
+      "epoch": 1.14,
+      "eval_loss": 0.3167419731616974,
+      "eval_runtime": 1.4463,
+      "eval_samples_per_second": 1519.733,
+      "eval_steps_per_second": 24.2,
+      "step": 49000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00029810053094886136,
+      "loss": 0.3345,
+      "step": 49500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00029802238899339473,
+      "loss": 0.3338,
+      "step": 50000
+    },
+    {
+      "epoch": 1.16,
+      "eval_loss": 0.3171909749507904,
+      "eval_runtime": 1.4105,
+      "eval_samples_per_second": 1558.283,
+      "eval_steps_per_second": 24.813,
+      "step": 50000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002979426829679962,
+      "loss": 0.3334,
+      "step": 50500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002978614137443183,
+      "loss": 0.3325,
+      "step": 51000
+    },
+    {
+      "epoch": 1.19,
+      "eval_loss": 0.31682637333869934,
+      "eval_runtime": 1.4213,
+      "eval_samples_per_second": 1546.472,
+      "eval_steps_per_second": 24.625,
+      "step": 51000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000297778582211108,
+      "loss": 0.3322,
+      "step": 51500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00029769418927419786,
+      "loss": 0.3316,
+      "step": 52000
+    },
+    {
+      "epoch": 1.21,
+      "eval_loss": 0.31410133838653564,
+      "eval_runtime": 1.4261,
+      "eval_samples_per_second": 1541.269,
+      "eval_steps_per_second": 24.542,
+      "step": 52000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002976082358564954,
+      "loss": 0.331,
+      "step": 52500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00029752072289797353,
+      "loss": 0.3305,
+      "step": 53000
+    },
+    {
+      "epoch": 1.23,
+      "eval_loss": 0.3114263713359833,
+      "eval_runtime": 1.4182,
+      "eval_samples_per_second": 1549.888,
+      "eval_steps_per_second": 24.68,
+      "step": 53000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00029743165135565986,
+      "loss": 0.3301,
+      "step": 53500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00029734102220362654,
+      "loss": 0.3303,
+      "step": 54000
+    },
+    {
+      "epoch": 1.26,
+      "eval_loss": 0.3110273778438568,
+      "eval_runtime": 1.3899,
+      "eval_samples_per_second": 1581.462,
+      "eval_steps_per_second": 25.183,
+      "step": 54000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00029724883643297937,
+      "loss": 0.3289,
+      "step": 54500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002971550950518473,
+      "loss": 0.3284,
+      "step": 55000
+    },
+    {
+      "epoch": 1.28,
+      "eval_loss": 0.3112471401691437,
+      "eval_runtime": 1.3713,
+      "eval_samples_per_second": 1602.906,
+      "eval_steps_per_second": 25.524,
+      "step": 55000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000297059799085371,
+      "loss": 0.3284,
+      "step": 55500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00029696294957569196,
+      "loss": 0.3276,
+      "step": 56000
+    },
+    {
+      "epoch": 1.3,
+      "eval_loss": 0.30989283323287964,
+      "eval_runtime": 1.4309,
+      "eval_samples_per_second": 1536.096,
+      "eval_steps_per_second": 24.46,
+      "step": 56000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00029686454758194076,
+      "loss": 0.3269,
+      "step": 56500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00029676459418022594,
+      "loss": 0.3264,
+      "step": 57000
+    },
+    {
+      "epoch": 1.33,
+      "eval_loss": 0.3089369833469391,
+      "eval_runtime": 1.4226,
+      "eval_samples_per_second": 1545.068,
+      "eval_steps_per_second": 24.603,
+      "step": 57000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002966630904636219,
+      "loss": 0.3261,
+      "step": 57500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002965600375421569,
+      "loss": 0.3256,
+      "step": 58000
+    },
+    {
+      "epoch": 1.35,
+      "eval_loss": 0.31043851375579834,
+      "eval_runtime": 1.4043,
+      "eval_samples_per_second": 1565.245,
+      "eval_steps_per_second": 24.924,
+      "step": 58000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002964554365428013,
+      "loss": 0.3251,
+      "step": 58500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00029634928860945486,
+      "loss": 0.3247,
+      "step": 59000
+    },
+    {
+      "epoch": 1.37,
+      "eval_loss": 0.3088381588459015,
+      "eval_runtime": 1.4281,
+      "eval_samples_per_second": 1539.081,
+      "eval_steps_per_second": 24.508,
+      "step": 59000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002962415949029343,
+      "loss": 0.3243,
+      "step": 59500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00029613235660096084,
+      "loss": 0.3241,
+      "step": 60000
+    },
+    {
+      "epoch": 1.4,
+      "eval_loss": 0.3055568337440491,
+      "eval_runtime": 1.4302,
+      "eval_samples_per_second": 1536.88,
+      "eval_steps_per_second": 24.473,
+      "step": 60000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00029602157489814693,
+      "loss": 0.3234,
+      "step": 60500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00029590925100598365,
+      "loss": 0.3229,
+      "step": 61000
+    },
+    {
+      "epoch": 1.42,
+      "eval_loss": 0.3047891855239868,
+      "eval_runtime": 1.4211,
+      "eval_samples_per_second": 1546.647,
+      "eval_steps_per_second": 24.628,
+      "step": 61000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000295795386152827,
+      "loss": 0.323,
+      "step": 61500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002956799815838848,
+      "loss": 0.3223,
+      "step": 62000
+    },
+    {
+      "epoch": 1.44,
+      "eval_loss": 0.30619609355926514,
+      "eval_runtime": 1.3977,
+      "eval_samples_per_second": 1572.599,
+      "eval_steps_per_second": 25.041,
+      "step": 62000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002955630385612029,
+      "loss": 0.3217,
+      "step": 62500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002954445583636515,
+      "loss": 0.3215,
+      "step": 63000
+    },
+    {
+      "epoch": 1.47,
+      "eval_loss": 0.3042140007019043,
+      "eval_runtime": 1.3779,
+      "eval_samples_per_second": 1595.193,
+      "eval_steps_per_second": 25.401,
+      "step": 63000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00029532454228691103,
+      "loss": 0.3212,
+      "step": 63500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002952029916434581,
+      "loss": 0.3208,
+      "step": 64000
+    },
+    {
+      "epoch": 1.49,
+      "eval_loss": 0.30290499329566956,
+      "eval_runtime": 1.3826,
+      "eval_samples_per_second": 1589.79,
+      "eval_steps_per_second": 25.315,
+      "step": 64000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00029507990776255107,
+      "loss": 0.3206,
+      "step": 64500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00029495529199021555,
+      "loss": 0.3201,
+      "step": 65000
+    },
+    {
+      "epoch": 1.51,
+      "eval_loss": 0.30093201994895935,
+      "eval_runtime": 1.3496,
+      "eval_samples_per_second": 1628.653,
+      "eval_steps_per_second": 25.934,
+      "step": 65000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002948291456892296,
+      "loss": 0.3199,
+      "step": 65500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00029470147023910907,
+      "loss": 0.3198,
+      "step": 66000
+    },
+    {
+      "epoch": 1.54,
+      "eval_loss": 0.3037390410900116,
+      "eval_runtime": 1.4125,
+      "eval_samples_per_second": 1556.101,
+      "eval_steps_per_second": 24.779,
+      "step": 66000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002945722670360921,
+      "loss": 0.3189,
+      "step": 66500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002944415374931243,
+      "loss": 0.3187,
+      "step": 67000
+    },
+    {
+      "epoch": 1.56,
+      "eval_loss": 0.30209919810295105,
+      "eval_runtime": 1.3711,
+      "eval_samples_per_second": 1603.144,
+      "eval_steps_per_second": 25.528,
+      "step": 67000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00029430928303984295,
+      "loss": 0.3184,
+      "step": 67500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002941755051225616,
+      "loss": 0.3179,
+      "step": 68000
+    },
+    {
+      "epoch": 1.58,
+      "eval_loss": 0.29896965622901917,
+      "eval_runtime": 1.5715,
+      "eval_samples_per_second": 1398.699,
+      "eval_steps_per_second": 22.272,
+      "step": 68000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00029404020520425417,
+      "loss": 0.3177,
+      "step": 68500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002939033847645388,
+      "loss": 0.3176,
+      "step": 69000
+    },
+    {
+      "epoch": 1.61,
+      "eval_loss": 0.30087536573410034,
+      "eval_runtime": 1.4497,
+      "eval_samples_per_second": 1516.182,
+      "eval_steps_per_second": 24.143,
+      "step": 69000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00029376504529966195,
+      "loss": 0.3172,
+      "step": 69500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00029362518832248184,
+      "loss": 0.3169,
+      "step": 70000
+    },
+    {
+      "epoch": 1.63,
+      "eval_loss": 0.3002066910266876,
+      "eval_runtime": 1.4297,
+      "eval_samples_per_second": 1537.403,
+      "eval_steps_per_second": 24.481,
+      "step": 70000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002934838153624519,
+      "loss": 0.3166,
+      "step": 70500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00029334092796560427,
+      "loss": 0.3161,
+      "step": 71000
+    },
+    {
+      "epoch": 1.65,
+      "eval_loss": 0.2996619939804077,
+      "eval_runtime": 1.4139,
+      "eval_samples_per_second": 1554.517,
+      "eval_steps_per_second": 24.753,
+      "step": 71000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002931965276945326,
+      "loss": 0.3159,
+      "step": 71500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002930506161283751,
+      "loss": 0.3156,
+      "step": 72000
+    },
+    {
+      "epoch": 1.68,
+      "eval_loss": 0.29820874333381653,
+      "eval_runtime": 1.408,
+      "eval_samples_per_second": 1561.093,
+      "eval_steps_per_second": 24.858,
+      "step": 72000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00029290319486279724,
+      "loss": 0.3152,
+      "step": 72500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002927542655099744,
+      "loss": 0.3148,
+      "step": 73000
+    },
+    {
+      "epoch": 1.7,
+      "eval_loss": 0.2982105612754822,
+      "eval_runtime": 1.3918,
+      "eval_samples_per_second": 1579.253,
+      "eval_steps_per_second": 25.147,
+      "step": 73000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00029260382969857417,
+      "loss": 0.3148,
+      "step": 73500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00029245188907373845,
+      "loss": 0.3143,
+      "step": 74000
+    },
+    {
+      "epoch": 1.72,
+      "eval_loss": 0.2960808277130127,
+      "eval_runtime": 1.4226,
+      "eval_samples_per_second": 1545.079,
+      "eval_steps_per_second": 24.603,
+      "step": 74000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002922984452970655,
+      "loss": 0.3141,
+      "step": 74500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000292143500046592,
+      "loss": 0.3137,
+      "step": 75000
+    },
+    {
+      "epoch": 1.75,
+      "eval_loss": 0.2935367524623871,
+      "eval_runtime": 1.3992,
+      "eval_samples_per_second": 1570.943,
+      "eval_steps_per_second": 25.015,
+      "step": 75000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002919870550167743,
+      "loss": 0.3133,
+      "step": 75500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002918291119184702,
+      "loss": 0.3132,
+      "step": 76000
+    },
+    {
+      "epoch": 1.77,
+      "eval_loss": 0.2974264919757843,
+      "eval_runtime": 1.3828,
+      "eval_samples_per_second": 1589.484,
+      "eval_steps_per_second": 25.31,
+      "step": 76000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002916696724789201,
+      "loss": 0.3134,
+      "step": 76500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00029150873844172823,
+      "loss": 0.313,
+      "step": 77000
+    },
+    {
+      "epoch": 1.79,
+      "eval_loss": 0.29650062322616577,
+      "eval_runtime": 1.4037,
+      "eval_samples_per_second": 1565.884,
+      "eval_steps_per_second": 24.934,
+      "step": 77000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00029134631156684334,
+      "loss": 0.3128,
+      "step": 77500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002911823936305398,
+      "loss": 0.3147,
+      "step": 78000
+    },
+    {
+      "epoch": 1.82,
+      "eval_loss": 0.2946864664554596,
+      "eval_runtime": 1.3928,
+      "eval_samples_per_second": 1578.113,
+      "eval_steps_per_second": 25.129,
+      "step": 78000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002910169864253979,
+      "loss": 0.3123,
+      "step": 78500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002908500917602842,
+      "loss": 0.3118,
+      "step": 79000
+    },
+    {
+      "epoch": 1.84,
+      "eval_loss": 0.2940049469470978,
+      "eval_runtime": 1.405,
+      "eval_samples_per_second": 1564.438,
+      "eval_steps_per_second": 24.911,
+      "step": 79000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00029068171146033226,
+      "loss": 0.3115,
+      "step": 79500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002905118473669218,
+      "loss": 0.3113,
+      "step": 80000
+    },
+    {
+      "epoch": 1.86,
+      "eval_loss": 0.293503075838089,
+      "eval_runtime": 1.4595,
+      "eval_samples_per_second": 1505.956,
+      "eval_steps_per_second": 23.98,
+      "step": 80000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00029034050133765947,
+      "loss": 0.3111,
+      "step": 80500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00029016767524635804,
+      "loss": 0.3109,
+      "step": 81000
+    },
+    {
+      "epoch": 1.89,
+      "eval_loss": 0.2935539484024048,
+      "eval_runtime": 1.3879,
+      "eval_samples_per_second": 1583.739,
+      "eval_steps_per_second": 25.219,
+      "step": 81000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00028999337098301585,
+      "loss": 0.3108,
+      "step": 81500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002898175904537964,
+      "loss": 0.3103,
+      "step": 82000
+    },
+    {
+      "epoch": 1.91,
+      "eval_loss": 0.2912616431713104,
+      "eval_runtime": 1.3921,
+      "eval_samples_per_second": 1578.853,
+      "eval_steps_per_second": 25.141,
+      "step": 82000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002896403355810075,
+      "loss": 0.3104,
+      "step": 82500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00028946160830307997,
+      "loss": 0.3102,
+      "step": 83000
+    },
+    {
+      "epoch": 1.93,
+      "eval_loss": 0.29547759890556335,
+      "eval_runtime": 1.4163,
+      "eval_samples_per_second": 1551.936,
+      "eval_steps_per_second": 24.712,
+      "step": 83000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00028928141057454665,
+      "loss": 0.3099,
+      "step": 83500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002890997443660211,
+      "loss": 0.3096,
+      "step": 84000
+    },
+    {
+      "epoch": 1.96,
+      "eval_loss": 0.29173827171325684,
+      "eval_runtime": 1.3888,
+      "eval_samples_per_second": 1582.667,
+      "eval_steps_per_second": 25.202,
+      "step": 84000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00028891661166417586,
+      "loss": 0.3089,
+      "step": 84500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00028873201447172074,
+      "loss": 0.3093,
+      "step": 85000
+    },
+    {
+      "epoch": 1.98,
+      "eval_loss": 0.2930682897567749,
+      "eval_runtime": 1.4086,
+      "eval_samples_per_second": 1560.366,
+      "eval_steps_per_second": 24.847,
+      "step": 85000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002885459548073812,
+      "loss": 0.309,
+      "step": 85500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002883584347058758,
+      "loss": 0.3084,
+      "step": 86000
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.2906832695007324,
+      "eval_runtime": 1.4896,
+      "eval_samples_per_second": 1475.535,
+      "eval_steps_per_second": 23.496,
+      "step": 86000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00028816945621789437,
+      "loss": 0.3086,
+      "step": 86500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002879790214100753,
+      "loss": 0.3086,
+      "step": 87000
+    },
+    {
+      "epoch": 2.03,
+      "eval_loss": 0.2908460795879364,
+      "eval_runtime": 1.4139,
+      "eval_samples_per_second": 1554.553,
+      "eval_steps_per_second": 24.754,
+      "step": 87000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002877871323649833,
+      "loss": 0.3078,
+      "step": 87500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002875937911810861,
+      "loss": 0.3077,
+      "step": 88000
+    },
+    {
+      "epoch": 2.05,
+      "eval_loss": 0.2881493866443634,
+      "eval_runtime": 1.4042,
+      "eval_samples_per_second": 1565.313,
+      "eval_steps_per_second": 24.925,
+      "step": 88000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.000287398999972732,
+      "loss": 0.3075,
+      "step": 88500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00028720276087012636,
+      "loss": 0.3087,
+      "step": 89000
+    },
+    {
+      "epoch": 2.07,
+      "eval_loss": 0.2877899706363678,
+      "eval_runtime": 1.3911,
+      "eval_samples_per_second": 1580.034,
+      "eval_steps_per_second": 25.16,
+      "step": 89000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002870050760193086,
+      "loss": 0.3067,
+      "step": 89500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00028680594758212854,
+      "loss": 0.3065,
+      "step": 90000
+    },
+    {
+      "epoch": 2.1,
+      "eval_loss": 0.2883204519748688,
+      "eval_runtime": 1.3997,
+      "eval_samples_per_second": 1570.392,
+      "eval_steps_per_second": 25.006,
+      "step": 90000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00028660537773622294,
+      "loss": 0.3073,
+      "step": 90500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00028640336867499143,
+      "loss": 0.3063,
+      "step": 91000
+    },
+    {
+      "epoch": 2.12,
+      "eval_loss": 0.28914499282836914,
+      "eval_runtime": 1.4031,
+      "eval_samples_per_second": 1566.538,
+      "eval_steps_per_second": 24.945,
+      "step": 91000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002861999226075728,
+      "loss": 0.306,
+      "step": 91500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002859950417588206,
+      "loss": 0.3057,
+      "step": 92000
+    },
+    {
+      "epoch": 2.14,
+      "eval_loss": 0.28657251596450806,
+      "eval_runtime": 1.3993,
+      "eval_samples_per_second": 1570.823,
+      "eval_steps_per_second": 25.013,
+      "step": 92000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00028578872836927904,
+      "loss": 0.3057,
+      "step": 92500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002855809846951582,
+      "loss": 0.3066,
+      "step": 93000
+    },
+    {
+      "epoch": 2.17,
+      "eval_loss": 0.2893889248371124,
+      "eval_runtime": 1.4241,
+      "eval_samples_per_second": 1543.421,
+      "eval_steps_per_second": 24.577,
+      "step": 93000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00028537181300830963,
+      "loss": 0.305,
+      "step": 93500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002851612155962014,
+      "loss": 0.3048,
+      "step": 94000
+    },
+    {
+      "epoch": 2.19,
+      "eval_loss": 0.2858821451663971,
+      "eval_runtime": 1.4056,
+      "eval_samples_per_second": 1563.797,
+      "eval_steps_per_second": 24.901,
+      "step": 94000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002849491947618932,
+      "loss": 0.3048,
+      "step": 94500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002847357528240107,
+      "loss": 0.3051,
+      "step": 95000
+    },
+    {
+      "epoch": 2.21,
+      "eval_loss": 0.28810954093933105,
+      "eval_runtime": 1.4039,
+      "eval_samples_per_second": 1565.629,
+      "eval_steps_per_second": 24.93,
+      "step": 95000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002845208921167208,
+      "loss": 0.3047,
+      "step": 95500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00028430461498970584,
+      "loss": 0.3041,
+      "step": 96000
+    },
+    {
+      "epoch": 2.24,
+      "eval_loss": 0.28588977456092834,
+      "eval_runtime": 1.4454,
+      "eval_samples_per_second": 1520.728,
+      "eval_steps_per_second": 24.215,
+      "step": 96000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00028408692380813775,
+      "loss": 0.3042,
+      "step": 96500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00028386782095265247,
+      "loss": 0.3039,
+      "step": 97000
+    },
+    {
+      "epoch": 2.26,
+      "eval_loss": 0.2837139666080475,
+      "eval_runtime": 1.414,
+      "eval_samples_per_second": 1554.434,
+      "eval_steps_per_second": 24.752,
+      "step": 97000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002836473088193237,
+      "loss": 0.3036,
+      "step": 97500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00028342538981963677,
+      "loss": 0.3032,
+      "step": 98000
+    },
+    {
+      "epoch": 2.28,
+      "eval_loss": 0.28603771328926086,
+      "eval_runtime": 1.4079,
+      "eval_samples_per_second": 1561.179,
+      "eval_steps_per_second": 24.86,
+      "step": 98000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002832020663804624,
+      "loss": 0.3035,
+      "step": 98500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00028297734094402986,
+      "loss": 0.3039,
+      "step": 99000
+    },
+    {
+      "epoch": 2.31,
+      "eval_loss": 0.2857549488544464,
+      "eval_runtime": 1.4042,
+      "eval_samples_per_second": 1565.268,
+      "eval_steps_per_second": 24.925,
+      "step": 99000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002827512159679005,
+      "loss": 0.3027,
+      "step": 99500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00028252369392494086,
+      "loss": 0.3025,
+      "step": 100000
+    },
+    {
+      "epoch": 2.33,
+      "eval_loss": 0.28487083315849304,
+      "eval_runtime": 1.4056,
+      "eval_samples_per_second": 1563.737,
+      "eval_steps_per_second": 24.9,
+      "step": 100000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002822947773032956,
+      "loss": 0.3022,
+      "step": 100500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002820644686063602,
+      "loss": 0.3023,
+      "step": 101000
+    },
+    {
+      "epoch": 2.35,
+      "eval_loss": 0.2856578528881073,
+      "eval_runtime": 1.4144,
+      "eval_samples_per_second": 1553.982,
+      "eval_steps_per_second": 24.745,
+      "step": 101000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00028183277035275363,
+      "loss": 0.3022,
+      "step": 101500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002815996850762909,
+      "loss": 0.3024,
+      "step": 102000
+    },
+    {
+      "epoch": 2.38,
+      "eval_loss": 0.28789734840393066,
+      "eval_runtime": 1.8256,
+      "eval_samples_per_second": 1203.971,
+      "eval_steps_per_second": 19.172,
+      "step": 102000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00028136521532595515,
+      "loss": 0.3022,
+      "step": 102500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00028112936366587023,
+      "loss": 0.3018,
+      "step": 103000
+    },
+    {
+      "epoch": 2.4,
+      "eval_loss": 0.28655195236206055,
+      "eval_runtime": 1.4347,
+      "eval_samples_per_second": 1532.011,
+      "eval_steps_per_second": 24.395,
+      "step": 103000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00028089213267527184,
+      "loss": 0.3018,
+      "step": 103500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002806535249484803,
+      "loss": 0.3015,
+      "step": 104000
+    },
+    {
+      "epoch": 2.42,
+      "eval_loss": 0.2860538959503174,
+      "eval_runtime": 1.4232,
+      "eval_samples_per_second": 1544.364,
+      "eval_steps_per_second": 24.592,
+      "step": 104000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00028041354309487135,
+      "loss": 0.3013,
+      "step": 104500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002801721897388482,
+      "loss": 0.301,
+      "step": 105000
+    },
+    {
+      "epoch": 2.45,
+      "eval_loss": 0.28530260920524597,
+      "eval_runtime": 1.4494,
+      "eval_samples_per_second": 1516.459,
+      "eval_steps_per_second": 24.147,
+      "step": 105000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002799294675198124,
+      "loss": 0.3018,
+      "step": 105500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00027968537909213524,
+      "loss": 0.3008,
+      "step": 106000
+    },
+    {
+      "epoch": 2.47,
+      "eval_loss": 0.285876989364624,
+      "eval_runtime": 1.4506,
+      "eval_samples_per_second": 1515.222,
+      "eval_steps_per_second": 24.128,
