diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,32356 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.2554355997955073,
+  "global_step": 525000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 14.3541,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 14.3676,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-06,
+      "loss": 14.2824,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 14.2217,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 13.7982,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 13.6274,
+      "step": 600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5000000000000004e-06,
+      "loss": 13.2412,
+      "step": 700
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 12.586,
+      "step": 800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-06,
+      "loss": 12.0465,
+      "step": 900
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 11.3204,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 10.3171,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-06,
+      "loss": 9.2676,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 8.2344,
+      "step": 1300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 6.8489,
+      "step": 1400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.5e-06,
+      "loss": 5.5019,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 4.6103,
+      "step": 1600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.500000000000002e-06,
+      "loss": 4.0935,
+      "step": 1700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-06,
+      "loss": 3.9416,
+      "step": 1800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.5e-06,
+      "loss": 3.7521,
+      "step": 1900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 3.6218,
+      "step": 2000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.05e-05,
+      "loss": 3.5332,
+      "step": 2100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 3.3914,
+      "step": 2200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1500000000000002e-05,
+      "loss": 3.3178,
+      "step": 2300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2e-05,
+      "loss": 3.2905,
+      "step": 2400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.25e-05,
+      "loss": 3.1829,
+      "step": 2500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 3.1098,
+      "step": 2600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 3.0184,
+      "step": 2700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 2.9732,
+      "step": 2800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.45e-05,
+      "loss": 2.9343,
+      "step": 2900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-05,
+      "loss": 2.9035,
+      "step": 3000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.55e-05,
+      "loss": 2.8675,
+      "step": 3100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 2.8479,
+      "step": 3200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.65e-05,
+      "loss": 2.8325,
+      "step": 3300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 2.7818,
+      "step": 3400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.75e-05,
+      "loss": 2.7728,
+      "step": 3500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8e-05,
+      "loss": 2.7172,
+      "step": 3600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.85e-05,
+      "loss": 2.7153,
+      "step": 3700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9e-05,
+      "loss": 2.6476,
+      "step": 3800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 2.6379,
+      "step": 3900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 2.6251,
+      "step": 4000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.05e-05,
+      "loss": 2.622,
+      "step": 4100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1e-05,
+      "loss": 2.5939,
+      "step": 4200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.15e-05,
+      "loss": 2.5749,
+      "step": 4300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 2.5746,
+      "step": 4400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.25e-05,
+      "loss": 2.5466,
+      "step": 4500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 2.5754,
+      "step": 4600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.35e-05,
+      "loss": 2.5217,
+      "step": 4700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4e-05,
+      "loss": 2.5456,
+      "step": 4800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.45e-05,
+      "loss": 2.4834,
+      "step": 4900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5e-05,
+      "loss": 2.49,
+      "step": 5000
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 1.9592857360839844,
+      "eval_runtime": 18.7264,
+      "eval_samples_per_second": 534.004,
+      "eval_steps_per_second": 16.714,
+      "step": 5000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5500000000000003e-05,
+      "loss": 2.4657,
+      "step": 5100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 2.457,
+      "step": 5200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 2.4793,
+      "step": 5300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 2.4802,
+      "step": 5400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 2.4615,
+      "step": 5500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 2.4107,
+      "step": 5600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 2.4498,
+      "step": 5700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9e-05,
+      "loss": 2.3964,
+      "step": 5800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.95e-05,
+      "loss": 2.4961,
+      "step": 5900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3e-05,
+      "loss": 2.4611,
+      "step": 6000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.05e-05,
+      "loss": 2.4297,
+      "step": 6100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1e-05,
+      "loss": 2.4205,
+      "step": 6200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.15e-05,
+      "loss": 2.4465,
+      "step": 6300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 2.4045,
+      "step": 6400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 2.4217,
+      "step": 6500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3e-05,
+      "loss": 2.3831,
+      "step": 6600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.35e-05,
+      "loss": 2.3653,
+      "step": 6700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 2.3325,
+      "step": 6800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.45e-05,
+      "loss": 2.38,
+      "step": 6900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5e-05,
+      "loss": 2.3534,
+      "step": 7000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.55e-05,
+      "loss": 2.3172,
+      "step": 7100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6e-05,
+      "loss": 2.3546,
+      "step": 7200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.65e-05,
+      "loss": 2.311,
+      "step": 7300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7e-05,
+      "loss": 2.2996,
+      "step": 7400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 2.3427,
+      "step": 7500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8e-05,
+      "loss": 2.2847,
+      "step": 7600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.85e-05,
+      "loss": 2.2906,
+      "step": 7700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 2.2957,
+      "step": 7800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9500000000000005e-05,
+      "loss": 2.2933,
+      "step": 7900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4e-05,
+      "loss": 2.2658,
+      "step": 8000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.05e-05,
+      "loss": 2.2802,
+      "step": 8100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1e-05,
+      "loss": 2.2505,
+      "step": 8200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.15e-05,
+      "loss": 2.2688,
+      "step": 8300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2e-05,
+      "loss": 2.2176,
+      "step": 8400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.25e-05,
+      "loss": 2.2627,
+      "step": 8500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3e-05,
+      "loss": 2.266,
+      "step": 8600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.35e-05,
+      "loss": 2.2386,
+      "step": 8700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 2.2339,
+      "step": 8800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 2.2431,
+      "step": 8900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.5e-05,
+      "loss": 2.2403,
+      "step": 9000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.55e-05,
+      "loss": 2.2199,
+      "step": 9100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 2.228,
+      "step": 9200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 2.2257,
+      "step": 9300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7e-05,
+      "loss": 2.2182,
+      "step": 9400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.75e-05,
+      "loss": 2.2455,
+      "step": 9500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8e-05,
+      "loss": 2.2127,
+      "step": 9600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.85e-05,
+      "loss": 2.1893,
+      "step": 9700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9e-05,
+      "loss": 2.1823,
+      "step": 9800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 2.1782,
+      "step": 9900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5e-05,
+      "loss": 2.1625,
+      "step": 10000
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 1.8114508390426636,
+      "eval_runtime": 18.7835,
+      "eval_samples_per_second": 532.381,
+      "eval_steps_per_second": 16.664,
+      "step": 10000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9994571613446764e-05,
+      "loss": 2.2058,
+      "step": 10100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998914322689353e-05,
+      "loss": 2.2157,
+      "step": 10200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9983714840340295e-05,
+      "loss": 2.1651,
+      "step": 10300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997828645378706e-05,
+      "loss": 2.2339,
+      "step": 10400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9972858067233825e-05,
+      "loss": 2.1859,
+      "step": 10500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9967429680680594e-05,
+      "loss": 2.1955,
+      "step": 10600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9962001294127356e-05,
+      "loss": 2.1904,
+      "step": 10700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9956572907574125e-05,
+      "loss": 2.1527,
+      "step": 10800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9951144521020886e-05,
+      "loss": 2.1523,
+      "step": 10900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994571613446765e-05,
+      "loss": 2.1323,
+      "step": 11000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994028774791442e-05,
+      "loss": 2.142,
+      "step": 11100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993485936136118e-05,
+      "loss": 2.1683,
+      "step": 11200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992943097480795e-05,
+      "loss": 2.1413,
+      "step": 11300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992400258825471e-05,
+      "loss": 2.1629,
+      "step": 11400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991857420170147e-05,
+      "loss": 2.1141,
+      "step": 11500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9913145815148247e-05,
+      "loss": 2.1713,
+      "step": 11600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.990771742859501e-05,
+      "loss": 2.1313,
+      "step": 11700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.990228904204177e-05,
+      "loss": 2.1351,
+      "step": 11800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.989686065548854e-05,
+      "loss": 2.1445,
+      "step": 11900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98914322689353e-05,
+      "loss": 2.1467,
+      "step": 12000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988600388238206e-05,
+      "loss": 2.1156,
+      "step": 12100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.988057549582883e-05,
+      "loss": 2.1247,
+      "step": 12200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.987514710927559e-05,
+      "loss": 2.1161,