+      "step": 106000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002794399271251287,
+      "loss": 0.3013,
+      "step": 106500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002791931143030162,
+      "loss": 0.3006,
+      "step": 107000
+    },
+    {
+      "epoch": 2.49,
+      "eval_loss": 0.28289294242858887,
+      "eval_runtime": 1.4223,
+      "eval_samples_per_second": 1545.438,
+      "eval_steps_per_second": 24.609,
+      "step": 107000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00027894494332490315,
+      "loss": 0.3001,
+      "step": 107500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0002786954169047476,
+      "loss": 0.3003,
+      "step": 108000
+    },
+    {
+      "epoch": 2.52,
+      "eval_loss": 0.2835444509983063,
+      "eval_runtime": 1.4445,
+      "eval_samples_per_second": 1521.603,
+      "eval_steps_per_second": 24.229,
+      "step": 108000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002784445377713306,
+      "loss": 0.2999,
+      "step": 108500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002781923086682261,
+      "loss": 0.2998,
+      "step": 109000
+    },
+    {
+      "epoch": 2.54,
+      "eval_loss": 0.27993467450141907,
+      "eval_runtime": 1.4177,
+      "eval_samples_per_second": 1550.426,
+      "eval_steps_per_second": 24.688,
+      "step": 109000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0002779387323537711,
+      "loss": 0.2994,
+      "step": 109500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002776838116010356,
+      "loss": 0.3001,
+      "step": 110000
+    },
+    {
+      "epoch": 2.56,
+      "eval_loss": 0.28160494565963745,
+      "eval_runtime": 1.4324,
+      "eval_samples_per_second": 1534.54,
+      "eval_steps_per_second": 24.435,
+      "step": 110000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0002774275491977922,
+      "loss": 0.2993,
+      "step": 110500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0002771699479464853,
+      "loss": 0.2991,
+      "step": 111000
+    },
+    {
+      "epoch": 2.59,
+      "eval_loss": 0.28305351734161377,
+      "eval_runtime": 1.4425,
+      "eval_samples_per_second": 1523.692,
+      "eval_steps_per_second": 24.263,
+      "step": 111000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00027691101066420104,
+      "loss": 0.2993,
+      "step": 111500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002766507401826361,
+      "loss": 0.2987,
+      "step": 112000
+    },
+    {
+      "epoch": 2.61,
+      "eval_loss": 0.2818717956542969,
+      "eval_runtime": 1.4177,
+      "eval_samples_per_second": 1550.356,
+      "eval_steps_per_second": 24.687,
+      "step": 112000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002763891393480666,
+      "loss": 0.2989,
+      "step": 112500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0002761262110213175,
+      "loss": 0.2985,
+      "step": 113000
+    },
+    {
+      "epoch": 2.63,
+      "eval_loss": 0.283016562461853,
+      "eval_runtime": 1.4225,
+      "eval_samples_per_second": 1545.128,
+      "eval_steps_per_second": 24.604,
+      "step": 113000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00027586195807773083,
+      "loss": 0.2989,
+      "step": 113500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00027559638340713435,
+      "loss": 0.2981,
+      "step": 114000
+    },
+    {
+      "epoch": 2.66,
+      "eval_loss": 0.2824258804321289,
+      "eval_runtime": 1.412,
+      "eval_samples_per_second": 1556.637,
+      "eval_steps_per_second": 24.787,
+      "step": 114000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00027532948991381025,
+      "loss": 0.2976,
+      "step": 114500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00027506128051646287,
+      "loss": 0.2978,
+      "step": 115000
+    },
+    {
+      "epoch": 2.68,
+      "eval_loss": 0.2790142595767975,
+      "eval_runtime": 1.436,
+      "eval_samples_per_second": 1530.608,
+      "eval_steps_per_second": 24.373,
+      "step": 115000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00027479175814818733,
+      "loss": 0.2982,
+      "step": 115500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.000274520925756437,
+      "loss": 0.2973,
+      "step": 116000
+    },
+    {
+      "epoch": 2.7,
+      "eval_loss": 0.27925553917884827,
+      "eval_runtime": 1.4557,
+      "eval_samples_per_second": 1509.89,
+      "eval_steps_per_second": 24.043,
+      "step": 116000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00027424878630299157,
+      "loss": 0.2971,
+      "step": 116500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002739753427639244,
+      "loss": 0.2977,
+      "step": 117000
+    },
+    {
+      "epoch": 2.73,
+      "eval_loss": 0.2807333171367645,
+      "eval_runtime": 1.4375,
+      "eval_samples_per_second": 1529.048,
+      "eval_steps_per_second": 24.348,
+      "step": 117000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002737005981295704,
+      "loss": 0.2973,
+      "step": 117500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002734245554044927,
+      "loss": 0.2969,
+      "step": 118000
+    },
+    {
+      "epoch": 2.75,
+      "eval_loss": 0.28166475892066956,
+      "eval_runtime": 1.4453,
+      "eval_samples_per_second": 1520.838,
+      "eval_steps_per_second": 24.217,
+      "step": 118000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0002731472176074504,
+      "loss": 0.2967,
+      "step": 118500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002728685877713653,
+      "loss": 0.2967,
+      "step": 119000
+    },
+    {
+      "epoch": 2.77,
+      "eval_loss": 0.2808319330215454,
+      "eval_runtime": 1.4174,
+      "eval_samples_per_second": 1550.705,
+      "eval_steps_per_second": 24.693,
+      "step": 119000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002725886689432884,
+      "loss": 0.2965,
+      "step": 119500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0002723074641843674,
+      "loss": 0.2965,
+      "step": 120000
+    },
+    {
+      "epoch": 2.8,
+      "eval_loss": 0.27865585684776306,
+      "eval_runtime": 1.4415,
+      "eval_samples_per_second": 1524.849,
+      "eval_steps_per_second": 24.281,
+      "step": 120000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0002720249765698123,
+      "loss": 0.2963,
+      "step": 120500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002717412091888626,
+      "loss": 0.2958,
+      "step": 121000
+    },
+    {
+      "epoch": 2.82,
+      "eval_loss": 0.2785983681678772,
+      "eval_runtime": 1.4123,
+      "eval_samples_per_second": 1556.34,
+      "eval_steps_per_second": 24.782,
+      "step": 121000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00027145616514475274,
+      "loss": 0.2963,
+      "step": 121500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0002711698475546788,
+      "loss": 0.2968,
+      "step": 122000
+    },
+    {
+      "epoch": 2.84,
+      "eval_loss": 0.27918389439582825,
+      "eval_runtime": 1.4238,
+      "eval_samples_per_second": 1543.71,
+      "eval_steps_per_second": 24.581,
+      "step": 122000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00027088225954976407,
+      "loss": 0.2965,
+      "step": 122500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0002705934042750249,
+      "loss": 0.2956,
+      "step": 123000
+    },
+    {
+      "epoch": 2.87,
+      "eval_loss": 0.27785566449165344,
+      "eval_runtime": 1.3994,
+      "eval_samples_per_second": 1570.623,
+      "eval_steps_per_second": 25.01,
+      "step": 123000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00027030328488933625,
+      "loss": 0.2962,
+      "step": 123500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00027001190456539726,
+      "loss": 0.2948,
+      "step": 124000
+    },
+    {
+      "epoch": 2.89,
+      "eval_loss": 0.2769564390182495,
+      "eval_runtime": 1.4165,
+      "eval_samples_per_second": 1551.741,
+      "eval_steps_per_second": 24.709,
+      "step": 124000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002697192664896965,
+      "loss": 0.295,
+      "step": 124500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00026942537386247706,
+      "loss": 0.2952,
+      "step": 125000
+    },
+    {
+      "epoch": 2.91,
+      "eval_loss": 0.27958372235298157,
+      "eval_runtime": 1.3901,
+      "eval_samples_per_second": 1581.168,
+      "eval_steps_per_second": 25.178,
+      "step": 125000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0002691302298977016,
+      "loss": 0.2948,
+      "step": 125500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0002688338378230173,
+      "loss": 0.2943,
+      "step": 126000
+    },
+    {
+      "epoch": 2.94,
+      "eval_loss": 0.2770977318286896,
+      "eval_runtime": 1.4225,
+      "eval_samples_per_second": 1545.165,
+      "eval_steps_per_second": 24.605,
+      "step": 126000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00026853620087972035,
+      "loss": 0.2945,
+      "step": 126500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00026823732232272065,
+      "loss": 0.2945,
+      "step": 127000
+    },
+    {
+      "epoch": 2.96,
+      "eval_loss": 0.27677854895591736,
+      "eval_runtime": 1.4513,
+      "eval_samples_per_second": 1514.548,
+      "eval_steps_per_second": 24.117,
+      "step": 127000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0002679372054205063,
+      "loss": 0.2949,
+      "step": 127500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0002676358534551076,
+      "loss": 0.2945,
+      "step": 128000
+    },
+    {
+      "epoch": 2.98,
+      "eval_loss": 0.27655109763145447,
+      "eval_runtime": 1.4264,
+      "eval_samples_per_second": 1540.932,
+      "eval_steps_per_second": 24.537,
+      "step": 128000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00026733326972206133,
+      "loss": 0.2942,
+      "step": 128500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0002670294575303748,
+      "loss": 0.2944,
+      "step": 129000
+    },
+    {
+      "epoch": 3.01,
+      "eval_loss": 0.2760108709335327,
+      "eval_runtime": 1.4342,
+      "eval_samples_per_second": 1532.608,
+      "eval_steps_per_second": 24.405,
+      "step": 129000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0002667244202024894,
+      "loss": 0.2942,
+      "step": 129500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00026641816107424453,
+      "loss": 0.2944,
+      "step": 130000
+    },
+    {
+      "epoch": 3.03,
+      "eval_loss": 0.2757641077041626,
+      "eval_runtime": 1.4126,
+      "eval_samples_per_second": 1555.976,
+      "eval_steps_per_second": 24.777,
+      "step": 130000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0002661106834948409,
+      "loss": 0.2939,
+      "step": 130500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0002658019908268041,
+      "loss": 0.2937,
+      "step": 131000
+    },
+    {
+      "epoch": 3.05,
+      "eval_loss": 0.27508604526519775,
+      "eval_runtime": 1.4205,
+      "eval_samples_per_second": 1547.302,
+      "eval_steps_per_second": 24.639,
+      "step": 131000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00026549208644594766,
+      "loss": 0.294,
+      "step": 131500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00026518097374133627,
+      "loss": 0.2935,
+      "step": 132000
+    },
+    {
+      "epoch": 3.08,
+      "eval_loss": 0.2736782431602478,
+      "eval_runtime": 1.4305,
+      "eval_samples_per_second": 1536.563,
+      "eval_steps_per_second": 24.468,
+      "step": 132000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00026486865611524853,
+      "loss": 0.2932,
+      "step": 132500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00026455513698314003,
+      "loss": 0.2931,
+      "step": 133000
+    },
+    {
+      "epoch": 3.1,
+      "eval_loss": 0.27545085549354553,
+      "eval_runtime": 1.4136,
+      "eval_samples_per_second": 1554.95,
+      "eval_steps_per_second": 24.76,
+      "step": 133000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0002642404197736058,
+      "loss": 0.2931,
+      "step": 133500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0002639245079283428,
+      "loss": 0.2936,
+      "step": 134000
+    },
+    {
+      "epoch": 3.12,
+      "eval_loss": 0.27676478028297424,
+      "eval_runtime": 1.4314,
+      "eval_samples_per_second": 1535.602,
+      "eval_steps_per_second": 24.452,
+      "step": 134000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00026360740490211234,
+      "loss": 0.2932,
+      "step": 134500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0002632891141627023,
+      "loss": 0.2934,
+      "step": 135000
+    },
+    {
+      "epoch": 3.15,
+      "eval_loss": 0.2775753438472748,
+      "eval_runtime": 1.4316,
+      "eval_samples_per_second": 1535.332,
+      "eval_steps_per_second": 24.448,
+      "step": 135000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00026296963919088923,
+      "loss": 0.2927,
+      "step": 135500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00026264898348040024,
+      "loss": 0.2925,
+      "step": 136000
+    },
+    {
+      "epoch": 3.17,
+      "eval_loss": 0.27160218358039856,
+      "eval_runtime": 1.4163,
+      "eval_samples_per_second": 1551.918,
+      "eval_steps_per_second": 24.712,
+      "step": 136000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0002623271505378748,
+      "loss": 0.2926,
+      "step": 136500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00026200414388282637,
+      "loss": 0.2924,
+      "step": 137000
+    },
+    {
+      "epoch": 3.19,
+      "eval_loss": 0.2737308442592621,
+      "eval_runtime": 1.4266,
+      "eval_samples_per_second": 1540.705,
+      "eval_steps_per_second": 24.534,
+      "step": 137000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00026167996704760406,
+      "loss": 0.2929,
+      "step": 137500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00026135462357735375,
+      "loss": 0.2931,
+      "step": 138000
+    },
+    {
+      "epoch": 3.22,
+      "eval_loss": 0.2752697467803955,
+      "eval_runtime": 1.4495,
+      "eval_samples_per_second": 1516.388,
+      "eval_steps_per_second": 24.146,
+      "step": 138000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002610281170299795,
+      "loss": 0.292,
+      "step": 138500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00026070045097610465,
+      "loss": 0.2917,
+      "step": 139000
+    },
+    {
+      "epoch": 3.24,
+      "eval_loss": 0.2744641602039337,
+      "eval_runtime": 1.4101,
+      "eval_samples_per_second": 1558.726,
+      "eval_steps_per_second": 24.82,
+      "step": 139000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0002603716289990326,
+      "loss": 0.2923,
+      "step": 139500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00026004165469470787,
+      "loss": 0.2918,
+      "step": 140000
+    },
+    {
+      "epoch": 3.26,
+      "eval_loss": 0.2773716449737549,
+      "eval_runtime": 1.4122,
+      "eval_samples_per_second": 1556.477,
+      "eval_steps_per_second": 24.785,
+      "step": 140000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0002597105316716766,
+      "loss": 0.2921,
+      "step": 140500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.000259378263551047,
+      "loss": 0.2914,
+      "step": 141000
+    },
+    {
+      "epoch": 3.29,
+      "eval_loss": 0.2755984961986542,
+      "eval_runtime": 1.4384,
+      "eval_samples_per_second": 1528.099,
+      "eval_steps_per_second": 24.333,
+      "step": 141000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0002590448539664501,
+      "loss": 0.2915,
+      "step": 141500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00025871030656399966,
+      "loss": 0.2913,
+      "step": 142000
+    },
+    {
+      "epoch": 3.31,
+      "eval_loss": 0.2742982804775238,
+      "eval_runtime": 1.3825,
+      "eval_samples_per_second": 1589.856,
+      "eval_steps_per_second": 25.316,
+      "step": 142000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00025837462500225255,
+      "loss": 0.2912,
+      "step": 142500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0002580378129521685,
+      "loss": 0.2909,
+      "step": 143000
+    },
+    {
+      "epoch": 3.33,
+      "eval_loss": 0.2741672694683075,
+      "eval_runtime": 1.3884,
+      "eval_samples_per_second": 1583.139,
+      "eval_steps_per_second": 25.209,
+      "step": 143000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0002576998740970701,
+      "loss": 0.2907,
+      "step": 143500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00025736081213260253,
+      "loss": 0.2911,
+      "step": 144000
+    },
+    {
+      "epoch": 3.36,
+      "eval_loss": 0.27347105741500854,
+      "eval_runtime": 1.4089,
+      "eval_samples_per_second": 1560.053,
+      "eval_steps_per_second": 24.842,
+      "step": 144000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0002570206307666931,
+      "loss": 0.2905,
+      "step": 144500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0002566793337195108,
+      "loss": 0.2904,
+      "step": 145000
+    },
+    {
+      "epoch": 3.38,
+      "eval_loss": 0.27271768450737,
+      "eval_runtime": 1.4129,
+      "eval_samples_per_second": 1555.627,
+      "eval_steps_per_second": 24.771,
+      "step": 145000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0002563369247234254,
+      "loss": 0.29,
+      "step": 145500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0002559934075229669,
+      "loss": 0.2903,
+      "step": 146000
+    },
+    {
+      "epoch": 3.4,
+      "eval_loss": 0.27478328347206116,
+      "eval_runtime": 1.4292,
+      "eval_samples_per_second": 1537.886,
+      "eval_steps_per_second": 24.489,
+      "step": 146000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0002556487858747843,
+      "loss": 0.2904,
+      "step": 146500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00025530306354760464,
+      "loss": 0.2899,
+      "step": 147000
+    },
+    {
+      "epoch": 3.43,
+      "eval_loss": 0.2697647213935852,
+      "eval_runtime": 1.4585,
+      "eval_samples_per_second": 1506.981,
+      "eval_steps_per_second": 23.997,
+      "step": 147000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.000254956244322192,
+      "loss": 0.2899,
+      "step": 147500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00025460833199130595,
+      "loss": 0.2896,
+      "step": 148000
+    },
+    {
+      "epoch": 3.45,
+      "eval_loss": 0.27176716923713684,
+      "eval_runtime": 1.4122,
+      "eval_samples_per_second": 1556.39,
+      "eval_steps_per_second": 24.783,
+      "step": 148000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00025425933035965983,
+      "loss": 0.2895,
+      "step": 148500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00025390924324387965,
+      "loss": 0.2895,
+      "step": 149000
+    },
+    {
+      "epoch": 3.47,
+      "eval_loss": 0.27247366309165955,
+      "eval_runtime": 1.4208,
+      "eval_samples_per_second": 1547.057,
+      "eval_steps_per_second": 24.635,
+      "step": 149000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0002535580744724621,
+      "loss": 0.2892,
+      "step": 149500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00025320582788573246,
+      "loss": 0.2892,
+      "step": 150000
+    },
+    {
+      "epoch": 3.49,
+      "eval_loss": 0.2716062366962433,
+      "eval_runtime": 1.4082,
+      "eval_samples_per_second": 1560.897,
+      "eval_steps_per_second": 24.855,
+      "step": 150000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.000252852507335803,
+      "loss": 0.2894,
+      "step": 150500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0002524981166865307,
+      "loss": 0.2889,
+      "step": 151000
+    },
+    {
+      "epoch": 3.52,
+      "eval_loss": 0.2722132205963135,
+      "eval_runtime": 1.426,
+      "eval_samples_per_second": 1541.329,
+      "eval_steps_per_second": 24.543,
+      "step": 151000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00025214265981347487,
+      "loss": 0.2889,
+      "step": 151500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.000251786140603855,
+      "loss": 0.289,
+      "step": 152000
+    },
+    {
+      "epoch": 3.54,
+      "eval_loss": 0.27176567912101746,
+      "eval_runtime": 1.4194,
+      "eval_samples_per_second": 1548.585,
+      "eval_steps_per_second": 24.659,
+      "step": 152000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00025142856295650795,
+      "loss": 0.2898,
+      "step": 152500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0002510699307818457,
+      "loss": 0.2888,
+      "step": 153000
+    },
+    {
+      "epoch": 3.56,
+      "eval_loss": 0.2718333601951599,
+      "eval_runtime": 1.4384,
+      "eval_samples_per_second": 1528.088,
+      "eval_steps_per_second": 24.333,
+      "step": 153000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00025071024800181214,
+      "loss": 0.2889,
+      "step": 153500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0002503495185498405,
+      "loss": 0.2887,
+      "step": 154000
+    },
+    {
+      "epoch": 3.59,
+      "eval_loss": 0.2725473940372467,
+      "eval_runtime": 1.4286,
+      "eval_samples_per_second": 1538.528,
+      "eval_steps_per_second": 24.499,
+      "step": 154000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00024998774637081044,
+      "loss": 0.2896,
+      "step": 154500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00024962493542100443,
+      "loss": 0.2888,
+      "step": 155000
+    },
+    {
+      "epoch": 3.61,
+      "eval_loss": 0.27104923129081726,
+      "eval_runtime": 1.4479,
+      "eval_samples_per_second": 1518.017,
+      "eval_steps_per_second": 24.172,
+      "step": 155000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0002492610896680649,
+      "loss": 0.2879,
+      "step": 155500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00024889621309095067,
+      "loss": 0.289,
+      "step": 156000
+    },
+    {
+      "epoch": 3.63,
+      "eval_loss": 0.2686387598514557,
+      "eval_runtime": 1.428,
+      "eval_samples_per_second": 1539.209,
+      "eval_steps_per_second": 24.51,
+      "step": 156000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00024853030967989366,
+      "loss": 0.288,
+      "step": 156500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00024816338343635485,
+      "loss": 0.2873,
+      "step": 157000
+    },
+    {
+      "epoch": 3.66,
+      "eval_loss": 0.2692621946334839,
+      "eval_runtime": 1.4361,
+      "eval_samples_per_second": 1530.504,
+      "eval_steps_per_second": 24.371,
+      "step": 157000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0002477954383729809,
+      "loss": 0.288,
+      "step": 157500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00024742647851355997,
+      "loss": 0.2876,
+      "step": 158000
+    },
+    {
+      "epoch": 3.68,
+      "eval_loss": 0.2716818153858185,
+      "eval_runtime": 1.4202,
+      "eval_samples_per_second": 1547.632,
+      "eval_steps_per_second": 24.644,
+      "step": 158000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0002470565078929781,
+      "loss": 0.2872,
+      "step": 158500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00024668553055717465,
+      "loss": 0.2873,
+      "step": 159000
+    },
+    {
+      "epoch": 3.7,
+      "eval_loss": 0.2711220681667328,
+      "eval_runtime": 1.4791,
+      "eval_samples_per_second": 1486.051,
+      "eval_steps_per_second": 23.663,
+      "step": 159000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0002463135505630984,
+      "loss": 0.2892,
+      "step": 159500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00024594057197866283,
+      "loss": 0.2883,
+      "step": 160000
+    },
+    {
+      "epoch": 3.73,
+      "eval_loss": 0.2702338993549347,
+      "eval_runtime": 1.4449,
+      "eval_samples_per_second": 1521.168,
+      "eval_steps_per_second": 24.222,
+      "step": 160000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0002455665988827021,
+      "loss": 0.2869,
+      "step": 160500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0002451916353649261,
+      "loss": 0.287,
+      "step": 161000
+    },
+    {
+      "epoch": 3.75,
+      "eval_loss": 0.27171802520751953,
+      "eval_runtime": 1.4254,
+      "eval_samples_per_second": 1542.077,
+      "eval_steps_per_second": 24.555,
+      "step": 161000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00024481568552587566,
+      "loss": 0.2871,
+      "step": 161500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002444387534768781,
+      "loss": 0.2868,
+      "step": 162000
+    },
+    {
+      "epoch": 3.77,
+      "eval_loss": 0.26901960372924805,
+      "eval_runtime": 1.4243,
+      "eval_samples_per_second": 1543.244,
+      "eval_steps_per_second": 24.574,
+      "step": 162000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002440608433400018,
+      "loss": 0.2869,
+      "step": 162500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00024368195924801158,
+      "loss": 0.2869,
+      "step": 163000
+    },
+    {
+      "epoch": 3.8,
+      "eval_loss": 0.26863837242126465,
+      "eval_runtime": 1.4601,
+      "eval_samples_per_second": 1505.401,