+      "step": 12300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9869718722722355e-05,
+      "loss": 2.112,
+      "step": 12400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9864290336169124e-05,
+      "loss": 2.1165,
+      "step": 12500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9858861949615886e-05,
+      "loss": 2.1256,
+      "step": 12600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9853433563062654e-05,
+      "loss": 2.0823,
+      "step": 12700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984800517650942e-05,
+      "loss": 2.1153,
+      "step": 12800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9842576789956185e-05,
+      "loss": 2.109,
+      "step": 12900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.983714840340295e-05,
+      "loss": 2.1053,
+      "step": 13000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9831720016849715e-05,
+      "loss": 2.1166,
+      "step": 13100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.982629163029648e-05,
+      "loss": 2.0674,
+      "step": 13200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9820863243743246e-05,
+      "loss": 2.1002,
+      "step": 13300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.981543485719001e-05,
+      "loss": 2.095,
+      "step": 13400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.981000647063677e-05,
+      "loss": 2.0961,
+      "step": 13500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980457808408354e-05,
+      "loss": 2.0961,
+      "step": 13600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979914969753031e-05,
+      "loss": 2.079,
+      "step": 13700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979372131097707e-05,
+      "loss": 2.0919,
+      "step": 13800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978829292442384e-05,
+      "loss": 2.1167,
+      "step": 13900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97828645378706e-05,
+      "loss": 2.0993,
+      "step": 14000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977743615131736e-05,
+      "loss": 2.0618,
+      "step": 14100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977200776476413e-05,
+      "loss": 2.0935,
+      "step": 14200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976657937821089e-05,
+      "loss": 2.0928,
+      "step": 14300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9761150991657654e-05,
+      "loss": 2.1033,
+      "step": 14400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975572260510442e-05,
+      "loss": 2.1083,
+      "step": 14500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9750294218551184e-05,
+      "loss": 2.0568,
+      "step": 14600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974486583199795e-05,
+      "loss": 2.039,
+      "step": 14700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973943744544472e-05,
+      "loss": 2.0834,
+      "step": 14800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973400905889148e-05,
+      "loss": 2.0732,
+      "step": 14900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9728580672338245e-05,
+      "loss": 2.0624,
+      "step": 15000
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 1.7244441509246826,
+      "eval_runtime": 18.7641,
+      "eval_samples_per_second": 532.933,
+      "eval_steps_per_second": 16.681,
+      "step": 15000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9723152285785014e-05,
+      "loss": 2.0532,
+      "step": 15100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9717723899231776e-05,
+      "loss": 2.0569,
+      "step": 15200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9712295512678544e-05,
+      "loss": 2.0622,
+      "step": 15300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9706867126125306e-05,
+      "loss": 2.0448,
+      "step": 15400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.970143873957207e-05,
+      "loss": 2.0594,
+      "step": 15500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969601035301884e-05,
+      "loss": 2.0486,
+      "step": 15600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.96905819664656e-05,
+      "loss": 2.0488,
+      "step": 15700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968515357991237e-05,
+      "loss": 2.03,
+      "step": 15800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9679725193359136e-05,
+      "loss": 2.0228,
+      "step": 15900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.96742968068059e-05,
+      "loss": 2.0109,
+      "step": 16000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966886842025266e-05,
+      "loss": 2.0503,
+      "step": 16100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.966344003369943e-05,
+      "loss": 2.0112,
+      "step": 16200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965801164714619e-05,
+      "loss": 2.0324,
+      "step": 16300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965258326059295e-05,
+      "loss": 2.0737,
+      "step": 16400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.964715487403972e-05,
+      "loss": 2.0184,
+      "step": 16500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.964172648748648e-05,
+      "loss": 2.0513,
+      "step": 16600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963629810093325e-05,
+      "loss": 2.0329,
+      "step": 16700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963086971438001e-05,
+      "loss": 2.0303,
+      "step": 16800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.962544132782678e-05,
+      "loss": 2.0042,
+      "step": 16900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9620012941273544e-05,
+      "loss": 2.0358,
+      "step": 17000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.961458455472031e-05,
+      "loss": 2.041,
+      "step": 17100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9609156168167074e-05,
+      "loss": 2.0349,
+      "step": 17200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.960372778161384e-05,
+      "loss": 2.0428,
+      "step": 17300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9598299395060605e-05,
+      "loss": 2.0209,
+      "step": 17400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.959287100850737e-05,
+      "loss": 2.0384,
+      "step": 17500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9587442621954135e-05,
+      "loss": 2.0594,
+      "step": 17600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.95820142354009e-05,
+      "loss": 2.0087,
+      "step": 17700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.957658584884766e-05,
+      "loss": 2.0458,
+      "step": 17800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9571157462294434e-05,
+      "loss": 2.0377,
+      "step": 17900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9565729075741196e-05,
+      "loss": 2.0111,
+      "step": 18000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.956030068918796e-05,
+      "loss": 2.0155,
+      "step": 18100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.955487230263473e-05,
+      "loss": 2.0355,
+      "step": 18200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954944391608149e-05,
+      "loss": 2.0127,
+      "step": 18300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.954401552952825e-05,
+      "loss": 1.9876,
+      "step": 18400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953858714297502e-05,
+      "loss": 2.0298,
+      "step": 18500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.953315875642178e-05,
+      "loss": 2.0306,
+      "step": 18600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952773036986855e-05,
+      "loss": 2.0032,
+      "step": 18700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.952230198331531e-05,
+      "loss": 1.9919,
+      "step": 18800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9516873596762074e-05,
+      "loss": 1.9956,
+      "step": 18900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.951144521020884e-05,
+      "loss": 1.9778,
+      "step": 19000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950601682365561e-05,
+      "loss": 1.9788,
+      "step": 19100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.950058843710237e-05,
+      "loss": 2.003,
+      "step": 19200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.949516005054914e-05,
+      "loss": 2.0157,
+      "step": 19300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.94897316639959e-05,
+      "loss": 2.0052,
+      "step": 19400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9484303277442665e-05,
+      "loss": 2.0187,
+      "step": 19500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9478874890889434e-05,
+      "loss": 1.9949,
+      "step": 19600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9473446504336196e-05,
+      "loss": 1.9735,
+      "step": 19700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.946801811778296e-05,
+      "loss": 1.9628,
+      "step": 19800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9462589731229726e-05,
+      "loss": 1.9955,
+      "step": 19900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9457161344676495e-05,
+      "loss": 1.9873,
+      "step": 20000
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 1.6870460510253906,
+      "eval_runtime": 18.8219,
+      "eval_samples_per_second": 531.295,
+      "eval_steps_per_second": 16.63,
+      "step": 20000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.945173295812326e-05,
+      "loss": 1.9887,
+      "step": 20100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9446304571570025e-05,
+      "loss": 2.0004,
+      "step": 20200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.944087618501679e-05,
+      "loss": 1.9967,
+      "step": 20300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.943544779846355e-05,
+      "loss": 2.0021,
+      "step": 20400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.943001941191032e-05,
+      "loss": 1.9617,
+      "step": 20500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.942459102535708e-05,
+      "loss": 1.9755,
+      "step": 20600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.941916263880385e-05,
+      "loss": 2.0077,
+      "step": 20700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.941373425225061e-05,
+      "loss": 2.0046,
+      "step": 20800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.940830586569737e-05,
+      "loss": 1.9379,
+      "step": 20900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.940287747914414e-05,
+      "loss": 2.0161,
+      "step": 21000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.939744909259091e-05,
+      "loss": 2.0048,
+      "step": 21100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.939202070603767e-05,
+      "loss": 1.9474,
+      "step": 21200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.938659231948444e-05,
+      "loss": 1.9633,
+      "step": 21300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.93811639329312e-05,
+      "loss": 1.9736,
+      "step": 21400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9375735546377964e-05,
+      "loss": 1.9645,
+      "step": 21500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.937030715982473e-05,
+      "loss": 1.9823,
+      "step": 21600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9364878773271494e-05,
+      "loss": 1.9318,
+      "step": 21700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9359450386718256e-05,
+      "loss": 1.9283,
+      "step": 21800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9354022000165025e-05,
+      "loss": 1.9433,
+      "step": 21900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9348593613611787e-05,
+      "loss": 1.9767,
+      "step": 22000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9343165227058555e-05,
+      "loss": 1.9709,
+      "step": 22100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9337736840505324e-05,
+      "loss": 1.9541,
+      "step": 22200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9332308453952086e-05,
+      "loss": 1.9767,
+      "step": 22300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.932688006739885e-05,
+      "loss": 1.959,
+      "step": 22400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9321451680845616e-05,
+      "loss": 1.9543,
+      "step": 22500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.931602329429238e-05,
+      "loss": 1.9563,
+      "step": 22600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.931059490773915e-05,
+      "loss": 1.94,