+      "eval_steps_per_second": 23.971,
+      "step": 163000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00024330210534432314,
+      "loss": 0.2862,
+      "step": 163500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002429212857829579,
+      "loss": 0.286,
+      "step": 164000
+    },
+    {
+      "epoch": 3.82,
+      "eval_loss": 0.26618143916130066,
+      "eval_runtime": 1.4166,
+      "eval_samples_per_second": 1551.57,
+      "eval_steps_per_second": 24.707,
+      "step": 164000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00024253950472849758,
+      "loss": 0.2863,
+      "step": 164500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002421567663560386,
+      "loss": 0.286,
+      "step": 165000
+    },
+    {
+      "epoch": 3.84,
+      "eval_loss": 0.26967230439186096,
+      "eval_runtime": 1.4363,
+      "eval_samples_per_second": 1530.302,
+      "eval_steps_per_second": 24.368,
+      "step": 165000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00024177307485114653,
+      "loss": 0.2864,
+      "step": 165500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002413884344098101,
+      "loss": 0.2859,
+      "step": 166000
+    },
+    {
+      "epoch": 3.87,
+      "eval_loss": 0.270435631275177,
+      "eval_runtime": 1.4339,
+      "eval_samples_per_second": 1532.847,
+      "eval_steps_per_second": 24.408,
+      "step": 166000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00024100284923839568,
+      "loss": 0.2855,
+      "step": 166500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002406163235536008,
+      "loss": 0.2858,
+      "step": 167000
+    },
+    {
+      "epoch": 3.89,
+      "eval_loss": 0.26777446269989014,
+      "eval_runtime": 1.4287,
+      "eval_samples_per_second": 1538.493,
+      "eval_steps_per_second": 24.498,
+      "step": 167000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024022886158240857,
+      "loss": 0.2856,
+      "step": 167500
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0002398404675620409,
+      "loss": 0.2854,
+      "step": 168000
+    },
+    {
+      "epoch": 3.91,
+      "eval_loss": 0.2669973075389862,
+      "eval_runtime": 1.4303,
+      "eval_samples_per_second": 1536.724,
+      "eval_steps_per_second": 24.47,
+      "step": 168000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0002394511457399126,
+      "loss": 0.2858,
+      "step": 168500
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00023906090037358478,
+      "loss": 0.2853,
+      "step": 169000
+    },
+    {
+      "epoch": 3.94,
+      "eval_loss": 0.2690945863723755,
+      "eval_runtime": 1.4107,
+      "eval_samples_per_second": 1558.141,
+      "eval_steps_per_second": 24.811,
+      "step": 169000
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0002386697357307182,
+      "loss": 0.285,
+      "step": 169500
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00023827765608902676,
+      "loss": 0.2849,
+      "step": 170000
+    },
+    {
+      "epoch": 3.96,
+      "eval_loss": 0.2683064937591553,
+      "eval_runtime": 1.4268,
+      "eval_samples_per_second": 1540.527,
+      "eval_steps_per_second": 24.531,
+      "step": 170000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0002378846657362306,
+      "loss": 0.285,
+      "step": 170500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00023749076897000928,
+      "loss": 0.2858,
+      "step": 171000
+    },
+    {
+      "epoch": 3.98,
+      "eval_loss": 0.26628705859184265,
+      "eval_runtime": 1.4334,
+      "eval_samples_per_second": 1533.445,
+      "eval_steps_per_second": 24.418,
+      "step": 171000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00023709597009795465,
+      "loss": 0.2847,
+      "step": 171500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.000236700273437524,
+      "loss": 0.2846,
+      "step": 172000
+    },
+    {
+      "epoch": 4.01,
+      "eval_loss": 0.2714168429374695,
+      "eval_runtime": 1.439,
+      "eval_samples_per_second": 1527.408,
+      "eval_steps_per_second": 24.322,
+      "step": 172000
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0002363036833159925,
+      "loss": 0.2843,
+      "step": 172500
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00023590620407040633,
+      "loss": 0.2846,
+      "step": 173000
+    },
+    {
+      "epoch": 4.03,
+      "eval_loss": 0.2668880820274353,
+      "eval_runtime": 1.4491,
+      "eval_samples_per_second": 1516.791,
+      "eval_steps_per_second": 24.153,
+      "step": 173000
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00023550784004753471,
+      "loss": 0.2843,
+      "step": 173500
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0002351085956038229,
+      "loss": 0.2844,
+      "step": 174000
+    },
+    {
+      "epoch": 4.05,
+      "eval_loss": 0.2690703272819519,
+      "eval_runtime": 1.4333,
+      "eval_samples_per_second": 1533.56,
+      "eval_steps_per_second": 24.42,
+      "step": 174000
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00023470847510534407,
+      "loss": 0.2837,
+      "step": 174500
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00023430748292775188,
+      "loss": 0.2839,
+      "step": 175000
+    },
+    {
+      "epoch": 4.08,
+      "eval_loss": 0.2685079574584961,
+      "eval_runtime": 1.4345,
+      "eval_samples_per_second": 1532.24,
+      "eval_steps_per_second": 24.399,
+      "step": 175000
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0002339056234562326,
+      "loss": 0.2842,
+      "step": 175500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00023350290108545694,
+      "loss": 0.284,
+      "step": 176000
+    },
+    {
+      "epoch": 4.1,
+      "eval_loss": 0.26691344380378723,
+      "eval_runtime": 1.4222,
+      "eval_samples_per_second": 1545.466,
+      "eval_steps_per_second": 24.609,
+      "step": 176000
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00023309932021953238,
+      "loss": 0.2837,
+      "step": 176500
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00023269488527195446,
+      "loss": 0.2835,
+      "step": 177000
+    },
+    {
+      "epoch": 4.12,
+      "eval_loss": 0.2659927308559418,
+      "eval_runtime": 1.4298,
+      "eval_samples_per_second": 1537.321,
+      "eval_steps_per_second": 24.48,
+      "step": 177000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00023228960066555907,
+      "loss": 0.2834,
+      "step": 177500
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00023188347083247365,
+      "loss": 0.2833,
+      "step": 178000
+    },
+    {
+      "epoch": 4.15,
+      "eval_loss": 0.2681402266025543,
+      "eval_runtime": 1.4006,
+      "eval_samples_per_second": 1569.372,
+      "eval_steps_per_second": 24.99,
+      "step": 178000
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00023147650021406905,
+      "loss": 0.2834,
+      "step": 178500
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00023106869326091075,
+      "loss": 0.2833,
+      "step": 179000
+    },
+    {
+      "epoch": 4.17,
+      "eval_loss": 0.2678356468677521,
+      "eval_runtime": 1.4225,
+      "eval_samples_per_second": 1545.199,
+      "eval_steps_per_second": 24.605,
+      "step": 179000
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00023066005443271017,
+      "loss": 0.2838,
+      "step": 179500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00023025058819827618,
+      "loss": 0.2839,
+      "step": 180000
+    },
+    {
+      "epoch": 4.19,
+      "eval_loss": 0.2670538127422333,
+      "eval_runtime": 1.4128,
+      "eval_samples_per_second": 1555.751,
+      "eval_steps_per_second": 24.773,
+      "step": 180000
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00022984029903546578,
+      "loss": 0.2835,
+      "step": 180500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00022942919143113572,
+      "loss": 0.2832,
+      "step": 181000
+    },
+    {
+      "epoch": 4.22,
+      "eval_loss": 0.26567918062210083,
+      "eval_runtime": 1.4421,
+      "eval_samples_per_second": 1524.212,
+      "eval_steps_per_second": 24.271,
+      "step": 181000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0002290172698810927,
+      "loss": 0.2832,
+      "step": 181500
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00022860453889004493,
+      "loss": 0.2828,
+      "step": 182000
+    },
+    {
+      "epoch": 4.24,
+      "eval_loss": 0.2642976641654968,
+      "eval_runtime": 1.4457,
+      "eval_samples_per_second": 1520.406,
+      "eval_steps_per_second": 24.21,
+      "step": 182000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00022819100297155235,
+      "loss": 0.2825,
+      "step": 182500
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0002277766666479774,
+      "loss": 0.2825,
+      "step": 183000
+    },
+    {
+      "epoch": 4.26,
+      "eval_loss": 0.26654452085494995,
+      "eval_runtime": 1.5184,
+      "eval_samples_per_second": 1447.589,
+      "eval_steps_per_second": 23.051,
+      "step": 183000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00022736153445043595,
+      "loss": 0.2833,
+      "step": 183500
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00022694561091874706,
+      "loss": 0.2854,
+      "step": 184000
+    },
+    {
+      "epoch": 4.29,
+      "eval_loss": 0.2664638161659241,
+      "eval_runtime": 1.3999,
+      "eval_samples_per_second": 1570.099,
+      "eval_steps_per_second": 25.002,
+      "step": 184000
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00022652890060138387,
+      "loss": 0.2833,
+      "step": 184500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00022611140805542366,
+      "loss": 0.2824,
+      "step": 185000
+    },
+    {
+      "epoch": 4.31,
+      "eval_loss": 0.26191577315330505,
+      "eval_runtime": 1.367,
+      "eval_samples_per_second": 1607.943,
+      "eval_steps_per_second": 25.604,
+      "step": 185000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00022569313784649798,
+      "loss": 0.2819,
+      "step": 185500
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0002252740945487429,
+      "loss": 0.2819,
+      "step": 186000
+    },
+    {
+      "epoch": 4.33,
+      "eval_loss": 0.26568353176116943,
+      "eval_runtime": 1.413,
+      "eval_samples_per_second": 1555.568,
+      "eval_steps_per_second": 24.77,
+      "step": 186000
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00022485428274474867,
+      "loss": 0.2817,
+      "step": 186500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00022443370702551,
+      "loss": 0.2824,
+      "step": 187000
+    },
+    {
+      "epoch": 4.36,
+      "eval_loss": 0.26549944281578064,
+      "eval_runtime": 1.4329,
+      "eval_samples_per_second": 1533.942,
+      "eval_steps_per_second": 24.426,
+      "step": 187000
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00022401237199037565,
+      "loss": 0.2822,
+      "step": 187500
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0002235902822469979,
+      "loss": 0.2815,
+      "step": 188000
+    },
+    {
+      "epoch": 4.38,
+      "eval_loss": 0.2621370553970337,
+      "eval_runtime": 1.4638,
+      "eval_samples_per_second": 1501.617,
+      "eval_steps_per_second": 23.911,
+      "step": 188000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00022316744241128268,
+      "loss": 0.2815,
+      "step": 188500
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00022274385710733855,
+      "loss": 0.2814,
+      "step": 189000
+    },
+    {
+      "epoch": 4.4,
+      "eval_loss": 0.2638288140296936,
+      "eval_runtime": 1.4301,
+      "eval_samples_per_second": 1536.915,
+      "eval_steps_per_second": 24.473,
+      "step": 189000
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00022231953096742672,
+      "loss": 0.2826,
+      "step": 189500
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00022189446863190974,
+      "loss": 0.2821,
+      "step": 190000
+    },
+    {
+      "epoch": 4.43,
+      "eval_loss": 0.26497727632522583,
+      "eval_runtime": 1.4514,
+      "eval_samples_per_second": 1514.384,
+      "eval_steps_per_second": 24.114,
+      "step": 190000
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00022146867474920118,
+      "loss": 0.2828,
+      "step": 190500
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00022104215397571484,
+      "loss": 0.281,
+      "step": 191000
+    },
+    {
+      "epoch": 4.45,
+      "eval_loss": 0.2647937834262848,
+      "eval_runtime": 1.4216,
+      "eval_samples_per_second": 1546.121,
+      "eval_steps_per_second": 24.62,
+      "step": 191000
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0002206149109758135,
+      "loss": 0.2816,
+      "step": 191500
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00022018695042175818,
+      "loss": 0.2812,
+      "step": 192000
+    },
+    {
+      "epoch": 4.47,
+      "eval_loss": 0.2635156512260437,
+      "eval_runtime": 1.4081,
+      "eval_samples_per_second": 1560.959,
+      "eval_steps_per_second": 24.856,
+      "step": 192000
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00021975827699365693,
+      "loss": 0.2807,
+      "step": 192500
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00021932889537941365,
+      "loss": 0.2809,
+      "step": 193000
+    },
+    {
+      "epoch": 4.5,
+      "eval_loss": 0.2628609240055084,
+      "eval_runtime": 1.4036,
+      "eval_samples_per_second": 1565.967,
+      "eval_steps_per_second": 24.936,
+      "step": 193000
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0002188988102746769,
+      "loss": 0.2806,
+      "step": 193500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0002184680263827885,
+      "loss": 0.2809,
+      "step": 194000
+    },
+    {
+      "epoch": 4.52,
+      "eval_loss": 0.26426324248313904,
+      "eval_runtime": 1.4613,
+      "eval_samples_per_second": 1504.163,
+      "eval_steps_per_second": 23.952,
+      "step": 194000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00021803654841473204,
+      "loss": 0.2805,
+      "step": 194500
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00021760438108908142,
+      "loss": 0.2805,
+      "step": 195000
+    },
+    {
+      "epoch": 4.54,
+      "eval_loss": 0.2624601125717163,
+      "eval_runtime": 1.4471,
+      "eval_samples_per_second": 1518.947,
+      "eval_steps_per_second": 24.187,
+      "step": 195000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0002171715291319494,
+      "loss": 0.2809,
+      "step": 195500
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0002167379972769355,
+      "loss": 0.2806,
+      "step": 196000
+    },
+    {
+      "epoch": 4.57,
+      "eval_loss": 0.26392462849617004,
+      "eval_runtime": 1.3978,
+      "eval_samples_per_second": 1572.46,
+      "eval_steps_per_second": 25.039,
+      "step": 196000
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0002163037902650747,
+      "loss": 0.2801,
+      "step": 196500
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0002158689128447853,
+      "loss": 0.2805,
+      "step": 197000
+    },
+    {
+      "epoch": 4.59,
+      "eval_loss": 0.262777715921402,
+      "eval_runtime": 1.4828,
+      "eval_samples_per_second": 1482.293,
+      "eval_steps_per_second": 23.603,
+      "step": 197000
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00021543336977181704,
+      "loss": 0.2805,
+      "step": 197500
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00021499716580919933,
+      "loss": 0.2799,
+      "step": 198000
+    },
+    {
+      "epoch": 4.61,
+      "eval_loss": 0.26294538378715515,
+      "eval_runtime": 1.4108,
+      "eval_samples_per_second": 1557.962,
+      "eval_steps_per_second": 24.808,
+      "step": 198000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00021456030572718866,
+      "loss": 0.2797,
+      "step": 198500
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.000214122794303217,
+      "loss": 0.2802,
+      "step": 199000
+    },
+    {
+      "epoch": 4.64,
+      "eval_loss": 0.2624069154262543,
+      "eval_runtime": 1.4063,
+      "eval_samples_per_second": 1562.932,
+      "eval_steps_per_second": 24.887,
+      "step": 199000
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00021368463632183912,
+      "loss": 0.2799,
+      "step": 199500
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00021324583657468055,
+      "loss": 0.2799,
+      "step": 200000
+    },
+    {
+      "epoch": 4.66,
+      "eval_loss": 0.2646183967590332,
+      "eval_runtime": 1.4232,
+      "eval_samples_per_second": 1544.399,
+      "eval_steps_per_second": 24.592,
+      "step": 200000
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.000212806399860385,
+      "loss": 0.2797,
+      "step": 200500
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00021236633098456196,
+      "loss": 0.2796,
+      "step": 201000
+    },
+    {
+      "epoch": 4.68,
+      "eval_loss": 0.26055699586868286,
+      "eval_runtime": 1.4119,
+      "eval_samples_per_second": 1556.768,
+      "eval_steps_per_second": 24.789,
+      "step": 201000
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0002119256347597342,
+      "loss": 0.2795,
+      "step": 201500
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.000211484316005285,
+      "loss": 0.2796,
+      "step": 202000
+    },
+    {
+      "epoch": 4.71,
+      "eval_loss": 0.264119952917099,
+      "eval_runtime": 1.437,
+      "eval_samples_per_second": 1529.564,
+      "eval_steps_per_second": 24.356,
+      "step": 202000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00021104237954740554,
+      "loss": 0.2796,
+      "step": 202500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00021059983021904215,
+      "loss": 0.2795,
+      "step": 203000
+    },
+    {
+      "epoch": 4.73,
+      "eval_loss": 0.2626611590385437,
+      "eval_runtime": 1.4477,
+      "eval_samples_per_second": 1518.231,
+      "eval_steps_per_second": 24.176,
+      "step": 203000
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00021015667285984336,
+      "loss": 0.2803,
+      "step": 203500
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00020971291231610707,
+      "loss": 0.2793,
+      "step": 204000
+    },
+    {
+      "epoch": 4.75,
+      "eval_loss": 0.2611912786960602,
+      "eval_runtime": 1.4639,
+      "eval_samples_per_second": 1501.46,
+      "eval_steps_per_second": 23.909,
+      "step": 204000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0002092685534407274,
+      "loss": 0.2792,
+      "step": 204500
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00020882360109314197,
+      "loss": 0.2789,
+      "step": 205000
+    },
+    {
+      "epoch": 4.78,
+      "eval_loss": 0.2591469883918762,
+      "eval_runtime": 1.4679,
+      "eval_samples_per_second": 1497.345,
+      "eval_steps_per_second": 23.843,
+      "step": 205000
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0002083780601392783,
+      "loss": 0.2788,
+      "step": 205500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0002079319354515008,
+      "loss": 0.2789,
+      "step": 206000
+    },
+    {
+      "epoch": 4.8,
+      "eval_loss": 0.26023101806640625,
+      "eval_runtime": 1.4041,
+      "eval_samples_per_second": 1565.36,
+      "eval_steps_per_second": 24.926,
+      "step": 206000
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00020748523190855772,
+      "loss": 0.2788,
+      "step": 206500
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0002070379543955273,
+      "loss": 0.2789,
+      "step": 207000
+    },
+    {
+      "epoch": 4.82,
+      "eval_loss": 0.2611112892627716,
+      "eval_runtime": 1.4444,
+      "eval_samples_per_second": 1521.751,
+      "eval_steps_per_second": 24.232,
+      "step": 207000
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00020659010780376487,
+      "loss": 0.2785,
+      "step": 207500
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00020614169703084896,
+      "loss": 0.2787,
+      "step": 208000
+    },
+    {
+      "epoch": 4.85,
+      "eval_loss": 0.2649895250797272,
+      "eval_runtime": 1.4305,
+      "eval_samples_per_second": 1536.472,
+      "eval_steps_per_second": 24.466,
+      "step": 208000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0002056927269805279,
+      "loss": 0.2782,
+      "step": 208500
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00020524320256266635,
+      "loss": 0.2784,
+      "step": 209000
+    },
+    {
+      "epoch": 4.87,
+      "eval_loss": 0.2615431845188141,
+      "eval_runtime": 1.4232,
+      "eval_samples_per_second": 1544.408,
+      "eval_steps_per_second": 24.592,
+      "step": 209000
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0002047931286931912,
+      "loss": 0.2785,
+      "step": 209500
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00020434251029403824,
+      "loss": 0.2782,
+      "step": 210000
+    },
+    {
+      "epoch": 4.89,
+      "eval_loss": 0.26237037777900696,
+      "eval_runtime": 1.4051,
+      "eval_samples_per_second": 1564.263,
+      "eval_steps_per_second": 24.909,
+      "step": 210000
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00020389135229309803,
+      "loss": 0.2781,
+      "step": 210500
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00020343965962416229,
+      "loss": 0.2783,
+      "step": 211000
+    },
+    {
+      "epoch": 4.92,
+      "eval_loss": 0.26213955879211426,
+      "eval_runtime": 1.4602,
+      "eval_samples_per_second": 1505.253,
+      "eval_steps_per_second": 23.969,
+      "step": 211000
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00020298743722686958,
+      "loss": 0.2778,
+      "step": 211500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0002025346900466516,
+      "loss": 0.2781,
+      "step": 212000
+    },
+    {
+      "epoch": 4.94,
+      "eval_loss": 0.2614113390445709,
+      "eval_runtime": 1.4246,
+      "eval_samples_per_second": 1542.853,
+      "eval_steps_per_second": 24.568,
+      "step": 212000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0002020814230346791,
+      "loss": 0.2778,
+      "step": 212500
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00020162764114780733,
+      "loss": 0.2779,
+      "step": 213000
+    },
+    {
+      "epoch": 4.96,
+      "eval_loss": 0.2597576975822449,
+      "eval_runtime": 1.4338,
+      "eval_samples_per_second": 1532.976,
+      "eval_steps_per_second": 24.41,
+      "step": 213000
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0002011733493485224,
+      "loss": 0.2778,
+      "step": 213500
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00020071855260488664,
+      "loss": 0.2776,
+      "step": 214000
+    },
+    {
+      "epoch": 4.99,
+      "eval_loss": 0.2594500780105591,
+      "eval_runtime": 1.4614,
+      "eval_samples_per_second": 1504.034,
+      "eval_steps_per_second": 23.95,
+      "step": 214000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0002002632558904843,
+      "loss": 0.2777,
+      "step": 214500
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00019980746418436736,
+      "loss": 0.2774,
+      "step": 215000
+    },
+    {
+      "epoch": 5.01,
+      "eval_loss": 0.2608456611633301,
+      "eval_runtime": 1.4098,
+      "eval_samples_per_second": 1559.139,
+      "eval_steps_per_second": 24.827,
+      "step": 215000
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00019935118247100088,
+      "loss": 0.2777,
+      "step": 215500
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00019889441574020864,
+      "loss": 0.2782,
+      "step": 216000
+    },
+    {
+      "epoch": 5.03,
+      "eval_loss": 0.26146388053894043,
+      "eval_runtime": 1.4335,
+      "eval_samples_per_second": 1533.337,
+      "eval_steps_per_second": 24.416,
+      "step": 216000
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0001984371689871183,
+      "loss": 0.2775,
+      "step": 216500
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00019797944721210725,
+      "loss": 0.2779,
+      "step": 217000
+    },
+    {
+      "epoch": 5.06,
+      "eval_loss": 0.26257631182670593,
+      "eval_runtime": 1.4244,
+      "eval_samples_per_second": 1543.156,
+      "eval_steps_per_second": 24.573,
+      "step": 217000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00019752125542074736,
+      "loss": 0.2785,
+      "step": 217500
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00019706259862375074,
+      "loss": 0.2768,
+      "step": 218000
+    },
+    {
+      "epoch": 5.08,
+      "eval_loss": 0.25935330986976624,
+      "eval_runtime": 1.4288,
+      "eval_samples_per_second": 1538.396,
+      "eval_steps_per_second": 24.497,
+      "step": 218000
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00019660348183691453,
+      "loss": 0.277,
+      "step": 218500
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0001961439100810664,
+      "loss": 0.2774,
+      "step": 219000
+    },
+    {
+      "epoch": 5.1,
+      "eval_loss": 0.2606256604194641,
+      "eval_runtime": 1.4284,
+      "eval_samples_per_second": 1538.738,
+      "eval_steps_per_second": 24.502,
+      "step": 219000
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00019568388838200952,
+      "loss": 0.277,
+      "step": 219500
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00019522342177046744,
+      "loss": 0.2767,
+      "step": 220000
+    },