+      "step": 22700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.930516652118591e-05,
+      "loss": 1.9529,
+      "step": 22800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.929973813463267e-05,
+      "loss": 1.9763,
+      "step": 22900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.929430974807944e-05,
+      "loss": 2.0103,
+      "step": 23000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.928888136152621e-05,
+      "loss": 1.9782,
+      "step": 23100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.928345297497297e-05,
+      "loss": 1.9247,
+      "step": 23200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.927802458841974e-05,
+      "loss": 1.9302,
+      "step": 23300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.92725962018665e-05,
+      "loss": 1.9572,
+      "step": 23400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.926716781531326e-05,
+      "loss": 1.9389,
+      "step": 23500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.926173942876003e-05,
+      "loss": 1.9863,
+      "step": 23600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.925631104220679e-05,
+      "loss": 1.9723,
+      "step": 23700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9250882655653555e-05,
+      "loss": 1.9609,
+      "step": 23800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.924545426910032e-05,
+      "loss": 1.9826,
+      "step": 23900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9240025882547085e-05,
+      "loss": 1.9683,
+      "step": 24000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9234597495993854e-05,
+      "loss": 1.9441,
+      "step": 24100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.922916910944062e-05,
+      "loss": 1.888,
+      "step": 24200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9223740722887384e-05,
+      "loss": 1.9307,
+      "step": 24300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9218312336334146e-05,
+      "loss": 1.9494,
+      "step": 24400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9212883949780915e-05,
+      "loss": 1.9249,
+      "step": 24500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.920745556322768e-05,
+      "loss": 1.936,
+      "step": 24600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9202027176674445e-05,
+      "loss": 1.9151,
+      "step": 24700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.919659879012121e-05,
+      "loss": 1.9327,
+      "step": 24800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.919117040356797e-05,
+      "loss": 1.9377,
+      "step": 24900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.918574201701474e-05,
+      "loss": 1.9319,
+      "step": 25000
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 1.659123420715332,
+      "eval_runtime": 18.8272,
+      "eval_samples_per_second": 531.147,
+      "eval_steps_per_second": 16.625,
+      "step": 25000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.91803136304615e-05,
+      "loss": 1.9475,
+      "step": 25100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.917488524390827e-05,
+      "loss": 1.9135,
+      "step": 25200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.916945685735504e-05,
+      "loss": 1.9213,
+      "step": 25300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.91640284708018e-05,
+      "loss": 1.9438,
+      "step": 25400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.915860008424856e-05,
+      "loss": 1.9821,
+      "step": 25500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.915317169769533e-05,
+      "loss": 1.9149,
+      "step": 25600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.914774331114209e-05,
+      "loss": 1.9356,
+      "step": 25700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.914231492458885e-05,
+      "loss": 1.9192,
+      "step": 25800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.913688653803562e-05,
+      "loss": 1.9404,
+      "step": 25900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9131458151482384e-05,
+      "loss": 1.9447,
+      "step": 26000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.912602976492915e-05,
+      "loss": 1.9286,
+      "step": 26100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9120601378375914e-05,
+      "loss": 1.9304,
+      "step": 26200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.911517299182268e-05,
+      "loss": 1.9031,
+      "step": 26300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9109744605269445e-05,
+      "loss": 1.9244,
+      "step": 26400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.910431621871621e-05,
+      "loss": 1.9265,
+      "step": 26500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9098887832162975e-05,
+      "loss": 1.9259,
+      "step": 26600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9093459445609744e-05,
+      "loss": 1.9633,
+      "step": 26700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9088031059056506e-05,
+      "loss": 1.949,
+      "step": 26800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.908260267250327e-05,
+      "loss": 1.9403,
+      "step": 26900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9077174285950036e-05,
+      "loss": 1.9391,
+      "step": 27000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.90717458993968e-05,
+      "loss": 1.9692,
+      "step": 27100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.906631751284356e-05,
+      "loss": 1.9442,
+      "step": 27200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9060889126290335e-05,
+      "loss": 1.9369,
+      "step": 27300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.90554607397371e-05,
+      "loss": 1.9239,
+      "step": 27400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.905003235318386e-05,
+      "loss": 1.9146,
+      "step": 27500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.904460396663063e-05,
+      "loss": 1.9086,
+      "step": 27600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.903917558007739e-05,
+      "loss": 1.9168,
+      "step": 27700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.903374719352415e-05,
+      "loss": 1.9262,
+      "step": 27800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.902831880697092e-05,
+      "loss": 1.9099,