+    {
+      "epoch": 5.13,
+      "eval_loss": 0.2605719566345215,
+      "eval_runtime": 1.4163,
+      "eval_samples_per_second": 1551.973,
+      "eval_steps_per_second": 24.713,
+      "step": 220000
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00019476251528202922,
+      "loss": 0.2773,
+      "step": 220500
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0001943011739570944,
+      "loss": 0.2763,
+      "step": 221000
+    },
+    {
+      "epoch": 5.15,
+      "eval_loss": 0.26054272055625916,
+      "eval_runtime": 1.4233,
+      "eval_samples_per_second": 1544.255,
+      "eval_steps_per_second": 24.59,
+      "step": 221000
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00019383940284081774,
+      "loss": 0.2766,
+      "step": 221500
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00019337720698305431,
+      "loss": 0.2766,
+      "step": 222000
+    },
+    {
+      "epoch": 5.17,
+      "eval_loss": 0.25992023944854736,
+      "eval_runtime": 1.4053,
+      "eval_samples_per_second": 1564.073,
+      "eval_steps_per_second": 24.906,
+      "step": 222000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0001929145914383038,
+      "loss": 0.2763,
+      "step": 222500
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00019245156126565586,
+      "loss": 0.2763,
+      "step": 223000
+    },
+    {
+      "epoch": 5.2,
+      "eval_loss": 0.2582224905490875,
+      "eval_runtime": 1.4422,
+      "eval_samples_per_second": 1524.087,
+      "eval_steps_per_second": 24.269,
+      "step": 223000
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00019198812152873416,
+      "loss": 0.2762,
+      "step": 223500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00019152427729564144,
+      "loss": 0.2764,
+      "step": 224000
+    },
+    {
+      "epoch": 5.22,
+      "eval_loss": 0.26058298349380493,
+      "eval_runtime": 1.4571,
+      "eval_samples_per_second": 1508.477,
+      "eval_steps_per_second": 24.02,
+      "step": 224000
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00019106003363890395,
+      "loss": 0.2761,
+      "step": 224500
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00019059539563541584,
+      "loss": 0.276,
+      "step": 225000
+    },
+    {
+      "epoch": 5.24,
+      "eval_loss": 0.25854378938674927,
+      "eval_runtime": 1.4732,
+      "eval_samples_per_second": 1491.985,
+      "eval_steps_per_second": 23.758,
+      "step": 225000
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.000190130368366384,
+      "loss": 0.2761,
+      "step": 225500
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00018966495691727207,
+      "loss": 0.2758,
+      "step": 226000
+    },
+    {
+      "epoch": 5.27,
+      "eval_loss": 0.26082664728164673,
+      "eval_runtime": 1.3852,
+      "eval_samples_per_second": 1586.815,
+      "eval_steps_per_second": 25.268,
+      "step": 226000
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0001891991663777451,
+      "loss": 0.2759,
+      "step": 226500
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00018873300184161387,
+      "loss": 0.2761,
+      "step": 227000
+    },
+    {
+      "epoch": 5.29,
+      "eval_loss": 0.2606270909309387,
+      "eval_runtime": 1.4133,
+      "eval_samples_per_second": 1555.208,
+      "eval_steps_per_second": 24.764,
+      "step": 227000
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00018826646840677894,
+      "loss": 0.2764,
+      "step": 227500
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00018779957117517532,
+      "loss": 0.2759,
+      "step": 228000
+    },
+    {
+      "epoch": 5.31,
+      "eval_loss": 0.25820019841194153,
+      "eval_runtime": 1.4725,
+      "eval_samples_per_second": 1492.708,
+      "eval_steps_per_second": 23.769,
+      "step": 228000
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00018733231525271625,
+      "loss": 0.2754,
+      "step": 228500
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00018686470574923766,
+      "loss": 0.2754,
+      "step": 229000
+    },
+    {
+      "epoch": 5.34,
+      "eval_loss": 0.25702887773513794,
+      "eval_runtime": 1.4623,
+      "eval_samples_per_second": 1503.12,
+      "eval_steps_per_second": 23.935,
+      "step": 229000
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00018639674777844224,
+      "loss": 0.2754,
+      "step": 229500
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00018592844645784327,
+      "loss": 0.2753,
+      "step": 230000
+    },
+    {
+      "epoch": 5.36,
+      "eval_loss": 0.25963670015335083,
+      "eval_runtime": 1.4406,
+      "eval_samples_per_second": 1525.752,
+      "eval_steps_per_second": 24.295,
+      "step": 230000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00018545980690870903,
+      "loss": 0.276,
+      "step": 230500
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00018499083425600648,
+      "loss": 0.2753,
+      "step": 231000
+    },
+    {
+      "epoch": 5.38,
+      "eval_loss": 0.2578088045120239,
+      "eval_runtime": 1.4301,
+      "eval_samples_per_second": 1536.977,
+      "eval_steps_per_second": 24.474,
+      "step": 231000
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00018452153362834552,
+      "loss": 0.275,
+      "step": 231500
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00018405191015792254,
+      "loss": 0.2749,
+      "step": 232000
+    },
+    {
+      "epoch": 5.41,
+      "eval_loss": 0.25682488083839417,
+      "eval_runtime": 1.4507,
+      "eval_samples_per_second": 1515.179,
+      "eval_steps_per_second": 24.127,
+      "step": 232000
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0001835819689804646,
+      "loss": 0.2749,
+      "step": 232500
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0001831117152351732,
+      "loss": 0.2749,
+      "step": 233000
+    },
+    {
+      "epoch": 5.43,
+      "eval_loss": 0.25751060247421265,
+      "eval_runtime": 1.429,
+      "eval_samples_per_second": 1538.092,
+      "eval_steps_per_second": 24.492,
+      "step": 233000
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00018264115406466778,
+      "loss": 0.2752,
+      "step": 233500
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00018217029061493007,
+      "loss": 0.2748,
+      "step": 234000
+    },
+    {
+      "epoch": 5.45,
+      "eval_loss": 0.2570262551307678,
+      "eval_runtime": 1.4679,
+      "eval_samples_per_second": 1497.347,
+      "eval_steps_per_second": 23.843,
+      "step": 234000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00018169913003524717,
+      "loss": 0.2748,
+      "step": 234500
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00018122767747815594,
+      "loss": 0.2744,
+      "step": 235000
+    },
+    {
+      "epoch": 5.48,
+      "eval_loss": 0.2594066262245178,
+      "eval_runtime": 1.449,
+      "eval_samples_per_second": 1516.917,
+      "eval_steps_per_second": 24.155,
+      "step": 235000
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00018075593809938574,
+      "loss": 0.2746,
+      "step": 235500
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00018028391705780295,
+      "loss": 0.2748,
+      "step": 236000
+    },
+    {
+      "epoch": 5.5,
+      "eval_loss": 0.25915682315826416,
+      "eval_runtime": 1.4494,
+      "eval_samples_per_second": 1516.483,
+      "eval_steps_per_second": 24.148,
+      "step": 236000
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0001798116195153541,
+      "loss": 0.2745,
+      "step": 236500
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0001793390506370094,
+      "loss": 0.2742,
+      "step": 237000
+    },
+    {
+      "epoch": 5.52,
+      "eval_loss": 0.25850772857666016,
+      "eval_runtime": 1.4467,
+      "eval_samples_per_second": 1519.275,
+      "eval_steps_per_second": 24.192,
+      "step": 237000
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00017886621559070638,
+      "loss": 0.2743,
+      "step": 237500
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00017839311954729337,
+      "loss": 0.2742,
+      "step": 238000
+    },
+    {
+      "epoch": 5.55,
+      "eval_loss": 0.25702983140945435,
+      "eval_runtime": 1.4245,
+      "eval_samples_per_second": 1542.944,
+      "eval_steps_per_second": 24.569,
+      "step": 238000
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00017791976768047292,
+      "loss": 0.2738,
+      "step": 238500
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00017744616516674518,
+      "loss": 0.2743,
+      "step": 239000
+    },
+    {
+      "epoch": 5.57,
+      "eval_loss": 0.2567313611507416,
+      "eval_runtime": 1.5014,
+      "eval_samples_per_second": 1463.976,
+      "eval_steps_per_second": 23.312,
+      "step": 239000
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00017697231718535132,
+      "loss": 0.2742,
+      "step": 239500
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00017649822891821707,
+      "loss": 0.274,
+      "step": 240000
+    },
+    {
+      "epoch": 5.59,
+      "eval_loss": 0.25702497363090515,
+      "eval_runtime": 1.442,
+      "eval_samples_per_second": 1524.238,
+      "eval_steps_per_second": 24.271,
+      "step": 240000
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00017602390554989563,
+      "loss": 0.274,
+      "step": 240500
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0001755493522675115,
+      "loss": 0.2739,
+      "step": 241000
+    },
+    {
+      "epoch": 5.62,
+      "eval_loss": 0.2584410309791565,
+      "eval_runtime": 1.4562,
+      "eval_samples_per_second": 1509.435,
+      "eval_steps_per_second": 24.036,
+      "step": 241000
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00017507457426070317,
+      "loss": 0.2737,
+      "step": 241500
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00017459957672156704,
+      "loss": 0.2738,
+      "step": 242000
+    },
+    {
+      "epoch": 5.64,
+      "eval_loss": 0.2564358711242676,
+      "eval_runtime": 1.4454,
+      "eval_samples_per_second": 1520.687,
+      "eval_steps_per_second": 24.215,
+      "step": 242000
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00017412436484459998,
+      "loss": 0.2746,
+      "step": 242500
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00017364894382664297,
+      "loss": 0.2737,
+      "step": 243000
+    },
+    {
+      "epoch": 5.66,
+      "eval_loss": 0.25592368841171265,
+      "eval_runtime": 1.453,
+      "eval_samples_per_second": 1512.695,
+      "eval_steps_per_second": 24.087,
+      "step": 243000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00017317331886682408,
+      "loss": 0.2735,
+      "step": 243500
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0001726974951665017,
+      "loss": 0.2732,
+      "step": 244000
+    },
+    {
+      "epoch": 5.69,
+      "eval_loss": 0.25524938106536865,
+      "eval_runtime": 1.4633,
+      "eval_samples_per_second": 1502.063,
+      "eval_steps_per_second": 23.918,
+      "step": 244000
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0001722214779292076,
+      "loss": 0.2731,
+      "step": 244500
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00017174527236058998,
+      "loss": 0.2733,
+      "step": 245000
+    },
+    {
+      "epoch": 5.71,
+      "eval_loss": 0.25596246123313904,
+      "eval_runtime": 1.4704,
+      "eval_samples_per_second": 1494.824,
+      "eval_steps_per_second": 23.803,
+      "step": 245000
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00017126888366835662,
+      "loss": 0.2737,
+      "step": 245500
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.000170792317062218,
+      "loss": 0.2733,
+      "step": 246000
+    },
+    {
+      "epoch": 5.73,
+      "eval_loss": 0.255545437335968,
+      "eval_runtime": 1.4656,
+      "eval_samples_per_second": 1499.768,
+      "eval_steps_per_second": 23.882,
+      "step": 246000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00017031557775383011,
+      "loss": 0.2729,
+      "step": 246500
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0001698386709567377,
+      "loss": 0.2732,
+      "step": 247000
+    },
+    {
+      "epoch": 5.76,
+      "eval_loss": 0.2539271414279938,
+      "eval_runtime": 1.4492,
+      "eval_samples_per_second": 1516.729,
+      "eval_steps_per_second": 24.152,
+      "step": 247000
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0001693616018863171,
+      "loss": 0.273,
+      "step": 247500
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00016888437575971913,
+      "loss": 0.2732,
+      "step": 248000
+    },
+    {
+      "epoch": 5.78,
+      "eval_loss": 0.2551625669002533,
+      "eval_runtime": 1.4356,
+      "eval_samples_per_second": 1531.074,
+      "eval_steps_per_second": 24.38,
+      "step": 248000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00016840699779581238,
+      "loss": 0.2734,
+      "step": 248500
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00016792947321512573,
+      "loss": 0.2729,
+      "step": 249000
+    },
+    {
+      "epoch": 5.8,
+      "eval_loss": 0.25349161028862,
+      "eval_runtime": 1.4447,
+      "eval_samples_per_second": 1521.428,
+      "eval_steps_per_second": 24.227,
+      "step": 249000
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00016745180723979144,
+      "loss": 0.2731,
+      "step": 249500
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00016697400509348818,
+      "loss": 0.2728,
+      "step": 250000
+    },
+    {
+      "epoch": 5.82,
+      "eval_loss": 0.25426918268203735,
+      "eval_runtime": 1.4253,
+      "eval_samples_per_second": 1542.185,
+      "eval_steps_per_second": 24.557,
+      "step": 250000
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00016649607200138356,
+      "loss": 0.2728,
+      "step": 250500
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00016601801319007743,
+      "loss": 0.2729,
+      "step": 251000
+    },
+    {
+      "epoch": 5.85,
+      "eval_loss": 0.25545501708984375,
+      "eval_runtime": 1.4675,
+      "eval_samples_per_second": 1497.796,
+      "eval_steps_per_second": 23.85,
+      "step": 251000
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00016553983388754428,
+      "loss": 0.2725,
+      "step": 251500
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00016506153932307636,
+      "loss": 0.2723,
+      "step": 252000
+    },
+    {
+      "epoch": 5.87,
+      "eval_loss": 0.25281623005867004,
+      "eval_runtime": 1.4602,
+      "eval_samples_per_second": 1505.31,
+      "eval_steps_per_second": 23.97,
+      "step": 252000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00016458313472722638,
+      "loss": 0.2722,
+      "step": 252500
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00016410462533175045,
+      "loss": 0.2722,
+      "step": 253000
+    },
+    {
+      "epoch": 5.89,
+      "eval_loss": 0.2557053565979004,
+      "eval_runtime": 1.4444,
+      "eval_samples_per_second": 1521.717,
+      "eval_steps_per_second": 24.231,
+      "step": 253000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00016362601636955049,
+      "loss": 0.2719,
+      "step": 253500
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00016314731307461754,
+      "loss": 0.2722,
+      "step": 254000
+    },
+    {
+      "epoch": 5.92,
+      "eval_loss": 0.2575713098049164,
+      "eval_runtime": 1.4237,
+      "eval_samples_per_second": 1543.898,
+      "eval_steps_per_second": 24.584,
+      "step": 254000
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0001626685206819742,
+      "loss": 0.2723,
+      "step": 254500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0001621896444276172,
+      "loss": 0.2718,
+      "step": 255000
+    },
+    {
+      "epoch": 5.94,
+      "eval_loss": 0.25522297620773315,
+      "eval_runtime": 1.5788,
+      "eval_samples_per_second": 1392.174,
+      "eval_steps_per_second": 22.168,
+      "step": 255000
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00016171068954846067,
+      "loss": 0.2721,
+      "step": 255500
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00016123166128227835,
+      "loss": 0.2715,
+      "step": 256000
+    },
+    {
+      "epoch": 5.96,
+      "eval_loss": 0.25572481751441956,
+      "eval_runtime": 1.4307,
+      "eval_samples_per_second": 1536.298,
+      "eval_steps_per_second": 24.463,
+      "step": 256000
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0001607525648676467,
+      "loss": 0.2725,
+      "step": 256500
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0001602734055438873,
+      "loss": 0.2725,
+      "step": 257000
+    },
+    {
+      "epoch": 5.99,
+      "eval_loss": 0.2527320981025696,
+      "eval_runtime": 1.4262,
+      "eval_samples_per_second": 1541.159,
+      "eval_steps_per_second": 24.541,
+      "step": 257000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00015979418855100963,
+      "loss": 0.2721,
+      "step": 257500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00015931491912965417,
+      "loss": 0.2717,
+      "step": 258000
+    },
+    {
+      "epoch": 6.01,
+      "eval_loss": 0.2517436146736145,
+      "eval_runtime": 1.7752,
+      "eval_samples_per_second": 1238.2,
+      "eval_steps_per_second": 19.717,
+      "step": 258000
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001588356025210344,
+      "loss": 0.2717,
+      "step": 258500
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00015835624396688,
+      "loss": 0.272,
+      "step": 259000
+    },
+    {
+      "epoch": 6.03,
+      "eval_loss": 0.2557397484779358,
+      "eval_runtime": 1.4282,
+      "eval_samples_per_second": 1538.991,
+      "eval_steps_per_second": 24.506,
+      "step": 259000
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00015787684870937924,
+      "loss": 0.2717,
+      "step": 259500
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00015739742199112196,
+      "loss": 0.2717,
+      "step": 260000
+    },
+    {
+      "epoch": 6.06,
+      "eval_loss": 0.25458213686943054,
+      "eval_runtime": 1.4334,
+      "eval_samples_per_second": 1533.386,
+      "eval_steps_per_second": 24.417,
+      "step": 260000
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00015691796905504187,
+      "loss": 0.2712,
+      "step": 260500
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00015643849514435944,
+      "loss": 0.2709,
+      "step": 261000
+    },
+    {
+      "epoch": 6.08,
+      "eval_loss": 0.2548723816871643,
+      "eval_runtime": 1.4314,
+      "eval_samples_per_second": 1535.607,
+      "eval_steps_per_second": 24.452,
+      "step": 261000
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00015595900550252463,
+      "loss": 0.2707,
+      "step": 261500
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00015547950537315926,
+      "loss": 0.2708,
+      "step": 262000
+    },
+    {
+      "epoch": 6.1,
+      "eval_loss": 0.2543224096298218,
+      "eval_runtime": 1.4239,
+      "eval_samples_per_second": 1543.639,
+      "eval_steps_per_second": 24.58,
+      "step": 262000
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00015499999999999997,
+      "loss": 0.2708,
+      "step": 262500
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00015452049462684068,
+      "loss": 0.2719,
+      "step": 263000
+    },
+    {
+      "epoch": 6.13,
+      "eval_loss": 0.25397995114326477,
+      "eval_runtime": 1.4586,
+      "eval_samples_per_second": 1506.925,
+      "eval_steps_per_second": 23.996,
+      "step": 263000
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00015404099449747535,
+      "loss": 0.2712,
+      "step": 263500
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001535615048556405,
+      "loss": 0.2706,
+      "step": 264000
+    },
+    {
+      "epoch": 6.15,
+      "eval_loss": 0.2522125244140625,
+      "eval_runtime": 1.474,
+      "eval_samples_per_second": 1491.135,
+      "eval_steps_per_second": 23.744,
+      "step": 264000
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0001530820309449581,
+      "loss": 0.2705,
+      "step": 264500
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00015260257800887798,
+      "loss": 0.2705,
+      "step": 265000
+    },
+    {
+      "epoch": 6.17,
+      "eval_loss": 0.25516122579574585,
+      "eval_runtime": 1.3844,
+      "eval_samples_per_second": 1587.692,
+      "eval_steps_per_second": 25.282,
+      "step": 265000
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001521231512906207,
+      "loss": 0.2709,
+      "step": 265500
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00015164375603311998,
+      "loss": 0.2708,
+      "step": 266000
+    },
+    {
+      "epoch": 6.2,
+      "eval_loss": 0.25247690081596375,
+      "eval_runtime": 1.4156,
+      "eval_samples_per_second": 1552.724,
+      "eval_steps_per_second": 24.725,
+      "step": 266000
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00015116439747896553,
+      "loss": 0.271,
+      "step": 266500
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00015068508087034578,
+      "loss": 0.2704,
+      "step": 267000
+    },
+    {
+      "epoch": 6.22,
+      "eval_loss": 0.25168919563293457,
+      "eval_runtime": 1.4279,
+      "eval_samples_per_second": 1539.31,
+      "eval_steps_per_second": 24.511,
+      "step": 267000
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00015020581144899027,
+      "loss": 0.2702,
+      "step": 267500
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001497265944561127,
+      "loss": 0.2701,
+      "step": 268000
+    },
+    {
+      "epoch": 6.24,
+      "eval_loss": 0.2535783052444458,
+      "eval_runtime": 1.4078,
+      "eval_samples_per_second": 1561.345,
+      "eval_steps_per_second": 24.862,
+      "step": 268000
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00014924743513235327,
+      "loss": 0.2703,
+      "step": 268500
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0001487683387177216,
+      "loss": 0.2704,
+      "step": 269000
+    },
+    {
+      "epoch": 6.27,
+      "eval_loss": 0.25389179587364197,
+      "eval_runtime": 1.4483,
+      "eval_samples_per_second": 1517.622,
+      "eval_steps_per_second": 24.166,
+      "step": 269000
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00014828931045153928,
+      "loss": 0.2703,
+      "step": 269500
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00014781035557238272,
+      "loss": 0.2702,
+      "step": 270000
+    },
+    {
+      "epoch": 6.29,
+      "eval_loss": 0.2548398971557617,
+      "eval_runtime": 1.4524,
+      "eval_samples_per_second": 1513.389,
+      "eval_steps_per_second": 24.099,
+      "step": 270000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00014733147931802578,
+      "loss": 0.27,
+      "step": 270500
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00014685268692538238,
+      "loss": 0.2698,
+      "step": 271000
+    },
+    {
+      "epoch": 6.31,
+      "eval_loss": 0.2508079409599304,
+      "eval_runtime": 1.42,
+      "eval_samples_per_second": 1547.911,
+      "eval_steps_per_second": 24.648,
+      "step": 271000
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00014637398363044946,
+      "loss": 0.2699,
+      "step": 271500
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00014589537466824955,
+      "loss": 0.2703,
+      "step": 272000
+    },
+    {
+      "epoch": 6.34,
+      "eval_loss": 0.2517754137516022,
+      "eval_runtime": 1.4219,
+      "eval_samples_per_second": 1545.829,
+      "eval_steps_per_second": 24.615,
+      "step": 272000
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00014541686527277356,
+      "loss": 0.27,
+      "step": 272500
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00014493846067692358,
+      "loss": 0.2693,
+      "step": 273000
+    },
+    {
+      "epoch": 6.36,
+      "eval_loss": 0.25326284766197205,
+      "eval_runtime": 1.4476,
+      "eval_samples_per_second": 1518.396,
+      "eval_steps_per_second": 24.178,
+      "step": 273000
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00014446016611245567,
+      "loss": 0.2694,
+      "step": 273500
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00014398198680992252,
+      "loss": 0.2698,
+      "step": 274000
+    },
+    {
+      "epoch": 6.38,
+      "eval_loss": 0.25300565361976624,
+      "eval_runtime": 1.4333,
+      "eval_samples_per_second": 1533.534,
+      "eval_steps_per_second": 24.419,
+      "step": 274000
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00014350392799861636,
+      "loss": 0.2699,
+      "step": 274500
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0001430259949065118,
+      "loss": 0.2697,
+      "step": 275000
+    },
+    {
+      "epoch": 6.41,
+      "eval_loss": 0.2506762742996216,
+      "eval_runtime": 1.4367,
+      "eval_samples_per_second": 1529.926,
+      "eval_steps_per_second": 24.362,
+      "step": 275000
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0001425481927602085,
+      "loss": 0.2694,
+      "step": 275500
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001420705267848743,
+      "loss": 0.2691,
+      "step": 276000
+    },
+    {
+      "epoch": 6.43,
+      "eval_loss": 0.25103089213371277,
+      "eval_runtime": 1.4586,
+      "eval_samples_per_second": 1506.907,
+      "eval_steps_per_second": 23.995,
+      "step": 276000
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00014159300220418757,
+      "loss": 0.2692,
+      "step": 276500
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0001411156242402808,
+      "loss": 0.2691,
+      "step": 277000
+    },
+    {
+      "epoch": 6.45,
+      "eval_loss": 0.2522771656513214,
+      "eval_runtime": 1.4334,
+      "eval_samples_per_second": 1533.4,
+      "eval_steps_per_second": 24.417,
+      "step": 277000
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0001406383981136829,
+      "loss": 0.269,
+      "step": 277500
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00014016132904326226,
+      "loss": 0.2689,
+      "step": 278000
+    },
+    {
+      "epoch": 6.48,
+      "eval_loss": 0.2521708011627197,
+      "eval_runtime": 1.4301,
+      "eval_samples_per_second": 1536.934,
+      "eval_steps_per_second": 24.473,
+      "step": 278000
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00013968442224616989,
+      "loss": 0.269,
+      "step": 278500
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00013920768293778195,
+      "loss": 0.2688,
+      "step": 279000
+    },
+    {
+      "epoch": 6.5,
+      "eval_loss": 0.25149086117744446,
+      "eval_runtime": 1.4307,
+      "eval_samples_per_second": 1536.357,
+      "eval_steps_per_second": 24.464,
+      "step": 279000
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00013873111633164336,
+      "loss": 0.2686,
+      "step": 279500
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00013825472763941,
+      "loss": 0.2686,
+      "step": 280000
+    },
+    {
+      "epoch": 6.52,
+      "eval_loss": 0.25161558389663696,
+      "eval_runtime": 1.4115,
+      "eval_samples_per_second": 1557.23,
+      "eval_steps_per_second": 24.797,
+      "step": 280000
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00013777852207079235,
+      "loss": 0.2691,
+      "step": 280500
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00013730250483349825,
+      "loss": 0.2686,
+      "step": 281000
+    },
+    {
+      "epoch": 6.55,
+      "eval_loss": 0.2525520324707031,
+      "eval_runtime": 1.4509,
+      "eval_samples_per_second": 1514.902,
+      "eval_steps_per_second": 24.123,
+      "step": 281000
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00013682668113317584,
+      "loss": 0.2688,
+      "step": 281500
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00013635105617335703,
+      "loss": 0.2686,
+      "step": 282000
+    },
+    {
+      "epoch": 6.57,
+      "eval_loss": 0.2530405819416046,
+      "eval_runtime": 1.446,
+      "eval_samples_per_second": 1520.041,
+      "eval_steps_per_second": 24.204,
+      "step": 282000
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00013587563515539996,
+      "loss": 0.2686,
+      "step": 282500
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00013540042327843296,
+      "loss": 0.2681,
+      "step": 283000
+    },
+    {
+      "epoch": 6.59,
+      "eval_loss": 0.252304345369339,
+      "eval_runtime": 1.4523,
+      "eval_samples_per_second": 1513.433,
+      "eval_steps_per_second": 24.099,
+      "step": 283000
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00013492542573929678,
+      "loss": 0.2686,
+      "step": 283500
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00013445064773248846,
+      "loss": 0.2683,
+      "step": 284000
+    },
+    {
+      "epoch": 6.62,
+      "eval_loss": 0.2517680823802948,
+      "eval_runtime": 1.4564,
+      "eval_samples_per_second": 1509.219,
+      "eval_steps_per_second": 24.032,
+      "step": 284000
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00013397609445010432,
+      "loss": 0.2684,
+      "step": 284500
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00013350177108178288,
+      "loss": 0.268,
+      "step": 285000
+    },
+    {
+      "epoch": 6.64,
+      "eval_loss": 0.25080546736717224,
+      "eval_runtime": 1.4547,
+      "eval_samples_per_second": 1510.937,
+      "eval_steps_per_second": 24.06,
+      "step": 285000
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00013302768281464863,
+      "loss": 0.2682,
+      "step": 285500
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0001325538348332548,
+      "loss": 0.2683,
+      "step": 286000
+    },
+    {
+      "epoch": 6.66,
+      "eval_loss": 0.25022614002227783,
+      "eval_runtime": 1.4198,
+      "eval_samples_per_second": 1548.074,
+      "eval_steps_per_second": 24.651,
+      "step": 286000
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00013208023231952706,
+      "loss": 0.2684,
+      "step": 286500
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001316068804527066,
+      "loss": 0.2681,
+      "step": 287000
+    },
+    {
+      "epoch": 6.69,
+      "eval_loss": 0.2507327198982239,
+      "eval_runtime": 1.423,
+      "eval_samples_per_second": 1544.594,
+      "eval_steps_per_second": 24.595,
+      "step": 287000
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00013113378440929353,
+      "loss": 0.2682,
+      "step": 287500
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00013066094936299056,
+      "loss": 0.2681,
+      "step": 288000
+    },
+    {
+      "epoch": 6.71,
+      "eval_loss": 0.25175246596336365,
+      "eval_runtime": 1.4465,
+      "eval_samples_per_second": 1519.496,
+      "eval_steps_per_second": 24.196,
+      "step": 288000
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00013018838048464582,
+      "loss": 0.2677,
+      "step": 288500
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00012971608294219702,
+      "loss": 0.2679,
+      "step": 289000
+    },
+    {
+      "epoch": 6.73,
+      "eval_loss": 0.2503984272480011,
+      "eval_runtime": 1.4389,
+      "eval_samples_per_second": 1527.568,
+      "eval_steps_per_second": 24.324,
+      "step": 289000
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00012924406190061423,
+      "loss": 0.2677,
+      "step": 289500
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0001287723225218441,
+      "loss": 0.2675,
+      "step": 290000
+    },
+    {
+      "epoch": 6.76,
+      "eval_loss": 0.250543475151062,
+      "eval_runtime": 1.4231,
+      "eval_samples_per_second": 1544.544,
+      "eval_steps_per_second": 24.595,
+      "step": 290000
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00012830086996475274,
+      "loss": 0.2677,
+      "step": 290500
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00012782970938506988,
+      "loss": 0.2679,
+      "step": 291000
+    },
+    {
+      "epoch": 6.78,
+      "eval_loss": 0.2502252459526062,
+      "eval_runtime": 1.4522,
+      "eval_samples_per_second": 1513.548,
+      "eval_steps_per_second": 24.101,
+      "step": 291000
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00012735884593533222,
+      "loss": 0.2674,
+      "step": 291500
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0001268882847648268,
+      "loss": 0.2675,
+      "step": 292000
+    },
+    {
+      "epoch": 6.8,
+      "eval_loss": 0.25172051787376404,
+      "eval_runtime": 1.4702,
+      "eval_samples_per_second": 1495.064,
+      "eval_steps_per_second": 23.807,
+      "step": 292000
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00012641803101953535,
+      "loss": 0.2671,
+      "step": 292500
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00012594808984207743,
+      "loss": 0.2672,
+      "step": 293000
+    },
+    {
+      "epoch": 6.83,
+      "eval_loss": 0.2516680657863617,
+      "eval_runtime": 1.4673,
+      "eval_samples_per_second": 1498.034,
+      "eval_steps_per_second": 23.854,
+      "step": 293000
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00012547846637165445,
+      "loss": 0.2672,
+      "step": 293500
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00012500916574399346,
+      "loss": 0.2673,
+      "step": 294000
+    },
+    {
+      "epoch": 6.85,
+      "eval_loss": 0.24834002554416656,
+      "eval_runtime": 1.4335,
+      "eval_samples_per_second": 1533.285,
+      "eval_steps_per_second": 24.415,
+      "step": 294000
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00012454019309129095,
+      "loss": 0.2676,
+      "step": 294500
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.0001240715535421567,
+      "loss": 0.2671,
+      "step": 295000
+    },
+    {
+      "epoch": 6.87,
+      "eval_loss": 0.24830973148345947,
+      "eval_runtime": 1.4223,
+      "eval_samples_per_second": 1545.387,
+      "eval_steps_per_second": 24.608,
+      "step": 295000
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00012360325222155773,
+      "loss": 0.2669,
+      "step": 295500
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00012313529425076228,
+      "loss": 0.2669,
+      "step": 296000
+    },
+    {
+      "epoch": 6.9,
+      "eval_loss": 0.25227677822113037,
+      "eval_runtime": 1.4524,
+      "eval_samples_per_second": 1513.37,
+      "eval_steps_per_second": 24.098,
+      "step": 296000
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00012266768474728372,
+      "loss": 0.2669,
+      "step": 296500
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00012220042882482468,
+      "loss": 0.2666,
+      "step": 297000
+    },
+    {
+      "epoch": 6.92,
+      "eval_loss": 0.2493974268436432,
+      "eval_runtime": 1.4257,
+      "eval_samples_per_second": 1541.745,
+      "eval_steps_per_second": 24.55,
+      "step": 297000
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00012173353159322102,
+      "loss": 0.2664,
+      "step": 297500
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00012126699815838609,
+      "loss": 0.2664,
+      "step": 298000
+    },
+    {
+      "epoch": 6.94,
+      "eval_loss": 0.24965964257717133,
+      "eval_runtime": 1.4552,
+      "eval_samples_per_second": 1510.413,
+      "eval_steps_per_second": 24.051,
+      "step": 298000
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00012080083362225484,
+      "loss": 0.2664,
+      "step": 298500
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00012033504308272786,
+      "loss": 0.2953,
+      "step": 299000
+    },
+    {
+      "epoch": 6.97,
+      "eval_loss": 0.6473129987716675,
+      "eval_runtime": 1.4394,
+      "eval_samples_per_second": 1527.02,
+      "eval_steps_per_second": 24.316,
+      "step": 299000
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.00011986963163361598,
+      "loss": 0.6484,
+      "step": 299500
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0001194046043645841,
+      "loss": 0.6483,
+      "step": 300000
+    },
+    {
+      "epoch": 6.99,
+      "eval_loss": 0.6471491456031799,
+      "eval_runtime": 1.4019,
+      "eval_samples_per_second": 1567.888,
+      "eval_steps_per_second": 24.966,
+      "step": 300000
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00011893996636109606,
+      "loss": 0.6483,
+      "step": 300500
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.00011847572270435852,
+      "loss": 0.6477,
+      "step": 301000
+    },
+    {
+      "epoch": 7.01,
+      "eval_loss": 0.6460827589035034,
+      "eval_runtime": 1.4908,
+      "eval_samples_per_second": 1474.348,
+      "eval_steps_per_second": 23.477,
+      "step": 301000
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.00011801187847126579,
+      "loss": 0.6478,
+      "step": 301500
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.00011754843873434411,
+      "loss": 0.6471,
+      "step": 302000
+    },
+    {
+      "epoch": 7.04,
+      "eval_loss": 0.6455149054527283,
+      "eval_runtime": 1.4371,
+      "eval_samples_per_second": 1529.505,
+      "eval_steps_per_second": 24.355,
+      "step": 302000
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.00011708540856169612,
+      "loss": 0.6443,
+      "step": 302500
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.00011662279301694567,
+      "loss": 0.4659,
+      "step": 303000
+    },
+    {
+      "epoch": 7.06,
+      "eval_loss": 0.25994545221328735,
+      "eval_runtime": 1.3921,
+      "eval_samples_per_second": 1578.885,
+      "eval_steps_per_second": 25.141,
+      "step": 303000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.0001161605971591822,
+      "loss": 0.2739,
+      "step": 303500
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00011569882604290559,
+      "loss": 0.2707,
+      "step": 304000
+    },
+    {
+      "epoch": 7.08,
+      "eval_loss": 0.25346240401268005,
+      "eval_runtime": 1.4298,
+      "eval_samples_per_second": 1537.235,
+      "eval_steps_per_second": 24.478,
+      "step": 304000
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00011523748471797075,
+      "loss": 0.2691,
+      "step": 304500
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00011477657822953255,
+      "loss": 0.2681,
+      "step": 305000
+    },
+    {
+      "epoch": 7.11,
+      "eval_loss": 0.2524736821651459,
+      "eval_runtime": 1.4268,
+      "eval_samples_per_second": 1540.462,
+      "eval_steps_per_second": 24.53,
+      "step": 305000
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.00011431611161799043,
+      "loss": 0.268,
+      "step": 305500
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.0001138560899189335,
+      "loss": 0.2678,
+      "step": 306000
+    },
+    {
+      "epoch": 7.13,
+      "eval_loss": 0.25040730834007263,
+      "eval_runtime": 1.4446,
+      "eval_samples_per_second": 1521.535,
+      "eval_steps_per_second": 24.228,
+      "step": 306000
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00011339651816308543,
+      "loss": 0.267,
+      "step": 306500
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.00011293740137624925,
+      "loss": 0.267,
+      "step": 307000
+    },
+    {
+      "epoch": 7.15,
+      "eval_loss": 0.24852143228054047,
+      "eval_runtime": 1.3837,
+      "eval_samples_per_second": 1588.479,
+      "eval_steps_per_second": 25.294,
+      "step": 307000
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.00011247874457925261,
+      "loss": 0.2668,
+      "step": 307500
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.0001120205527878927,
+      "loss": 0.2663,
+      "step": 308000
+    },
+    {
+      "epoch": 7.18,
+      "eval_loss": 0.2509649991989136,
+      "eval_runtime": 1.4317,
+      "eval_samples_per_second": 1535.233,
+      "eval_steps_per_second": 24.446,
+      "step": 308000
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.00011156283101288165,
+      "loss": 0.266,
+      "step": 308500
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.00011110558425979132,
+      "loss": 0.266,
+      "step": 309000
+    },
+    {
+      "epoch": 7.2,
+      "eval_loss": 0.24989214539527893,
+      "eval_runtime": 1.446,
+      "eval_samples_per_second": 1520.105,
+      "eval_steps_per_second": 24.205,
+      "step": 309000
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00011064881752899906,
+      "loss": 0.2661,
+      "step": 309500
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00011019253581563262,
+      "loss": 0.2658,
+      "step": 310000
+    },
+    {
+      "epoch": 7.22,
+      "eval_loss": 0.2480825036764145,
+      "eval_runtime": 1.4466,
+      "eval_samples_per_second": 1519.474,
+      "eval_steps_per_second": 24.195,
+      "step": 310000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.00010973674410951567,
+      "loss": 0.2657,
+      "step": 310500
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.00010928144739511337,
+      "loss": 0.2656,
+      "step": 311000
+    },
+    {
+      "epoch": 7.25,
+      "eval_loss": 0.2480616420507431,
+      "eval_runtime": 1.4455,
+      "eval_samples_per_second": 1520.549,
+      "eval_steps_per_second": 24.213,
+      "step": 311000
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.00010882665065147757,
+      "loss": 0.2655,
+      "step": 311500
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00010837235885219267,
+      "loss": 0.2653,
+      "step": 312000
+    },
+    {
+      "epoch": 7.27,
+      "eval_loss": 0.2518889009952545,
+      "eval_runtime": 1.4252,
+      "eval_samples_per_second": 1542.203,
+      "eval_steps_per_second": 24.557,
+      "step": 312000
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.00010791857696532089,
+      "loss": 0.2653,
+      "step": 312500
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.00010746530995334832,
+      "loss": 0.2652,
+      "step": 313000
+    },
+    {
+      "epoch": 7.29,
+      "eval_loss": 0.24834561347961426,
+      "eval_runtime": 1.454,
+      "eval_samples_per_second": 1511.735,
+      "eval_steps_per_second": 24.072,
+      "step": 313000
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.0001070125627731304,
+      "loss": 0.2653,
+      "step": 313500
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.0001065603403758377,
+      "loss": 0.2654,
+      "step": 314000
+    },
+    {
+      "epoch": 7.32,
+      "eval_loss": 0.24683910608291626,
+      "eval_runtime": 1.4569,
+      "eval_samples_per_second": 1508.633,
+      "eval_steps_per_second": 24.023,
+      "step": 314000
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.00010610864770690196,
+      "loss": 0.2652,
+      "step": 314500
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.00010565748970596172,
+      "loss": 0.2653,
+      "step": 315000
+    },
+    {
+      "epoch": 7.34,
+      "eval_loss": 0.2493169754743576,
+      "eval_runtime": 1.4715,
+      "eval_samples_per_second": 1493.749,
+      "eval_steps_per_second": 23.786,
+      "step": 315000
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.00010520687130680884,
+      "loss": 0.2653,
+      "step": 315500
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.00010475679743733364,
+      "loss": 0.265,
+      "step": 316000
+    },
+    {
+      "epoch": 7.36,
+      "eval_loss": 0.24726563692092896,
+      "eval_runtime": 1.4143,
+      "eval_samples_per_second": 1554.111,
+      "eval_steps_per_second": 24.747,
+      "step": 316000
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.00010430727301947202,
+      "loss": 0.2647,
+      "step": 316500
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.00010385830296915104,
+      "loss": 0.2646,
+      "step": 317000
+    },
+    {
+      "epoch": 7.39,
+      "eval_loss": 0.24790264666080475,
+      "eval_runtime": 1.4368,
+      "eval_samples_per_second": 1529.828,
+      "eval_steps_per_second": 24.36,
+      "step": 317000
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.00010340989219623508,
+      "loss": 0.2649,
+      "step": 317500
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.0001029620456044727,
+      "loss": 0.2647,
+      "step": 318000
+    },
+    {
+      "epoch": 7.41,
+      "eval_loss": 0.24833044409751892,
+      "eval_runtime": 1.4568,
+      "eval_samples_per_second": 1508.735,
+      "eval_steps_per_second": 24.024,
+      "step": 318000
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00010251476809144226,
+      "loss": 0.2643,
+      "step": 318500
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.00010206806454849917,
+      "loss": 0.2645,
+      "step": 319000
+    },
+    {
+      "epoch": 7.43,
+      "eval_loss": 0.2483694851398468,
+      "eval_runtime": 1.465,
+      "eval_samples_per_second": 1500.321,
+      "eval_steps_per_second": 23.89,
+      "step": 319000
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00010162193986072167,
+      "loss": 0.2644,
+      "step": 319500
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.00010117639890685795,
+      "loss": 0.2642,
+      "step": 320000
+    },
+    {
+      "epoch": 7.46,
+      "eval_loss": 0.2466782182455063,
+      "eval_runtime": 1.4627,
+      "eval_samples_per_second": 1502.721,
+      "eval_steps_per_second": 23.929,
+      "step": 320000
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00010073144655927253,
+      "loss": 0.2641,
+      "step": 320500
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.0001002870876838929,
+      "loss": 0.2643,
+      "step": 321000
+    },
+    {
+      "epoch": 7.48,
+      "eval_loss": 0.24826975166797638,
+      "eval_runtime": 1.437,
+      "eval_samples_per_second": 1529.624,
+      "eval_steps_per_second": 24.357,
+      "step": 321000
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 9.984332714015662e-05,
+      "loss": 0.264,
+      "step": 321500
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 9.94001697809578e-05,
+      "loss": 0.2643,
+      "step": 322000
+    },
+    {
+      "epoch": 7.5,
+      "eval_loss": 0.2487555295228958,
+      "eval_runtime": 1.4486,
+      "eval_samples_per_second": 1517.36,
+      "eval_steps_per_second": 24.162,
+      "step": 322000
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 9.895762045259445e-05,
+      "loss": 0.2639,
+      "step": 322500
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 9.851568399471498e-05,
+      "loss": 0.2638,
+      "step": 323000
+    },
+    {
+      "epoch": 7.53,
+      "eval_loss": 0.2484089434146881,
+      "eval_runtime": 1.4155,
+      "eval_samples_per_second": 1552.76,
+      "eval_steps_per_second": 24.725,
+      "step": 323000
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 9.807436524026574e-05,
+      "loss": 0.2639,
+      "step": 323500
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 9.763366901543801e-05,
+      "loss": 0.2636,
+      "step": 324000
+    },
+    {
+      "epoch": 7.55,
+      "eval_loss": 0.24783042073249817,
+      "eval_runtime": 1.4248,
+      "eval_samples_per_second": 1542.655,
+      "eval_steps_per_second": 24.565,
+      "step": 324000
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 9.719360013961495e-05,
+      "loss": 0.2634,
+      "step": 324500
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 9.675416342531944e-05,
+      "loss": 0.2633,
+      "step": 325000
+    },
+    {
+      "epoch": 7.57,
+      "eval_loss": 0.24799397587776184,
+      "eval_runtime": 1.4677,
+      "eval_samples_per_second": 1497.557,
+      "eval_steps_per_second": 23.846,
+      "step": 325000
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 9.631536367816086e-05,
+      "loss": 0.2629,
+      "step": 325500
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 9.587720569678299e-05,
+      "loss": 0.2633,
+      "step": 326000
+    },
+    {
+      "epoch": 7.6,
+      "eval_loss": 0.24824275076389313,
+      "eval_runtime": 1.4202,
+      "eval_samples_per_second": 1547.658,
+      "eval_steps_per_second": 24.644,
+      "step": 326000
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 9.543969427281131e-05,
+      "loss": 0.2637,
+      "step": 326500
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 9.500283419080062e-05,
+      "loss": 0.2637,
+      "step": 327000
+    },
+    {
+      "epoch": 7.62,
+      "eval_loss": 0.24855847656726837,
+      "eval_runtime": 1.3917,
+      "eval_samples_per_second": 1579.351,
+      "eval_steps_per_second": 25.149,
+      "step": 327000
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 9.45666302281829e-05,
+      "loss": 0.264,
+      "step": 327500
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 9.413108715521467e-05,
+      "loss": 0.2634,
+      "step": 328000
+    },
+    {
+      "epoch": 7.64,
+      "eval_loss": 0.24637793004512787,
+      "eval_runtime": 1.4458,
+      "eval_samples_per_second": 1520.249,
+      "eval_steps_per_second": 24.208,
+      "step": 328000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 9.369620973492525e-05,
+      "loss": 0.2638,
+      "step": 328500
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 9.326200272306445e-05,
+      "loss": 0.2631,
+      "step": 329000
+    },
+    {
+      "epoch": 7.67,
+      "eval_loss": 0.24531778693199158,
+      "eval_runtime": 1.4556,
+      "eval_samples_per_second": 1510.044,
+      "eval_steps_per_second": 24.045,
+      "step": 329000
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 9.282847086805059e-05,
+      "loss": 0.2636,
+      "step": 329500
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 9.239561891091853e-05,
+      "loss": 0.2631,
+      "step": 330000
+    },
+    {
+      "epoch": 7.69,
+      "eval_loss": 0.24822315573692322,
+      "eval_runtime": 1.4463,
+      "eval_samples_per_second": 1519.716,
+      "eval_steps_per_second": 24.199,
+      "step": 330000
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 9.196345158526793e-05,
+      "loss": 0.2632,
+      "step": 330500
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 9.153197361721149e-05,
+      "loss": 0.263,
+      "step": 331000
+    },
+    {
+      "epoch": 7.71,
+      "eval_loss": 0.24357490241527557,
+      "eval_runtime": 1.4413,
+      "eval_samples_per_second": 1524.993,
+      "eval_steps_per_second": 24.283,
+      "step": 331000
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 9.110118972532302e-05,
+      "loss": 0.2627,
+      "step": 331500
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 9.067110462058634e-05,
+      "loss": 0.2631,
+      "step": 332000
+    },
+    {
+      "epoch": 7.74,
+      "eval_loss": 0.2456502616405487,
+      "eval_runtime": 1.4523,
+      "eval_samples_per_second": 1513.463,
+      "eval_steps_per_second": 24.1,
+      "step": 332000
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 9.024172300634305e-05,
+      "loss": 0.2629,
+      "step": 332500
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 8.981304957824182e-05,
+      "loss": 0.2629,
+      "step": 333000
+    },