+      "step": 27900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.902289042041768e-05,
+      "loss": 1.9157,
+      "step": 28000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.901746203386445e-05,
+      "loss": 1.9182,
+      "step": 28100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.901203364731121e-05,
+      "loss": 1.8918,
+      "step": 28200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9006605260757974e-05,
+      "loss": 1.9396,
+      "step": 28300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.900117687420474e-05,
+      "loss": 1.898,
+      "step": 28400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.899574848765151e-05,
+      "loss": 1.9377,
+      "step": 28500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8990320101098274e-05,
+      "loss": 1.9022,
+      "step": 28600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.898489171454504e-05,
+      "loss": 1.9118,
+      "step": 28700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8979463327991804e-05,
+      "loss": 1.9007,
+      "step": 28800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8974034941438566e-05,
+      "loss": 1.9042,
+      "step": 28900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8968606554885335e-05,
+      "loss": 1.8985,
+      "step": 29000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8963178168332097e-05,
+      "loss": 1.935,
+      "step": 29100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.895774978177886e-05,
+      "loss": 1.8826,
+      "step": 29200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.895232139522563e-05,
+      "loss": 1.9218,
+      "step": 29300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8946893008672396e-05,
+      "loss": 1.8877,
+      "step": 29400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.894146462211916e-05,
+      "loss": 1.8943,
+      "step": 29500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8936036235565926e-05,
+      "loss": 1.9168,
+      "step": 29600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.893060784901269e-05,
+      "loss": 1.8982,
+      "step": 29700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.892517946245945e-05,
+      "loss": 1.9107,
+      "step": 29800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.891975107590622e-05,
+      "loss": 1.9192,
+      "step": 29900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.891432268935298e-05,
+      "loss": 1.8701,
+      "step": 30000
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 1.6432205438613892,
+      "eval_runtime": 18.7697,
+      "eval_samples_per_second": 532.772,
+      "eval_steps_per_second": 16.676,
+      "step": 30000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.890889430279975e-05,
+      "loss": 1.8962,
+      "step": 30100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.890346591624651e-05,
+      "loss": 1.9178,
+      "step": 30200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.889803752969327e-05,
+      "loss": 1.9007,
+      "step": 30300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.889260914314004e-05,
+      "loss": 1.9306,
+      "step": 30400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.888718075658681e-05,
+      "loss": 1.8875,
+      "step": 30500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.888175237003357e-05,
+      "loss": 1.9071,
+      "step": 30600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.887632398348034e-05,
+      "loss": 1.9057,
+      "step": 30700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.88708955969271e-05,
+      "loss": 1.9082,
+      "step": 30800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8865467210373864e-05,
+      "loss": 1.9019,
+      "step": 30900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.886003882382063e-05,
+      "loss": 1.9186,
+      "step": 31000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8854610437267395e-05,
+      "loss": 1.9171,
+      "step": 31100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.884918205071416e-05,
+      "loss": 1.9104,
+      "step": 31200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8843753664160926e-05,
+      "loss": 1.8954,
+      "step": 31300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.883832527760769e-05,
+      "loss": 1.9037,
+      "step": 31400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8832896891054456e-05,
+      "loss": 1.8636,
+      "step": 31500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8827468504501225e-05,
+      "loss": 1.8734,
+      "step": 31600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8822040117947987e-05,
+      "loss": 1.8882,
+      "step": 31700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.881661173139475e-05,
+      "loss": 1.8936,
+      "step": 31800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.881118334484152e-05,
+      "loss": 1.8927,
+      "step": 31900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.880575495828828e-05,
+      "loss": 1.8856,
+      "step": 32000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.880032657173505e-05,
+      "loss": 1.8895,
+      "step": 32100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.879489818518181e-05,
+      "loss": 1.855,
+      "step": 32200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.878946979862857e-05,
+      "loss": 1.9009,
+      "step": 32300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.878404141207534e-05,
+      "loss": 1.8999,
+      "step": 32400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.87786130255221e-05,
+      "loss": 1.9017,
+      "step": 32500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.877318463896887e-05,
+      "loss": 1.9007,
+      "step": 32600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.876775625241564e-05,
+      "loss": 1.8886,
+      "step": 32700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.87623278658624e-05,
+      "loss": 1.8951,
+      "step": 32800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.875689947930916e-05,
+      "loss": 1.9086,
+      "step": 32900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.875147109275593e-05,
+      "loss": 1.8772,
+      "step": 33000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8746042706202693e-05,
+      "loss": 1.849,