+    {
+      "epoch": 7.76,
+      "eval_loss": 0.2470385730266571,
+      "eval_runtime": 1.4527,
+      "eval_samples_per_second": 1513.011,
+      "eval_steps_per_second": 24.093,
+      "step": 333000
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 8.938508902418643e-05,
+      "loss": 0.2623,
+      "step": 333500
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 8.89578460242851e-05,
+      "loss": 0.2626,
+      "step": 334000
+    },
+    {
+      "epoch": 7.78,
+      "eval_loss": 0.24880406260490417,
+      "eval_runtime": 1.4245,
+      "eval_samples_per_second": 1543.039,
+      "eval_steps_per_second": 24.571,
+      "step": 334000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 8.85313252507988e-05,
+      "loss": 0.2621,
+      "step": 334500
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 8.810553136809027e-05,
+      "loss": 0.2626,
+      "step": 335000
+    },
+    {
+      "epoch": 7.81,
+      "eval_loss": 0.2465057224035263,
+      "eval_runtime": 1.453,
+      "eval_samples_per_second": 1512.782,
+      "eval_steps_per_second": 24.089,
+      "step": 335000
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 8.76804690325733e-05,
+      "loss": 0.2622,
+      "step": 335500
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 8.725614289266137e-05,
+      "loss": 0.2623,
+      "step": 336000
+    },
+    {
+      "epoch": 7.83,
+      "eval_loss": 0.24810338020324707,
+      "eval_runtime": 1.4394,
+      "eval_samples_per_second": 1526.974,
+      "eval_steps_per_second": 24.315,
+      "step": 336000
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 8.683255758871734e-05,
+      "loss": 0.2627,
+      "step": 336500
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 8.640971775300207e-05,
+      "loss": 0.2624,
+      "step": 337000
+    },
+    {
+      "epoch": 7.85,
+      "eval_loss": 0.24493196606636047,
+      "eval_runtime": 1.4281,
+      "eval_samples_per_second": 1539.133,
+      "eval_steps_per_second": 24.508,
+      "step": 337000
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 8.598762800962431e-05,
+      "loss": 0.2625,
+      "step": 337500
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 8.55662929744899e-05,
+      "loss": 0.2623,
+      "step": 338000
+    },
+    {
+      "epoch": 7.88,
+      "eval_loss": 0.2459965944290161,
+      "eval_runtime": 1.463,
+      "eval_samples_per_second": 1502.342,
+      "eval_steps_per_second": 23.923,
+      "step": 338000
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 8.514571725525124e-05,
+      "loss": 0.2626,
+      "step": 338500
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 8.47259054512571e-05,
+      "loss": 0.2626,
+      "step": 339000
+    },
+    {
+      "epoch": 7.9,
+      "eval_loss": 0.24472270905971527,
+      "eval_runtime": 1.4079,
+      "eval_samples_per_second": 1561.143,
+      "eval_steps_per_second": 24.859,
+      "step": 339000
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 8.430686215350198e-05,
+      "loss": 0.2624,
+      "step": 339500
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 8.388859194457636e-05,
+      "loss": 0.262,
+      "step": 340000
+    },
+    {
+      "epoch": 7.92,
+      "eval_loss": 0.24556826055049896,
+      "eval_runtime": 1.3884,
+      "eval_samples_per_second": 1583.125,
+      "eval_steps_per_second": 25.209,
+      "step": 340000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 8.347109939861605e-05,
+      "loss": 0.2623,
+      "step": 340500
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 8.305438908125285e-05,
+      "loss": 0.2624,
+      "step": 341000
+    },
+    {
+      "epoch": 7.95,
+      "eval_loss": 0.24358774721622467,
+      "eval_runtime": 1.4516,
+      "eval_samples_per_second": 1514.19,
+      "eval_steps_per_second": 24.111,
+      "step": 341000
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 8.263846554956402e-05,
+      "loss": 0.2621,
+      "step": 341500
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 8.222333335202254e-05,
+      "loss": 0.2618,
+      "step": 342000
+    },
+    {
+      "epoch": 7.97,
+      "eval_loss": 0.24512818455696106,
+      "eval_runtime": 1.4448,
+      "eval_samples_per_second": 1521.327,
+      "eval_steps_per_second": 24.225,
+      "step": 342000
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 8.18089970284477e-05,
+      "loss": 0.2616,
+      "step": 342500
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 8.1395461109955e-05,
+      "loss": 0.2617,
+      "step": 343000
+    },
+    {
+      "epoch": 7.99,
+      "eval_loss": 0.24306875467300415,
+      "eval_runtime": 1.4195,
+      "eval_samples_per_second": 1548.387,
+      "eval_steps_per_second": 24.656,
+      "step": 343000
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 8.098273011890726e-05,
+      "loss": 0.2614,
+      "step": 343500
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 8.057080856886426e-05,
+      "loss": 0.2614,
+      "step": 344000
+    },
+    {
+      "epoch": 8.02,
+      "eval_loss": 0.24457786977291107,
+      "eval_runtime": 1.4376,
+      "eval_samples_per_second": 1528.919,
+      "eval_steps_per_second": 24.346,
+      "step": 344000
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 8.015970096453414e-05,
+      "loss": 0.2616,
+      "step": 344500
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 7.974941180172382e-05,
+      "loss": 0.2617,
+      "step": 345000
+    },
+    {
+      "epoch": 8.04,
+      "eval_loss": 0.24457286298274994,
+      "eval_runtime": 1.4399,
+      "eval_samples_per_second": 1526.508,
+      "eval_steps_per_second": 24.307,
+      "step": 345000
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 7.933994556728976e-05,
+      "loss": 0.2614,
+      "step": 345500
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 7.893130673908927e-05,
+      "loss": 0.2612,
+      "step": 346000
+    },
+    {
+      "epoch": 8.06,
+      "eval_loss": 0.24429433047771454,
+      "eval_runtime": 1.4483,
+      "eval_samples_per_second": 1517.626,
+      "eval_steps_per_second": 24.166,
+      "step": 346000
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 7.852349978593091e-05,
+      "loss": 0.2616,
+      "step": 346500
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 7.811652916752633e-05,
+      "loss": 0.261,
+      "step": 347000
+    },
+    {
+      "epoch": 8.08,
+      "eval_loss": 0.24454987049102783,
+      "eval_runtime": 1.4519,
+      "eval_samples_per_second": 1513.913,
+      "eval_steps_per_second": 24.107,
+      "step": 347000
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 7.771039933444092e-05,
+      "loss": 0.261,
+      "step": 347500
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 7.730511472804544e-05,
+      "loss": 0.261,
+      "step": 348000
+    },
+    {
+      "epoch": 8.11,
+      "eval_loss": 0.24601994454860687,
+      "eval_runtime": 1.4399,
+      "eval_samples_per_second": 1526.461,
+      "eval_steps_per_second": 24.307,
+      "step": 348000
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 7.690067978046758e-05,
+      "loss": 0.2611,
+      "step": 348500
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 7.649709891454298e-05,
+      "loss": 0.261,
+      "step": 349000
+    },
+    {
+      "epoch": 8.13,
+      "eval_loss": 0.24198506772518158,
+      "eval_runtime": 1.4613,
+      "eval_samples_per_second": 1504.096,
+      "eval_steps_per_second": 23.951,
+      "step": 349000
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 7.609437654376742e-05,
+      "loss": 0.2606,
+      "step": 349500
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 7.569251707224812e-05,
+      "loss": 0.2608,
+      "step": 350000
+    },
+    {
+      "epoch": 8.15,
+      "eval_loss": 0.24344965815544128,
+      "eval_runtime": 1.444,
+      "eval_samples_per_second": 1522.152,
+      "eval_steps_per_second": 24.238,
+      "step": 350000
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 7.529152489465592e-05,
+      "loss": 0.2607,
+      "step": 350500
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 7.489140439617708e-05,
+      "loss": 0.2607,
+      "step": 351000
+    },
+    {
+      "epoch": 8.18,
+      "eval_loss": 0.24649150669574738,
+      "eval_runtime": 1.45,
+      "eval_samples_per_second": 1515.881,
+      "eval_steps_per_second": 24.138,
+      "step": 351000
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 7.449215995246522e-05,
+      "loss": 0.2606,
+      "step": 351500
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 7.409379592959367e-05,
+      "loss": 0.261,
+      "step": 352000
+    },
+    {
+      "epoch": 8.2,
+      "eval_loss": 0.24395301938056946,
+      "eval_runtime": 1.3957,
+      "eval_samples_per_second": 1574.783,
+      "eval_steps_per_second": 25.076,
+      "step": 352000
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 7.369631668400746e-05,
+      "loss": 0.2605,
+      "step": 352500
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 7.3299726562476e-05,
+      "loss": 0.2605,
+      "step": 353000
+    },
+    {
+      "epoch": 8.22,
+      "eval_loss": 0.24368144571781158,
+      "eval_runtime": 1.4474,
+      "eval_samples_per_second": 1518.547,
+      "eval_steps_per_second": 24.181,
+      "step": 353000
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 7.290402990204531e-05,
+      "loss": 0.2602,
+      "step": 353500
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 7.250923102999073e-05,
+      "loss": 0.2605,
+      "step": 354000
+    },
+    {
+      "epoch": 8.25,
+      "eval_loss": 0.24442052841186523,
+      "eval_runtime": 1.4221,
+      "eval_samples_per_second": 1545.568,
+      "eval_steps_per_second": 24.611,
+      "step": 354000
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 7.211533426376934e-05,
+      "loss": 0.2603,
+      "step": 354500
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 7.172234391097317e-05,
+      "loss": 0.2603,
+      "step": 355000
+    },
+    {
+      "epoch": 8.27,
+      "eval_loss": 0.2446802258491516,
+      "eval_runtime": 1.4551,
+      "eval_samples_per_second": 1510.511,
+      "eval_steps_per_second": 24.053,
+      "step": 355000
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 7.133026426928173e-05,
+      "loss": 0.2603,
+      "step": 355500
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 7.093909962641514e-05,
+      "loss": 0.2599,
+      "step": 356000
+    },
+    {
+      "epoch": 8.29,
+      "eval_loss": 0.24162663519382477,
+      "eval_runtime": 1.4448,
+      "eval_samples_per_second": 1521.367,
+      "eval_steps_per_second": 24.226,
+      "step": 356000
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 7.054885426008737e-05,
+      "loss": 0.2602,
+      "step": 356500
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 7.015953243795907e-05,
+      "loss": 0.2601,
+      "step": 357000
+    },
+    {
+      "epoch": 8.32,
+      "eval_loss": 0.2434534877538681,
+      "eval_runtime": 1.4796,
+      "eval_samples_per_second": 1485.514,
+      "eval_steps_per_second": 23.655,
+      "step": 357000
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 6.97711384175914e-05,
+      "loss": 0.2603,
+      "step": 357500
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 6.938367644639911e-05,
+      "loss": 0.26,
+      "step": 358000
+    },
+    {
+      "epoch": 8.34,
+      "eval_loss": 0.24282395839691162,
+      "eval_runtime": 1.4529,
+      "eval_samples_per_second": 1512.882,
+      "eval_steps_per_second": 24.09,
+      "step": 358000
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 6.899715076160425e-05,
+      "loss": 0.26,
+      "step": 358500
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 6.861156559018986e-05,
+      "loss": 0.26,
+      "step": 359000
+    },
+    {
+      "epoch": 8.36,
+      "eval_loss": 0.2448093742132187,
+      "eval_runtime": 1.4776,
+      "eval_samples_per_second": 1487.575,
+      "eval_steps_per_second": 23.688,
+      "step": 359000
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 6.822692514885346e-05,
+      "loss": 0.2598,
+      "step": 359500
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 6.784323364396135e-05,
+      "loss": 0.2596,
+      "step": 360000
+    },
+    {
+      "epoch": 8.39,
+      "eval_loss": 0.24318400025367737,
+      "eval_runtime": 1.4613,
+      "eval_samples_per_second": 1504.148,
+      "eval_steps_per_second": 23.951,
+      "step": 360000
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 6.746049527150238e-05,
+      "loss": 0.2596,
+      "step": 360500
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 6.707871421704209e-05,
+      "loss": 0.2596,
+      "step": 361000
+    },
+    {
+      "epoch": 8.41,
+      "eval_loss": 0.24351955950260162,
+      "eval_runtime": 1.4786,
+      "eval_samples_per_second": 1486.571,
+      "eval_steps_per_second": 23.672,
+      "step": 361000
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 6.669789465567683e-05,
+      "loss": 0.2593,
+      "step": 361500
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 6.631804075198838e-05,
+      "loss": 0.2595,
+      "step": 362000
+    },
+    {
+      "epoch": 8.43,
+      "eval_loss": 0.2438839077949524,
+      "eval_runtime": 1.4636,
+      "eval_samples_per_second": 1501.733,
+      "eval_steps_per_second": 23.913,
+      "step": 362000
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 6.593915665999816e-05,
+      "loss": 0.2593,
+      "step": 362500
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 6.55612465231219e-05,
+      "loss": 0.2593,
+      "step": 363000
+    },
+    {
+      "epoch": 8.46,
+      "eval_loss": 0.24523594975471497,
+      "eval_runtime": 1.4077,
+      "eval_samples_per_second": 1561.38,
+      "eval_steps_per_second": 24.863,
+      "step": 363000
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 6.518431447412434e-05,
+      "loss": 0.2593,
+      "step": 363500
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 6.480836463507392e-05,
+      "loss": 0.2593,
+      "step": 364000
+    },
+    {
+      "epoch": 8.48,
+      "eval_loss": 0.24123093485832214,
+      "eval_runtime": 1.4526,
+      "eval_samples_per_second": 1513.098,
+      "eval_steps_per_second": 24.094,
+      "step": 364000
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 6.443340111729786e-05,
+      "loss": 0.259,
+      "step": 364500
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 6.405942802133713e-05,
+      "loss": 0.2592,
+      "step": 365000
+    },
+    {
+      "epoch": 8.5,
+      "eval_loss": 0.24300608038902283,
+      "eval_runtime": 1.5063,
+      "eval_samples_per_second": 1459.248,
+      "eval_steps_per_second": 23.236,
+      "step": 365000
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 6.36864494369016e-05,
+      "loss": 0.259,
+      "step": 365500
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 6.331446944282534e-05,
+      "loss": 0.2587,
+      "step": 366000
+    },
+    {
+      "epoch": 8.53,
+      "eval_loss": 0.24051520228385925,
+      "eval_runtime": 1.4527,
+      "eval_samples_per_second": 1513.01,
+      "eval_steps_per_second": 24.093,
+      "step": 366000
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 6.294349210702188e-05,
+      "loss": 0.2589,
+      "step": 366500
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 6.257352148643998e-05,
+      "loss": 0.2591,
+      "step": 367000
+    },
+    {
+      "epoch": 8.55,
+      "eval_loss": 0.2446797788143158,
+      "eval_runtime": 1.4184,
+      "eval_samples_per_second": 1549.604,
+      "eval_steps_per_second": 24.675,
+      "step": 367000
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 6.220456162701908e-05,
+      "loss": 0.2589,
+      "step": 367500
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 6.183661656364515e-05,
+      "loss": 0.2586,
+      "step": 368000
+    },
+    {
+      "epoch": 8.57,
+      "eval_loss": 0.24278691411018372,
+      "eval_runtime": 1.4508,
+      "eval_samples_per_second": 1515.055,
+      "eval_steps_per_second": 24.125,
+      "step": 368000
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 6.146969032010631e-05,
+      "loss": 0.2589,
+      "step": 368500
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 6.110378690904928e-05,
+      "loss": 0.259,
+      "step": 369000
+    },
+    {
+      "epoch": 8.6,
+      "eval_loss": 0.24149444699287415,
+      "eval_runtime": 1.4509,
+      "eval_samples_per_second": 1514.912,
+      "eval_steps_per_second": 24.123,
+      "step": 369000
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.073891033193507e-05,
+      "loss": 0.259,
+      "step": 369500
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.037506457899553e-05,
+      "loss": 0.2588,
+      "step": 370000
+    },
+    {
+      "epoch": 8.62,
+      "eval_loss": 0.23884567618370056,
+      "eval_runtime": 1.4545,
+      "eval_samples_per_second": 1511.147,
+      "eval_steps_per_second": 24.063,
+      "step": 370000
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.0012253629189544e-05,
+      "loss": 0.2584,
+      "step": 370500
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.965048145015944e-05,
+      "loss": 0.2585,
+      "step": 371000
+    },
+    {
+      "epoch": 8.64,
+      "eval_loss": 0.24364076554775238,
+      "eval_runtime": 1.4292,
+      "eval_samples_per_second": 1537.961,
+      "eval_steps_per_second": 24.49,
+      "step": 371000
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.928975199818785e-05,
+      "loss": 0.2583,
+      "step": 371500
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.893006921815428e-05,
+      "loss": 0.2582,
+      "step": 372000
+    },
+    {
+      "epoch": 8.67,
+      "eval_loss": 0.24061721563339233,
+      "eval_runtime": 1.4626,
+      "eval_samples_per_second": 1502.828,
+      "eval_steps_per_second": 23.93,
+      "step": 372000
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 5.857143704349198e-05,
+      "loss": 0.2584,
+      "step": 372500
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.8213859396144986e-05,
+      "loss": 0.2583,
+      "step": 373000
+    },
+    {
+      "epoch": 8.69,
+      "eval_loss": 0.24275849759578705,
+      "eval_runtime": 1.4296,
+      "eval_samples_per_second": 1537.531,
+      "eval_steps_per_second": 24.483,
+      "step": 373000
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.785734018652507e-05,
+      "loss": 0.2584,
+      "step": 373500
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 5.750188331346927e-05,
+      "loss": 0.2585,
+      "step": 374000
+    },
+    {
+      "epoch": 8.71,
+      "eval_loss": 0.24333250522613525,
+      "eval_runtime": 1.4555,
+      "eval_samples_per_second": 1510.18,
+      "eval_steps_per_second": 24.047,
+      "step": 374000
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.714749266419695e-05,
+      "loss": 0.2584,
+      "step": 374500
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 5.6794172114267566e-05,
+      "loss": 0.2578,
+      "step": 375000
+    },
+    {
+      "epoch": 8.74,
+      "eval_loss": 0.24110642075538635,
+      "eval_runtime": 1.4471,
+      "eval_samples_per_second": 1518.91,
+      "eval_steps_per_second": 24.186,
+      "step": 375000
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.6441925527537914e-05,
+      "loss": 0.2578,
+      "step": 375500
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 5.60907567561203e-05,
+      "loss": 0.2582,
+      "step": 376000
+    },
+    {
+      "epoch": 8.76,
+      "eval_loss": 0.242658793926239,
+      "eval_runtime": 1.4703,
+      "eval_samples_per_second": 1494.982,
+      "eval_steps_per_second": 23.805,
+      "step": 376000
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.574066964034012e-05,
+      "loss": 0.2581,
+      "step": 376500
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.539166800869402e-05,
+      "loss": 0.258,
+      "step": 377000
+    },
+    {
+      "epoch": 8.78,
+      "eval_loss": 0.2416759431362152,
+      "eval_runtime": 1.4756,
+      "eval_samples_per_second": 1489.53,
+      "eval_steps_per_second": 23.719,
+      "step": 377000
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.5043755677807955e-05,
+      "loss": 0.2578,
+      "step": 377500
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.4696936452395344e-05,
+      "loss": 0.2576,
+      "step": 378000
+    },
+    {
+      "epoch": 8.81,
+      "eval_loss": 0.23985832929611206,
+      "eval_runtime": 1.4569,
+      "eval_samples_per_second": 1508.675,
+      "eval_steps_per_second": 24.023,
+      "step": 378000
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.435121412521576e-05,
+      "loss": 0.2579,
+      "step": 378500
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.400659247703307e-05,
+      "loss": 0.2574,
+      "step": 379000
+    },
+    {
+      "epoch": 8.83,
+      "eval_loss": 0.24152863025665283,
+      "eval_runtime": 1.4464,
+      "eval_samples_per_second": 1519.59,
+      "eval_steps_per_second": 24.197,
+      "step": 379000
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.36630752765745e-05,
+      "loss": 0.2576,
+      "step": 379500
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.3320666280489146e-05,
+      "loss": 0.2579,
+      "step": 380000
+    },
+    {
+      "epoch": 8.85,
+      "eval_loss": 0.24166275560855865,
+      "eval_runtime": 1.4454,
+      "eval_samples_per_second": 1520.737,
+      "eval_steps_per_second": 24.216,
+      "step": 380000
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.2979369233306834e-05,
+      "loss": 0.2576,
+      "step": 380500
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.26391878673975e-05,
+      "loss": 0.2573,
+      "step": 381000
+    },
+    {
+      "epoch": 8.88,
+      "eval_loss": 0.2407015562057495,
+      "eval_runtime": 1.4434,
+      "eval_samples_per_second": 1522.761,
+      "eval_steps_per_second": 24.248,
+      "step": 381000
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.230012590292987e-05,
+      "loss": 0.2575,
+      "step": 381500
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.1962187047831517e-05,
+      "loss": 0.2575,
+      "step": 382000
+    },
+    {
+      "epoch": 8.9,
+      "eval_loss": 0.2418326437473297,
+      "eval_runtime": 1.4362,
+      "eval_samples_per_second": 1530.452,
+      "eval_steps_per_second": 24.37,
+      "step": 382000
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.162537499774743e-05,
+      "loss": 0.2574,
+      "step": 382500
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.128969343600032e-05,
+      "loss": 0.2573,
+      "step": 383000
+    },
+    {
+      "epoch": 8.92,
+      "eval_loss": 0.2397920787334442,
+      "eval_runtime": 1.4505,
+      "eval_samples_per_second": 1515.368,
+      "eval_steps_per_second": 24.13,
+      "step": 383000
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.09551460335499e-05,
+      "loss": 0.2571,
+      "step": 383500
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.062173644895296e-05,
+      "loss": 0.2573,
+      "step": 384000
+    },
+    {
+      "epoch": 8.95,
+      "eval_loss": 0.23995496332645416,
+      "eval_runtime": 1.4711,
+      "eval_samples_per_second": 1494.15,
+      "eval_steps_per_second": 23.792,
+      "step": 384000
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.0289468328323434e-05,
+      "loss": 0.2569,
+      "step": 384500
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.995834530529208e-05,
+      "loss": 0.2569,
+      "step": 385000
+    },
+    {
+      "epoch": 8.97,
+      "eval_loss": 0.23903849720954895,
+      "eval_runtime": 1.4433,
+      "eval_samples_per_second": 1522.946,
+      "eval_steps_per_second": 24.251,
+      "step": 385000
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.9628371000967394e-05,
+      "loss": 0.2573,
+      "step": 385500
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.929954902389534e-05,
+      "loss": 0.2572,
+      "step": 386000
+    },
+    {
+      "epoch": 8.99,
+      "eval_loss": 0.23725509643554688,
+      "eval_runtime": 1.4864,
+      "eval_samples_per_second": 1478.786,
+      "eval_steps_per_second": 23.548,
+      "step": 386000
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.897188297002046e-05,
+      "loss": 0.2571,
+      "step": 386500
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.8645376422646226e-05,
+      "loss": 0.2566,
+      "step": 387000
+    },
+    {
+      "epoch": 9.02,
+      "eval_loss": 0.24057132005691528,
+      "eval_runtime": 1.4509,
+      "eval_samples_per_second": 1514.911,
+      "eval_steps_per_second": 24.123,
+      "step": 387000
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.832003295239591e-05,
+      "loss": 0.2568,
+      "step": 387500
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.7995856117173624e-05,
+      "loss": 0.2565,
+      "step": 388000
+    },
+    {
+      "epoch": 9.04,
+      "eval_loss": 0.23878733813762665,
+      "eval_runtime": 1.3617,
+      "eval_samples_per_second": 1614.217,
+      "eval_steps_per_second": 25.704,
+      "step": 388000
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.767284946212521e-05,
+      "loss": 0.2567,
+      "step": 388500
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.735101651959977e-05,
+      "loss": 0.2566,
+      "step": 389000
+    },
+    {
+      "epoch": 9.06,
+      "eval_loss": 0.23982606828212738,
+      "eval_runtime": 1.4089,
+      "eval_samples_per_second": 1560.06,
+      "eval_steps_per_second": 24.842,
+      "step": 389000
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.7030360809110754e-05,
+      "loss": 0.2564,
+      "step": 389500
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.6710885837297726e-05,
+      "loss": 0.2564,
+      "step": 390000
+    },
+    {
+      "epoch": 9.09,
+      "eval_loss": 0.24110108613967896,
+      "eval_runtime": 1.4757,
+      "eval_samples_per_second": 1489.497,
+      "eval_steps_per_second": 23.718,
+      "step": 390000
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.639259509788768e-05,
+      "loss": 0.2564,
+      "step": 390500
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.60754920716572e-05,
+      "loss": 0.2566,
+      "step": 391000
+    },
+    {
+      "epoch": 9.11,
+      "eval_loss": 0.23735950887203217,
+      "eval_runtime": 1.4271,
+      "eval_samples_per_second": 1540.156,
+      "eval_steps_per_second": 24.525,
+      "step": 391000
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.5759580226394167e-05,
+      "loss": 0.2566,
+      "step": 391500
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.544486301685993e-05,
+      "loss": 0.2564,
+      "step": 392000
+    },
+    {
+      "epoch": 9.13,
+      "eval_loss": 0.2416989952325821,
+      "eval_runtime": 1.449,
+      "eval_samples_per_second": 1516.918,
+      "eval_steps_per_second": 24.155,
+      "step": 392000
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.5131343884751484e-05,
+      "loss": 0.2562,
+      "step": 392500
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.4819026258663774e-05,
+      "loss": 0.2565,
+      "step": 393000
+    },
+    {
+      "epoch": 9.16,
+      "eval_loss": 0.24084173142910004,
+      "eval_runtime": 1.456,
+      "eval_samples_per_second": 1509.619,
+      "eval_steps_per_second": 24.039,
+      "step": 393000
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.450791355405234e-05,
+      "loss": 0.2565,
+      "step": 393500
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.419800917319588e-05,
+      "loss": 0.2566,
+      "step": 394000
+    },
+    {
+      "epoch": 9.18,
+      "eval_loss": 0.23908619582653046,
+      "eval_runtime": 1.4522,
+      "eval_samples_per_second": 1513.612,
+      "eval_steps_per_second": 24.102,
+      "step": 394000
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.3889316505159056e-05,
+      "loss": 0.2564,
+      "step": 394500
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.3581838925755465e-05,
+      "loss": 0.2561,
+      "step": 395000
+    },
+    {
+      "epoch": 9.2,
+      "eval_loss": 0.23751436173915863,
+      "eval_runtime": 1.4847,
+      "eval_samples_per_second": 1480.475,
+      "eval_steps_per_second": 23.574,
+      "step": 395000
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.327557979751057e-05,
+      "loss": 0.2562,
+      "step": 395500
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.297054246962517e-05,
+      "loss": 0.2562,
+      "step": 396000
+    },
+    {
+      "epoch": 9.23,
+      "eval_loss": 0.23929938673973083,
+      "eval_runtime": 1.4334,
+      "eval_samples_per_second": 1533.416,
+      "eval_steps_per_second": 24.417,
+      "step": 396000
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.266673027793864e-05,
+      "loss": 0.2562,
+      "step": 396500
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.236414654489242e-05,
+      "loss": 0.256,
+      "step": 397000
+    },
+    {
+      "epoch": 9.25,
+      "eval_loss": 0.24186982214450836,
+      "eval_runtime": 1.4512,
+      "eval_samples_per_second": 1514.618,
+      "eval_steps_per_second": 24.118,
+      "step": 397000
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.206279457949371e-05,
+      "loss": 0.2558,
+      "step": 397500
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.1762677677279335e-05,
+      "loss": 0.2561,
+      "step": 398000
+    },
+    {
+      "epoch": 9.27,
+      "eval_loss": 0.24197684228420258,
+      "eval_runtime": 1.4439,
+      "eval_samples_per_second": 1522.288,
+      "eval_steps_per_second": 24.24,
+      "step": 398000
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.146379912027964e-05,
+      "loss": 0.2561,
+      "step": 398500
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.1166162176982664e-05,
+      "loss": 0.2554,
+      "step": 399000
+    },
+    {
+      "epoch": 9.3,
+      "eval_loss": 0.24182839691638947,
+      "eval_runtime": 1.4359,
+      "eval_samples_per_second": 1530.716,
+      "eval_steps_per_second": 24.374,
+      "step": 399000
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.086977010229838e-05,
+      "loss": 0.2557,
+      "step": 399500
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.057462613752294e-05,
+      "loss": 0.2559,
+      "step": 400000
+    },
+    {
+      "epoch": 9.32,
+      "eval_loss": 0.23793195188045502,
+      "eval_runtime": 1.413,
+      "eval_samples_per_second": 1555.592,
+      "eval_steps_per_second": 24.771,
+      "step": 400000
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.0280733510303475e-05,
+      "loss": 0.2558,
+      "step": 400500
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.9988095434602716e-05,
+      "loss": 0.2555,
+      "step": 401000
+    },
+    {
+      "epoch": 9.34,
+      "eval_loss": 0.2393871247768402,
+      "eval_runtime": 1.4557,
+      "eval_samples_per_second": 1509.958,
+      "eval_steps_per_second": 24.044,
+      "step": 401000
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.9696715110663726e-05,
+      "loss": 0.2557,
+      "step": 401500
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.9406595724975116e-05,
+      "loss": 0.2553,
+      "step": 402000
+    },
+    {
+      "epoch": 9.37,
+      "eval_loss": 0.2388046830892563,
+      "eval_runtime": 1.444,
+      "eval_samples_per_second": 1522.119,
+      "eval_steps_per_second": 24.238,
+      "step": 402000
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.9117740450235914e-05,
+      "loss": 0.2555,
+      "step": 402500
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.8830152445321163e-05,
+      "loss": 0.2553,
+      "step": 403000
+    },
+    {
+      "epoch": 9.39,
+      "eval_loss": 0.23731090128421783,
+      "eval_runtime": 1.5043,
+      "eval_samples_per_second": 1461.141,
+      "eval_steps_per_second": 23.267,
+      "step": 403000
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.854383485524724e-05,
+      "loss": 0.2553,
+      "step": 403500
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.8258790811137425e-05,
+      "loss": 0.2553,
+      "step": 404000
+    },
+    {
+      "epoch": 9.41,
+      "eval_loss": 0.2364571988582611,
+      "eval_runtime": 1.4662,
+      "eval_samples_per_second": 1499.102,
+      "eval_steps_per_second": 23.871,
+      "step": 404000
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 3.7975023430187676e-05,
+      "loss": 0.2552,
+      "step": 404500
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.7692535815632624e-05,
+      "loss": 0.2555,
+      "step": 405000
+    },
+    {
+      "epoch": 9.44,
+      "eval_loss": 0.23983800411224365,
+      "eval_runtime": 1.4385,
+      "eval_samples_per_second": 1527.964,
+      "eval_steps_per_second": 24.331,
+      "step": 405000
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 3.741133105671159e-05,
+      "loss": 0.2554,
+      "step": 405500
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 3.713141222863474e-05,
+      "loss": 0.2553,
+      "step": 406000
+    },
+    {
+      "epoch": 9.46,
+      "eval_loss": 0.23924601078033447,
+      "eval_runtime": 1.4347,
+      "eval_samples_per_second": 1532.043,
+      "eval_steps_per_second": 24.396,
+      "step": 406000
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.6852782392549584e-05,
+      "loss": 0.2556,
+      "step": 406500
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 3.657544459550729e-05,
+      "loss": 0.255,
+      "step": 407000
+    },
+    {
+      "epoch": 9.48,
+      "eval_loss": 0.23706746101379395,
+      "eval_runtime": 1.4741,
+      "eval_samples_per_second": 1491.033,
+      "eval_steps_per_second": 23.743,
+      "step": 407000
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 3.6299401870429606e-05,
+      "loss": 0.2549,
+      "step": 407500
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 3.6024657236075546e-05,
+      "loss": 0.2548,
+      "step": 408000
+    },
+    {
+      "epoch": 9.51,
+      "eval_loss": 0.23815912008285522,
+      "eval_runtime": 1.4566,
+      "eval_samples_per_second": 1508.999,
+      "eval_steps_per_second": 24.029,
+      "step": 408000
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 3.575121369700841e-05,
+      "loss": 0.2549,
+      "step": 408500
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 3.5479074243562995e-05,
+      "loss": 0.2548,
+      "step": 409000
+    },
+    {
+      "epoch": 9.53,
+      "eval_loss": 0.23738771677017212,
+      "eval_runtime": 1.4607,
+      "eval_samples_per_second": 1504.709,
+      "eval_steps_per_second": 23.96,
+      "step": 409000
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.5208241851812644e-05,
+      "loss": 0.2549,
+      "step": 409500
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 3.493871948353709e-05,
+      "loss": 0.2548,
+      "step": 410000
+    },
+    {
+      "epoch": 9.55,
+      "eval_loss": 0.23903900384902954,
+      "eval_runtime": 1.4733,
+      "eval_samples_per_second": 1491.858,
+      "eval_steps_per_second": 23.756,
+      "step": 410000
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 3.4670510086189736e-05,
+      "loss": 0.2551,
+      "step": 410500
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 3.440361659286563e-05,
+      "loss": 0.2547,
+      "step": 411000
+    },
+    {
+      "epoch": 9.58,
+      "eval_loss": 0.23947188258171082,
+      "eval_runtime": 1.4381,
+      "eval_samples_per_second": 1528.365,
+      "eval_steps_per_second": 24.337,
+      "step": 411000
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.413804192226918e-05,
+      "loss": 0.2547,
+      "step": 411500
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.387378897868246e-05,
+      "loss": 0.2544,
+      "step": 412000
+    },
+    {
+      "epoch": 9.6,
+      "eval_loss": 0.23671407997608185,
+      "eval_runtime": 1.4569,
+      "eval_samples_per_second": 1508.674,
+      "eval_steps_per_second": 24.023,
+      "step": 412000
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.361086065193336e-05,
+      "loss": 0.2549,
+      "step": 412500
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.334925981736389e-05,
+      "loss": 0.2548,
+      "step": 413000
+    },
+    {
+      "epoch": 9.62,
+      "eval_loss": 0.23614807426929474,
+      "eval_runtime": 1.4517,
+      "eval_samples_per_second": 1514.052,
+      "eval_steps_per_second": 24.109,
+      "step": 413000
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.3088989335798925e-05,
+      "loss": 0.2549,
+      "step": 413500
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.283005205351467e-05,
+      "loss": 0.2545,
+      "step": 414000
+    },
+    {
+      "epoch": 9.65,
+      "eval_loss": 0.2389170378446579,
+      "eval_runtime": 1.4416,
+      "eval_samples_per_second": 1524.726,
+      "eval_steps_per_second": 24.279,
+      "step": 414000
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.2572450802207845e-05,
+      "loss": 0.2546,
+      "step": 414500
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.2316188398964344e-05,
+      "loss": 0.2547,
+      "step": 415000
+    },
+    {
+      "epoch": 9.67,
+      "eval_loss": 0.24014273285865784,
+      "eval_runtime": 1.4644,
+      "eval_samples_per_second": 1500.945,
+      "eval_steps_per_second": 23.9,
+      "step": 415000
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.206126764622888e-05,
+      "loss": 0.2543,
+      "step": 415500
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 3.180769133177392e-05,
+      "loss": 0.2545,
+      "step": 416000
+    },
+    {
+      "epoch": 9.69,
+      "eval_loss": 0.2388986498117447,
+      "eval_runtime": 1.5029,
+      "eval_samples_per_second": 1462.487,
+      "eval_steps_per_second": 23.288,
+      "step": 416000
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 3.155546222866939e-05,
+      "loss": 0.2543,
+      "step": 416500
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.130458309525239e-05,
+      "loss": 0.2543,
+      "step": 417000
+    },
+    {
+      "epoch": 9.72,
+      "eval_loss": 0.2375379502773285,
+      "eval_runtime": 1.4847,
+      "eval_samples_per_second": 1480.402,
+      "eval_steps_per_second": 23.573,
+      "step": 417000
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.1055056675096826e-05,
+      "loss": 0.2544,
+      "step": 417500
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.0806885696983816e-05,
+      "loss": 0.2543,
+      "step": 418000
+    },
+    {
+      "epoch": 9.74,
+      "eval_loss": 0.23962582647800446,
+      "eval_runtime": 1.4976,
+      "eval_samples_per_second": 1467.703,
+      "eval_steps_per_second": 23.371,
+      "step": 418000
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.056007287487128e-05,
+      "loss": 0.2542,
+      "step": 418500
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.0314620907864744e-05,
+      "loss": 0.2541,
+      "step": 419000
+    },
+    {
+      "epoch": 9.76,
+      "eval_loss": 0.23849867284297943,
+      "eval_runtime": 1.4273,
+      "eval_samples_per_second": 1539.928,
+      "eval_steps_per_second": 24.521,
+      "step": 419000
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 3.0070532480187637e-05,
+      "loss": 0.2541,
+      "step": 419500
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.9827810261151784e-05,
+      "loss": 0.2541,
+      "step": 420000
+    },
+    {
+      "epoch": 9.79,
+      "eval_loss": 0.2357112467288971,
+      "eval_runtime": 1.4438,
+      "eval_samples_per_second": 1522.319,
+      "eval_steps_per_second": 24.241,
+      "step": 420000
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.9586456905128618e-05,
+      "loss": 0.2539,
+      "step": 420500
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 2.9346475051519687e-05,
+      "loss": 0.2539,
+      "step": 421000
+    },
+    {
+      "epoch": 9.81,
+      "eval_loss": 0.23778527975082397,
+      "eval_runtime": 1.4672,
+      "eval_samples_per_second": 1498.116,
+      "eval_steps_per_second": 23.855,
+      "step": 421000
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 2.910786732472815e-05,
+      "loss": 0.2538,
+      "step": 421500
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 2.887063633412981e-05,
+      "loss": 0.254,
+      "step": 422000
+    },
+    {
+      "epoch": 9.83,
+      "eval_loss": 0.23553605377674103,
+      "eval_runtime": 1.4917,
+      "eval_samples_per_second": 1473.465,
+      "eval_steps_per_second": 23.463,
+      "step": 422000
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 2.863478467404478e-05,
+      "loss": 0.2538,
+      "step": 422500
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 2.8400314923709112e-05,
+      "loss": 0.2537,
+      "step": 423000
+    },
+    {
+      "epoch": 9.86,
+      "eval_loss": 0.23579563200473785,
+      "eval_runtime": 1.4277,
+      "eval_samples_per_second": 1539.579,
+      "eval_steps_per_second": 24.516,
+      "step": 423000
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 2.816722964724636e-05,
+      "loss": 0.2537,
+      "step": 423500
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 2.793553139363981e-05,
+      "loss": 0.2536,
+      "step": 424000
+    },
+    {
+      "epoch": 9.88,
+      "eval_loss": 0.23610466718673706,
+      "eval_runtime": 1.4604,
+      "eval_samples_per_second": 1505.035,
+      "eval_steps_per_second": 23.966,
+      "step": 424000
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 2.7705222696704366e-05,
+      "loss": 0.2536,
+      "step": 424500
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 2.7476306075059096e-05,
+      "loss": 0.2534,
+      "step": 425000
+    },
+    {
+      "epoch": 9.9,
+      "eval_loss": 0.2384994477033615,
+      "eval_runtime": 1.4421,
+      "eval_samples_per_second": 1524.185,
+      "eval_steps_per_second": 24.27,
+      "step": 425000
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.7248784032099478e-05,
+      "loss": 0.2536,
+      "step": 425500
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.7022659055970144e-05,
+      "loss": 0.2534,
+      "step": 426000
+    },
+    {
+      "epoch": 9.93,
+      "eval_loss": 0.23720860481262207,
+      "eval_runtime": 1.4451,
+      "eval_samples_per_second": 1521.023,
+      "eval_steps_per_second": 24.22,
+      "step": 426000
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.6797933619537604e-05,
+      "loss": 0.2534,
+      "step": 426500
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.6574610180363166e-05,
+      "loss": 0.2535,
+      "step": 427000
+    },
+    {
+      "epoch": 9.95,
+      "eval_loss": 0.23618370294570923,
+      "eval_runtime": 1.4698,
+      "eval_samples_per_second": 1495.473,
+      "eval_steps_per_second": 23.813,
+      "step": 427000
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.6352691180676286e-05,
+      "loss": 0.2532,
+      "step": 427500
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 2.6132179047347505e-05,
+      "loss": 0.2535,
+      "step": 428000
+    },
+    {
+      "epoch": 9.97,
+      "eval_loss": 0.23612357676029205,
+      "eval_runtime": 1.4655,
+      "eval_samples_per_second": 1499.84,
+      "eval_steps_per_second": 23.883,
+      "step": 428000
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 2.5913076191862238e-05,
+      "loss": 0.2534,
+      "step": 428500
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.5695385010294165e-05,
+      "loss": 0.2532,
+      "step": 429000
+    },
+    {
+      "epoch": 10.0,
+      "eval_loss": 0.23688232898712158,
+      "eval_runtime": 1.4807,
+      "eval_samples_per_second": 1484.441,
+      "eval_steps_per_second": 23.638,
+      "step": 429000
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 2.5479107883279144e-05,
+      "loss": 0.2533,
+      "step": 429500
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 2.5264247175989292e-05,
+      "loss": 0.2533,
+      "step": 430000
+    },
+    {
+      "epoch": 10.02,
+      "eval_loss": 0.2365807592868805,
+      "eval_runtime": 1.4314,
+      "eval_samples_per_second": 1535.576,
+      "eval_steps_per_second": 24.452,
+      "step": 430000
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 2.5050805238106804e-05,
+      "loss": 0.253,
+      "step": 430500
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 2.4838784403798542e-05,
+      "loss": 0.2529,
+      "step": 431000
+    },
+    {
+      "epoch": 10.04,
+      "eval_loss": 0.23809103667736053,
+      "eval_runtime": 1.4751,
+      "eval_samples_per_second": 1490.054,
+      "eval_steps_per_second": 23.727,
+      "step": 431000
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 2.4628186991690346e-05,
+      "loss": 0.2528,
+      "step": 431500
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 2.4419015304841797e-05,
+      "loss": 0.2532,
+      "step": 432000
+    },
+    {
+      "epoch": 10.07,
+      "eval_loss": 0.23703138530254364,
+      "eval_runtime": 1.4419,
+      "eval_samples_per_second": 1524.361,
+      "eval_steps_per_second": 24.273,
+      "step": 432000
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 2.4211271630720957e-05,
+      "loss": 0.2532,
+      "step": 432500
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 2.4004958241179347e-05,
+      "loss": 0.253,
+      "step": 433000
+    },
+    {
+      "epoch": 10.09,
+      "eval_loss": 0.2378602772951126,
+      "eval_runtime": 1.476,
+      "eval_samples_per_second": 1489.124,
+      "eval_steps_per_second": 23.712,
+      "step": 433000
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 2.3800077392427193e-05,
+      "loss": 0.2533,
+      "step": 433500
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 2.3596631325008536e-05,
+      "loss": 0.2531,
+      "step": 434000
+    },
+    {
+      "epoch": 10.11,
+      "eval_loss": 0.23943325877189636,
+      "eval_runtime": 1.4371,
+      "eval_samples_per_second": 1529.446,
+      "eval_steps_per_second": 24.354,
+      "step": 434000
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 2.3394622263777042e-05,
+      "loss": 0.2529,
+      "step": 434500
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 2.3194052417871433e-05,
+      "loss": 0.2528,
+      "step": 435000
+    },
+    {
+      "epoch": 10.14,
+      "eval_loss": 0.2356664538383484,
+      "eval_runtime": 1.4719,
+      "eval_samples_per_second": 1493.299,
+      "eval_steps_per_second": 23.779,
+      "step": 435000
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 2.2994923980691425e-05,
+      "loss": 0.253,
+      "step": 435500
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 2.279723912987365e-05,
+      "loss": 0.253,
+      "step": 436000
+    },
+    {
+      "epoch": 10.16,
+      "eval_loss": 0.23836100101470947,
+      "eval_runtime": 1.4427,
+      "eval_samples_per_second": 1523.571,
+      "eval_steps_per_second": 24.261,
+      "step": 436000
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 2.2601000027268006e-05,
+      "loss": 0.2526,
+      "step": 436500
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 2.2406208818913857e-05,
+      "loss": 0.2525,
+      "step": 437000
+    },
+    {
+      "epoch": 10.18,
+      "eval_loss": 0.23866206407546997,
+      "eval_runtime": 1.4308,
+      "eval_samples_per_second": 1536.237,
+      "eval_steps_per_second": 24.462,
+      "step": 437000
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 2.221286763501666e-05,
+      "loss": 0.2527,
+      "step": 437500
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 2.2020978589924673e-05,
+      "loss": 0.2526,
+      "step": 438000
+    },
+    {
+      "epoch": 10.21,
+      "eval_loss": 0.23435133695602417,
+      "eval_runtime": 1.417,
+      "eval_samples_per_second": 1551.208,
+      "eval_steps_per_second": 24.701,
+      "step": 438000
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 2.1830543782105647e-05,
+      "loss": 0.2528,
+      "step": 438500
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 2.1641565294124206e-05,
+      "loss": 0.2527,
+      "step": 439000
+    },
+    {
+      "epoch": 10.23,
+      "eval_loss": 0.23615118861198425,
+      "eval_runtime": 1.4413,
+      "eval_samples_per_second": 1524.993,
+      "eval_steps_per_second": 24.283,
+      "step": 439000
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 2.1454045192618794e-05,
+      "loss": 0.2525,
+      "step": 439500
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 2.1267985528279212e-05,
+      "loss": 0.2528,
+      "step": 440000
+    },
+    {
+      "epoch": 10.25,
+      "eval_loss": 0.23707140982151031,
+      "eval_runtime": 1.4358,
+      "eval_samples_per_second": 1530.896,
+      "eval_steps_per_second": 24.377,
+      "step": 440000
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 2.1083388335824145e-05,
+      "loss": 0.2526,
+      "step": 440500
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 2.0900255633978873e-05,
+      "loss": 0.2525,
+      "step": 441000
+    },
+    {
+      "epoch": 10.28,
+      "eval_loss": 0.23552300035953522,
+      "eval_runtime": 1.4323,
+      "eval_samples_per_second": 1534.608,
+      "eval_steps_per_second": 24.436,
+      "step": 441000
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.0718589425453314e-05,
+      "loss": 0.2527,
+      "step": 441500
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 2.0538391696920015e-05,
+      "loss": 0.2521,
+      "step": 442000
+    },
+    {
+      "epoch": 10.3,
+      "eval_loss": 0.23650765419006348,
+      "eval_runtime": 1.4235,
+      "eval_samples_per_second": 1544.133,
+      "eval_steps_per_second": 24.588,
+      "step": 442000
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.035966441899249e-05,
+      "loss": 0.2523,
+      "step": 442500
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 2.0182409546203555e-05,
+      "loss": 0.2523,
+      "step": 443000
+    },
+    {
+      "epoch": 10.32,
+      "eval_loss": 0.23561330139636993,
+      "eval_runtime": 1.4199,
+      "eval_samples_per_second": 1548.043,
+      "eval_steps_per_second": 24.65,
+      "step": 443000
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.000662901698415e-05,
+      "loss": 0.2521,
+      "step": 443500
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.983232475364195e-05,
+      "loss": 0.2523,
+      "step": 444000
+    },
+    {
+      "epoch": 10.35,
+      "eval_loss": 0.23662236332893372,
+      "eval_runtime": 1.4145,
+      "eval_samples_per_second": 1553.873,
+      "eval_steps_per_second": 24.743,
+      "step": 444000
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.9659498662340474e-05,
+      "loss": 0.2524,
+      "step": 444500
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.948815263307819e-05,
+      "loss": 0.2519,
+      "step": 445000
+    },
+    {
+      "epoch": 10.37,
+      "eval_loss": 0.23521380126476288,