+      "step": 33100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8740614319649455e-05,
+      "loss": 1.9026,
+      "step": 33200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8735185933096224e-05,
+      "loss": 1.8813,
+      "step": 33300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8729757546542986e-05,
+      "loss": 1.8758,
+      "step": 33400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.872432915998975e-05,
+      "loss": 1.8896,
+      "step": 33500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.871890077343652e-05,
+      "loss": 1.8888,
+      "step": 33600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8713472386883285e-05,
+      "loss": 1.8835,
+      "step": 33700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.870804400033005e-05,
+      "loss": 1.8602,
+      "step": 33800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8702615613776816e-05,
+      "loss": 1.9211,
+      "step": 33900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.869718722722358e-05,
+      "loss": 1.8607,
+      "step": 34000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8691758840670346e-05,
+      "loss": 1.9135,
+      "step": 34100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.868633045411711e-05,
+      "loss": 1.8718,
+      "step": 34200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.868090206756387e-05,
+      "loss": 1.8732,
+      "step": 34300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.867547368101064e-05,
+      "loss": 1.8962,
+      "step": 34400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.86700452944574e-05,
+      "loss": 1.8708,
+      "step": 34500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.866461690790416e-05,
+      "loss": 1.8837,
+      "step": 34600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.865918852135094e-05,
+      "loss": 1.8801,
+      "step": 34700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.86537601347977e-05,
+      "loss": 1.8832,
+      "step": 34800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.864833174824446e-05,
+      "loss": 1.8599,
+      "step": 34900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.864290336169123e-05,
+      "loss": 1.878,
+      "step": 35000
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 1.6231168508529663,
+      "eval_runtime": 18.8376,
+      "eval_samples_per_second": 530.853,
+      "eval_steps_per_second": 16.616,
+      "step": 35000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.863747497513799e-05,
+      "loss": 1.8869,
+      "step": 35100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8632046588584754e-05,
+      "loss": 1.8941,
+      "step": 35200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.862661820203152e-05,
+      "loss": 1.8673,
+      "step": 35300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8621189815478284e-05,
+      "loss": 1.8841,
+      "step": 35400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8615761428925046e-05,
+      "loss": 1.8753,
+      "step": 35500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8610333042371815e-05,
+      "loss": 1.8634,
+      "step": 35600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8604904655818584e-05,
+      "loss": 1.8727,
+      "step": 35700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8599476269265345e-05,
+      "loss": 1.858,
+      "step": 35800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8594047882712114e-05,
+      "loss": 1.9326,
+      "step": 35900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8588619496158876e-05,
+      "loss": 1.8632,
+      "step": 36000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8583191109605645e-05,
+      "loss": 1.8698,
+      "step": 36100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8577762723052406e-05,
+      "loss": 1.8401,
+      "step": 36200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.857233433649917e-05,
+      "loss": 1.8499,
+      "step": 36300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.856690594994594e-05,
+      "loss": 1.8542,
+      "step": 36400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.85614775633927e-05,
+      "loss": 1.8607,
+      "step": 36500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.855604917683946e-05,
+      "loss": 1.8488,
+      "step": 36600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.855062079028623e-05,
+      "loss": 1.8528,
+      "step": 36700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8545192403733e-05,
+      "loss": 1.8522,
+      "step": 36800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.853976401717976e-05,
+      "loss": 1.8263,
+      "step": 36900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.853433563062653e-05,
+      "loss": 1.8421,
+      "step": 37000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.852890724407329e-05,
+      "loss": 1.8677,
+      "step": 37100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.852347885752005e-05,
+      "loss": 1.8743,
+      "step": 37200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.851805047096682e-05,
+      "loss": 1.8847,
+      "step": 37300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.851262208441358e-05,
+      "loss": 1.8881,
+      "step": 37400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8507193697860345e-05,
+      "loss": 1.8622,
+      "step": 37500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8501765311307113e-05,
+      "loss": 1.8695,
+      "step": 37600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8496336924753875e-05,
+      "loss": 1.8583,
+      "step": 37700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8490908538200644e-05,
+      "loss": 1.8674,
+      "step": 37800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.848548015164741e-05,
+      "loss": 1.8767,
+      "step": 37900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8480051765094174e-05,
+      "loss": 1.8579,
+      "step": 38000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.847462337854094e-05,
+      "loss": 1.8644,
+      "step": 38100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8469194991987705e-05,
+      "loss": 1.8259,
+      "step": 38200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.846376660543447e-05,
+      "loss": 1.8685,