+      "eval_runtime": 1.4054,
+      "eval_samples_per_second": 1563.94,
+      "eval_steps_per_second": 24.903,
+      "step": 445000
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.9318288539667765e-05,
+      "loss": 0.2524,
+      "step": 445500
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.914990823971574e-05,
+      "loss": 0.2523,
+      "step": 446000
+    },
+    {
+      "epoch": 10.39,
+      "eval_loss": 0.2357960045337677,
+      "eval_runtime": 1.4368,
+      "eval_samples_per_second": 1529.784,
+      "eval_steps_per_second": 24.36,
+      "step": 446000
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.8983013574602096e-05,
+      "loss": 0.2521,
+      "step": 446500
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.8817606369460156e-05,
+      "loss": 0.2521,
+      "step": 447000
+    },
+    {
+      "epoch": 10.41,
+      "eval_loss": 0.23412732779979706,
+      "eval_runtime": 1.4282,
+      "eval_samples_per_second": 1538.96,
+      "eval_steps_per_second": 24.506,
+      "step": 447000
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.865368843315663e-05,
+      "loss": 0.252,
+      "step": 447500
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.8491261558271762e-05,
+      "loss": 0.2519,
+      "step": 448000
+    },
+    {
+      "epoch": 10.44,
+      "eval_loss": 0.23749284446239471,
+      "eval_runtime": 1.4343,
+      "eval_samples_per_second": 1532.461,
+      "eval_steps_per_second": 24.402,
+      "step": 448000
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.833032752107986e-05,
+      "loss": 0.2518,
+      "step": 448500
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.817088808152978e-05,
+      "loss": 0.2519,
+      "step": 449000
+    },
+    {
+      "epoch": 10.46,
+      "eval_loss": 0.2349672168493271,
+      "eval_runtime": 1.4127,
+      "eval_samples_per_second": 1555.927,
+      "eval_steps_per_second": 24.776,
+      "step": 449000
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.801294498322569e-05,
+      "loss": 0.2517,
+      "step": 449500
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.7856499953407978e-05,
+      "loss": 0.2519,
+      "step": 450000
+    },
+    {
+      "epoch": 10.48,
+      "eval_loss": 0.2356027066707611,
+      "eval_runtime": 1.4379,
+      "eval_samples_per_second": 1528.604,
+      "eval_steps_per_second": 24.341,
+      "step": 450000
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.770155470293445e-05,
+      "loss": 0.2519,
+      "step": 450500
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.7548110926261522e-05,
+      "loss": 0.2518,
+      "step": 451000
+    },
+    {
+      "epoch": 10.51,
+      "eval_loss": 0.23447871208190918,
+      "eval_runtime": 1.4307,
+      "eval_samples_per_second": 1536.276,
+      "eval_steps_per_second": 24.463,
+      "step": 451000
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.7396170301425777e-05,
+      "loss": 0.2518,
+      "step": 451500
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.7245734490025544e-05,
+      "loss": 0.252,
+      "step": 452000
+    },
+    {
+      "epoch": 10.53,
+      "eval_loss": 0.23647595942020416,
+      "eval_runtime": 1.4161,
+      "eval_samples_per_second": 1552.16,
+      "eval_steps_per_second": 24.716,
+      "step": 452000
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.7096805137202738e-05,
+      "loss": 0.2518,
+      "step": 452500
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.6949383871624917e-05,
+      "loss": 0.2517,
+      "step": 453000
+    },
+    {
+      "epoch": 10.55,
+      "eval_loss": 0.23766835033893585,
+      "eval_runtime": 1.4533,
+      "eval_samples_per_second": 1512.402,
+      "eval_steps_per_second": 24.083,
+      "step": 453000
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.6803472305467368e-05,
+      "loss": 0.2515,
+      "step": 453500
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.665907203439568e-05,
+      "loss": 0.2516,
+      "step": 454000
+    },
+    {
+      "epoch": 10.58,
+      "eval_loss": 0.23621916770935059,
+      "eval_runtime": 1.476,
+      "eval_samples_per_second": 1489.127,
+      "eval_steps_per_second": 23.712,
+      "step": 454000
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.6516184637548058e-05,
+      "loss": 0.2515,
+      "step": 454500
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.6374811677518142e-05,
+      "loss": 0.2517,
+      "step": 455000
+    },
+    {
+      "epoch": 10.6,
+      "eval_loss": 0.23542185127735138,
+      "eval_runtime": 1.4176,
+      "eval_samples_per_second": 1550.543,
+      "eval_steps_per_second": 24.69,
+      "step": 455000
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.6234954700338025e-05,
+      "loss": 0.2516,
+      "step": 455500
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.6096615235461148e-05,
+      "loss": 0.2515,
+      "step": 456000
+    },
+    {
+      "epoch": 10.62,
+      "eval_loss": 0.2339724451303482,
+      "eval_runtime": 1.3908,
+      "eval_samples_per_second": 1580.393,
+      "eval_steps_per_second": 25.165,
+      "step": 456000
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.59597947957458e-05,
+      "loss": 0.2519,
+      "step": 456500
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.5824494877438344e-05,
+      "loss": 0.2515,
+      "step": 457000
+    },
+    {
+      "epoch": 10.65,
+      "eval_loss": 0.23481862246990204,
+      "eval_runtime": 1.4326,
+      "eval_samples_per_second": 1534.248,
+      "eval_steps_per_second": 24.431,
+      "step": 457000
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.569071696015702e-05,
+      "loss": 0.2513,
+      "step": 457500
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.555846250687569e-05,
+      "loss": 0.2517,
+      "step": 458000
+    },
+    {
+      "epoch": 10.67,
+      "eval_loss": 0.23701262474060059,
+      "eval_runtime": 1.4718,
+      "eval_samples_per_second": 1493.443,
+      "eval_steps_per_second": 23.781,
+      "step": 458000
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.542773296390789e-05,
+      "loss": 0.2518,
+      "step": 458500
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.5298529760890945e-05,
+      "loss": 0.2514,
+      "step": 459000
+    },
+    {
+      "epoch": 10.69,
+      "eval_loss": 0.23413069546222687,
+      "eval_runtime": 1.4496,
+      "eval_samples_per_second": 1516.277,
+      "eval_steps_per_second": 24.145,
+      "step": 459000
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.5170854310770376e-05,
+      "loss": 0.251,
+      "step": 459500
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.5044708009784457e-05,
+      "loss": 0.2514,
+      "step": 460000
+    },
+    {
+      "epoch": 10.72,
+      "eval_loss": 0.2361001968383789,
+      "eval_runtime": 1.4246,
+      "eval_samples_per_second": 1542.873,
+      "eval_steps_per_second": 24.568,
+      "step": 460000
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4920092237448903e-05,
+      "loss": 0.2514,
+      "step": 460500
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.4797008356541874e-05,
+      "loss": 0.2513,
+      "step": 461000
+    },
+    {
+      "epoch": 10.74,
+      "eval_loss": 0.2357269674539566,
+      "eval_runtime": 1.4097,
+      "eval_samples_per_second": 1559.248,
+      "eval_steps_per_second": 24.829,
+      "step": 461000
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.4675457713088947e-05,
+      "loss": 0.2513,
+      "step": 461500
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.4555441636348494e-05,
+      "loss": 0.2516,
+      "step": 462000
+    },
+    {
+      "epoch": 10.76,
+      "eval_loss": 0.2352001965045929,
+      "eval_runtime": 1.4391,
+      "eval_samples_per_second": 1527.384,
+      "eval_steps_per_second": 24.321,
+      "step": 462000
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.4436961438797095e-05,
+      "loss": 0.2512,
+      "step": 462500
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.4320018416115206e-05,
+      "loss": 0.2511,
+      "step": 463000
+    },
+    {
+      "epoch": 10.79,
+      "eval_loss": 0.23254527151584625,
+      "eval_runtime": 1.4508,
+      "eval_samples_per_second": 1514.977,
+      "eval_steps_per_second": 24.124,
+      "step": 463000
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.4204613847173003e-05,
+      "loss": 0.2513,
+      "step": 463500
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.4090748994016354e-05,
+      "loss": 0.2515,
+      "step": 464000
+    },
+    {
+      "epoch": 10.81,
+      "eval_loss": 0.23643817007541656,
+      "eval_runtime": 1.4271,
+      "eval_samples_per_second": 1540.18,
+      "eval_steps_per_second": 24.525,
+      "step": 464000
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.3978425101853049e-05,
+      "loss": 0.2513,
+      "step": 464500
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.3867643399039165e-05,
+      "loss": 0.2509,
+      "step": 465000
+    },
+    {
+      "epoch": 10.83,
+      "eval_loss": 0.2379205971956253,
+      "eval_runtime": 1.4615,
+      "eval_samples_per_second": 1503.93,
+      "eval_steps_per_second": 23.948,
+      "step": 465000
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.3758405097065648e-05,
+      "loss": 0.2515,
+      "step": 465500
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.3650711390545131e-05,
+      "loss": 0.2511,
+      "step": 466000
+    },
+    {
+      "epoch": 10.86,
+      "eval_loss": 0.2360716015100479,
+      "eval_runtime": 1.4528,
+      "eval_samples_per_second": 1512.979,
+      "eval_steps_per_second": 24.092,
+      "step": 466000
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.3544563457198657e-05,
+      "loss": 0.2514,
+      "step": 466500
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.343996245784307e-05,
+      "loss": 0.2513,
+      "step": 467000
+    },
+    {
+      "epoch": 10.88,
+      "eval_loss": 0.2349902093410492,
+      "eval_runtime": 1.4197,
+      "eval_samples_per_second": 1548.226,
+      "eval_steps_per_second": 24.653,
+      "step": 467000
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.3336909536378107e-05,
+      "loss": 0.2511,
+      "step": 467500
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.3235405819774022e-05,
+      "loss": 0.2509,
+      "step": 468000
+    },
+    {
+      "epoch": 10.9,
+      "eval_loss": 0.23496830463409424,
+      "eval_runtime": 1.4553,
+      "eval_samples_per_second": 1510.368,
+      "eval_steps_per_second": 24.05,
+      "step": 468000
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.3135452418059208e-05,
+      "loss": 0.2513,
+      "step": 468500
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.3037050424308027e-05,
+      "loss": 0.2512,
+      "step": 469000
+    },
+    {
+      "epoch": 10.93,
+      "eval_loss": 0.23394964635372162,
+      "eval_runtime": 1.4438,
+      "eval_samples_per_second": 1522.345,
+      "eval_steps_per_second": 24.241,
+      "step": 469000
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.2940200914628945e-05,
+      "loss": 0.251,
+      "step": 469500
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.2844904948152644e-05,
+      "loss": 0.2508,
+      "step": 470000
+    },
+    {
+      "epoch": 10.95,
+      "eval_loss": 0.23325932025909424,
+      "eval_runtime": 1.4563,
+      "eval_samples_per_second": 1509.326,
+      "eval_steps_per_second": 24.034,
+      "step": 470000
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.2751163567020592e-05,
+      "loss": 0.251,
+      "step": 470500
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.2658977796373478e-05,
+      "loss": 0.2508,
+      "step": 471000
+    },
+    {
+      "epoch": 10.97,
+      "eval_loss": 0.2342199683189392,
+      "eval_runtime": 1.4356,
+      "eval_samples_per_second": 1531.079,
+      "eval_steps_per_second": 24.38,
+      "step": 471000
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.2568348644340153e-05,
+      "loss": 0.2509,
+      "step": 471500
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.2479277102026465e-05,
+      "loss": 0.2508,
+      "step": 472000
+    },
+    {
+      "epoch": 11.0,
+      "eval_loss": 0.23464734852313995,
+      "eval_runtime": 1.4438,
+      "eval_samples_per_second": 1522.38,
+      "eval_steps_per_second": 24.242,
+      "step": 472000
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.2391764143504556e-05,
+      "loss": 0.2504,
+      "step": 472500
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.2305810725802118e-05,
+      "loss": 0.2509,
+      "step": 473000
+    },
+    {
+      "epoch": 11.02,
+      "eval_loss": 0.23320455849170685,
+      "eval_runtime": 1.4791,
+      "eval_samples_per_second": 1486.056,
+      "eval_steps_per_second": 23.663,
+      "step": 473000
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.222141778889195e-05,
+      "loss": 0.2508,
+      "step": 473500
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2138586255681707e-05,
+      "loss": 0.2506,
+      "step": 474000
+    },
+    {
+      "epoch": 11.04,
+      "eval_loss": 0.23511675000190735,
+      "eval_runtime": 1.4624,
+      "eval_samples_per_second": 1503.027,
+      "eval_steps_per_second": 23.934,
+      "step": 474000
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.2057317032003731e-05,
+      "loss": 0.2508,
+      "step": 474500
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.1977611006605263e-05,
+      "loss": 0.2506,
+      "step": 475000
+    },
+    {
+      "epoch": 11.07,
+      "eval_loss": 0.23173680901527405,
+      "eval_runtime": 1.4544,
+      "eval_samples_per_second": 1511.25,
+      "eval_steps_per_second": 24.064,
+      "step": 475000
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.1899469051138602e-05,
+      "loss": 0.2507,
+      "step": 475500
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.1822892020151667e-05,
+      "loss": 0.2508,
+      "step": 476000
+    },
+    {
+      "epoch": 11.09,
+      "eval_loss": 0.23393820226192474,
+      "eval_runtime": 1.4377,
+      "eval_samples_per_second": 1528.825,
+      "eval_steps_per_second": 24.344,
+      "step": 476000
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.1747880751078614e-05,
+      "loss": 0.2506,
+      "step": 476500
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.1674436064230637e-05,
+      "loss": 0.2506,
+      "step": 477000
+    },
+    {
+      "epoch": 11.11,
+      "eval_loss": 0.2342890352010727,
+      "eval_runtime": 1.4636,
+      "eval_samples_per_second": 1501.77,
+      "eval_steps_per_second": 23.914,
+      "step": 477000
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.1602558762787069e-05,
+      "loss": 0.2507,
+      "step": 477500
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.1532249632786582e-05,
+      "loss": 0.2506,
+      "step": 478000
+    },
+    {
+      "epoch": 11.14,
+      "eval_loss": 0.23357579112052917,
+      "eval_runtime": 1.4711,
+      "eval_samples_per_second": 1494.153,
+      "eval_steps_per_second": 23.792,
+      "step": 478000
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.1463509443118552e-05,
+      "loss": 0.2506,
+      "step": 478500
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.1396338945514663e-05,
+      "loss": 0.2508,
+      "step": 479000
+    },
+    {
+      "epoch": 11.16,
+      "eval_loss": 0.23314166069030762,
+      "eval_runtime": 1.4516,
+      "eval_samples_per_second": 1514.175,
+      "eval_steps_per_second": 24.111,
+      "step": 479000
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.133073887454072e-05,
+      "loss": 0.2505,
+      "step": 479500
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.1266709947588599e-05,
+      "loss": 0.2507,
+      "step": 480000
+    },
+    {
+      "epoch": 11.18,
+      "eval_loss": 0.2337002009153366,
+      "eval_runtime": 1.4605,
+      "eval_samples_per_second": 1504.998,
+      "eval_steps_per_second": 23.965,
+      "step": 480000
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.1204252864868377e-05,
+      "loss": 0.2506,
+      "step": 480500
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.1143368309400725e-05,
+      "loss": 0.2506,
+      "step": 481000
+    },
+    {
+      "epoch": 11.21,
+      "eval_loss": 0.2344614714384079,
+      "eval_runtime": 1.4496,
+      "eval_samples_per_second": 1516.33,
+      "eval_steps_per_second": 24.145,
+      "step": 481000
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.1084056947009348e-05,
+      "loss": 0.2507,
+      "step": 481500
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.1026319426313837e-05,
+      "loss": 0.2505,
+      "step": 482000
+    },
+    {
+      "epoch": 11.23,
+      "eval_loss": 0.2328985631465912,
+      "eval_runtime": 1.452,
+      "eval_samples_per_second": 1513.771,
+      "eval_steps_per_second": 24.105,
+      "step": 482000
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.097015637872247e-05,
+      "loss": 0.2505,
+      "step": 482500
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.0915568418425301e-05,
+      "loss": 0.2505,
+      "step": 483000
+    },
+    {
+      "epoch": 11.25,
+      "eval_loss": 0.2360752671957016,
+      "eval_runtime": 1.4507,
+      "eval_samples_per_second": 1515.088,
+      "eval_steps_per_second": 24.126,
+      "step": 483000
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.0862556142387571e-05,
+      "loss": 0.2506,
+      "step": 483500
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.081112013034298e-05,
+      "loss": 0.2505,
+      "step": 484000
+    },
+    {
+      "epoch": 11.28,
+      "eval_loss": 0.23406526446342468,
+      "eval_runtime": 1.4844,
+      "eval_samples_per_second": 1480.738,
+      "eval_steps_per_second": 23.579,
+      "step": 484000
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.0761260944787561e-05,
+      "loss": 0.2505,
+      "step": 484500
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.0712979130973347e-05,
+      "loss": 0.2503,
+      "step": 485000
+    },
+    {
+      "epoch": 11.3,
+      "eval_loss": 0.23385389149188995,
+      "eval_runtime": 1.4509,
+      "eval_samples_per_second": 1514.909,
+      "eval_steps_per_second": 24.123,
+      "step": 485000
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.0666275216902535e-05,
+      "loss": 0.2505,
+      "step": 485500
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.0621149713321656e-05,
+      "loss": 0.2505,
+      "step": 486000
+    },
+    {
+      "epoch": 11.32,
+      "eval_loss": 0.23407165706157684,
+      "eval_runtime": 1.4713,
+      "eval_samples_per_second": 1493.956,
+      "eval_steps_per_second": 23.789,
+      "step": 486000
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.0577603113715964e-05,
+      "loss": 0.2502,
+      "step": 486500
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.0535635894304106e-05,
+      "loss": 0.2505,
+      "step": 487000
+    },
+    {
+      "epoch": 11.35,
+      "eval_loss": 0.23341643810272217,
+      "eval_runtime": 1.4073,
+      "eval_samples_per_second": 1561.856,
+      "eval_steps_per_second": 24.87,
+      "step": 487000
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.0495248514032875e-05,
+      "loss": 0.2503,
+      "step": 487500
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.045644141457218e-05,
+      "loss": 0.2504,
+      "step": 488000
+    },
+    {
+      "epoch": 11.37,
+      "eval_loss": 0.23371146619319916,
+      "eval_runtime": 1.4249,
+      "eval_samples_per_second": 1542.542,
+      "eval_steps_per_second": 24.563,
+      "step": 488000
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.0419215020310254e-05,
+      "loss": 0.2505,
+      "step": 488500
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.0383569738348988e-05,
+      "loss": 0.2504,
+      "step": 489000
+    },
+    {
+      "epoch": 11.39,
+      "eval_loss": 0.23561736941337585,
+      "eval_runtime": 1.4254,
+      "eval_samples_per_second": 1542.027,
+      "eval_steps_per_second": 24.555,
+      "step": 489000
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.0349505958499436e-05,
+      "loss": 0.2501,
+      "step": 489500
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.0317024053277693e-05,
+      "loss": 0.2506,
+      "step": 490000
+    },
+    {
+      "epoch": 11.42,
+      "eval_loss": 0.23189207911491394,
+      "eval_runtime": 1.4243,
+      "eval_samples_per_second": 1543.227,
+      "eval_steps_per_second": 24.574,
+      "step": 490000
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.0286124377900624e-05,
+      "loss": 0.2504,
+      "step": 490500
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.0256807270282153e-05,
+      "loss": 0.2503,
+      "step": 491000
+    },
+    {
+      "epoch": 11.44,
+      "eval_loss": 0.2337212711572647,
+      "eval_runtime": 1.4623,
+      "eval_samples_per_second": 1503.072,
+      "eval_steps_per_second": 23.934,
+      "step": 491000
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.0229073051029455e-05,
+      "loss": 0.25,
+      "step": 491500
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.020292202343952e-05,
+      "loss": 0.2501,
+      "step": 492000
+    },
+    {
+      "epoch": 11.46,
+      "eval_loss": 0.23177069425582886,
+      "eval_runtime": 1.4821,
+      "eval_samples_per_second": 1482.991,
+      "eval_steps_per_second": 23.615,
+      "step": 492000
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.0178354473495813e-05,
+      "loss": 0.2503,
+      "step": 492500
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.0155370669865077e-05,
+      "loss": 0.2505,
+      "step": 493000
+    },
+    {
+      "epoch": 11.49,
+      "eval_loss": 0.23253387212753296,
+      "eval_runtime": 1.4831,
+      "eval_samples_per_second": 1482.001,
+      "eval_steps_per_second": 23.599,
+      "step": 493000
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.0133970863894557e-05,
+      "loss": 0.2502,
+      "step": 493500
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.0114155289609061e-05,
+      "loss": 0.2504,
+      "step": 494000
+    },
+    {
+      "epoch": 11.51,
+      "eval_loss": 0.23376800119876862,
+      "eval_runtime": 1.4253,
+      "eval_samples_per_second": 1542.116,
+      "eval_steps_per_second": 24.556,
+      "step": 494000
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.0095924163708572e-05,
+      "loss": 0.2503,
+      "step": 494500
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.0079277685565724e-05,
+      "loss": 0.2503,
+      "step": 495000
+    },
+    {
+      "epoch": 11.53,
+      "eval_loss": 0.23174437880516052,
+      "eval_runtime": 1.4318,
+      "eval_samples_per_second": 1535.107,
+      "eval_steps_per_second": 24.444,
+      "step": 495000
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 1.0064216037223772e-05,
+      "loss": 0.2501,
+      "step": 495500
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.0050739383394454e-05,
+      "loss": 0.2502,
+      "step": 496000
+    },
+    {
+      "epoch": 11.56,
+      "eval_loss": 0.23404622077941895,
+      "eval_runtime": 1.4342,
+      "eval_samples_per_second": 1532.524,
+      "eval_steps_per_second": 24.403,
+      "step": 496000
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.003884787145633e-05,
+      "loss": 0.2505,
+      "step": 496500
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.002854163145305e-05,
+      "loss": 0.25,
+      "step": 497000
+    },
+    {
+      "epoch": 11.58,
+      "eval_loss": 0.2340567409992218,
+      "eval_runtime": 1.4613,
+      "eval_samples_per_second": 1504.132,
+      "eval_steps_per_second": 23.951,
+      "step": 497000
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.0019820776091995e-05,
+      "loss": 0.2503,
+      "step": 497500
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.0012685400743077e-05,
+      "loss": 0.2502,
+      "step": 498000
+    },
+    {
+      "epoch": 11.6,
+      "eval_loss": 0.2324497401714325,
+      "eval_runtime": 1.4433,
+      "eval_samples_per_second": 1522.91,
+      "eval_steps_per_second": 24.25,
+      "step": 498000
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.0007135583437572e-05,
+      "loss": 0.25,
+      "step": 498500
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.0003171384867436e-05,
+      "loss": 0.2504,
+      "step": 499000
+    },
+    {
+      "epoch": 11.63,
+      "eval_loss": 0.2359778881072998,
+      "eval_runtime": 1.4253,
+      "eval_samples_per_second": 1542.145,
+      "eval_steps_per_second": 24.556,
+      "step": 499000
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.0000792848384467e-05,
+      "loss": 0.2501,
+      "step": 499500
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1e-05,
+      "loss": 0.2501,
+      "step": 500000
+    },
+    {
+      "epoch": 11.65,
+      "eval_loss": 0.23462386429309845,
+      "eval_runtime": 1.444,
+      "eval_samples_per_second": 1522.115,
+      "eval_steps_per_second": 24.238,
+      "step": 500000
+    },
+    {
+      "epoch": 11.65,
+      "step": 500000,
+      "total_flos": 1.5974329471825026e+22,
+      "train_loss": 0.2941521398468018,
+      "train_runtime": 140747.1243,
+      "train_samples_per_second": 1818.865,
+      "train_steps_per_second": 3.552
+    }
+  ],
+  "max_steps": 500000,
+  "num_train_epochs": 12,
+  "total_flos": 1.5974329471825026e+22,
+  "trial_name": null,
+  "trial_params": null
+}