+      "step": 38300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8458338218881236e-05,
+      "loss": 1.8588,
+      "step": 38400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8452909832328e-05,
+      "loss": 1.8768,
+      "step": 38500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.844748144577476e-05,
+      "loss": 1.8565,
+      "step": 38600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.844205305922153e-05,
+      "loss": 1.8883,
+      "step": 38700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.843662467266829e-05,
+      "loss": 1.8884,
+      "step": 38800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.843119628611506e-05,
+      "loss": 1.8314,
+      "step": 38900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.842576789956183e-05,
+      "loss": 1.8166,
+      "step": 39000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.842033951300859e-05,
+      "loss": 1.8752,
+      "step": 39100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.841491112645535e-05,
+      "loss": 1.8429,
+      "step": 39200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.840948273990212e-05,
+      "loss": 1.841,
+      "step": 39300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.840405435334888e-05,
+      "loss": 1.8476,
+      "step": 39400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.839862596679564e-05,
+      "loss": 1.8538,
+      "step": 39500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.839319758024241e-05,
+      "loss": 1.8662,
+      "step": 39600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8387769193689174e-05,
+      "loss": 1.8432,
+      "step": 39700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.838234080713594e-05,
+      "loss": 1.8317,
+      "step": 39800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.837691242058271e-05,
+      "loss": 1.8308,
+      "step": 39900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.837148403402947e-05,
+      "loss": 1.8797,
+      "step": 40000
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 1.6136231422424316,
+      "eval_runtime": 18.8524,
+      "eval_samples_per_second": 530.437,
+      "eval_steps_per_second": 16.603,
+      "step": 40000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.836605564747624e-05,
+      "loss": 1.857,
+      "step": 40100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8360627260923003e-05,
+      "loss": 1.8894,
+      "step": 40200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8355198874369765e-05,
+      "loss": 1.881,
+      "step": 40300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8349770487816534e-05,
+      "loss": 1.8252,
+      "step": 40400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8344342101263296e-05,
+      "loss": 1.8565,
+      "step": 40500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.833891371471006e-05,
+      "loss": 1.8664,
+      "step": 40600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8333485328156826e-05,
+      "loss": 1.8833,
+      "step": 40700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.832805694160359e-05,
+      "loss": 1.8351,
+      "step": 40800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.832262855505036e-05,
+      "loss": 1.8889,
+      "step": 40900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8317200168497126e-05,
+      "loss": 1.8882,
+      "step": 41000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.831177178194389e-05,
+      "loss": 1.8759,
+      "step": 41100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.830634339539065e-05,
+      "loss": 1.8267,
+      "step": 41200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.830091500883742e-05,
+      "loss": 1.8654,
+      "step": 41300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.829548662228418e-05,
+      "loss": 1.8388,
+      "step": 41400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.829005823573094e-05,
+      "loss": 1.8393,
+      "step": 41500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.828462984917771e-05,
+      "loss": 1.8669,
+      "step": 41600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.827920146262447e-05,
+      "loss": 1.8393,
+      "step": 41700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.827377307607124e-05,
+      "loss": 1.8635,
+      "step": 41800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8268344689518e-05,
+      "loss": 1.8317,
+      "step": 41900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.826291630296477e-05,
+      "loss": 1.8476,
+      "step": 42000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.825748791641154e-05,
+      "loss": 1.8106,
+      "step": 42100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.82520595298583e-05,
+      "loss": 1.8654,
+      "step": 42200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8246631143305064e-05,
+      "loss": 1.8358,
+      "step": 42300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.824120275675183e-05,
+      "loss": 1.8609,
+      "step": 42400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8235774370198594e-05,
+      "loss": 1.8331,
+      "step": 42500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8230345983645356e-05,
+      "loss": 1.8536,
+      "step": 42600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8224917597092125e-05,
+      "loss": 1.8018,
+      "step": 42700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.821948921053889e-05,
+      "loss": 1.8408,
+      "step": 42800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.821406082398565e-05,
+      "loss": 1.803,
+      "step": 42900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8208632437432424e-05,
+      "loss": 1.8361,
+      "step": 43000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8203204050879186e-05,
+      "loss": 1.8156,
+      "step": 43100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.819777566432595e-05,
+      "loss": 1.8568,
+      "step": 43200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8192347277772716e-05,
+      "loss": 1.8422,
+      "step": 43300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.818691889121948e-05,
+      "loss": 1.8498,
+      "step": 43400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.818149050466624e-05,
+      "loss": 1.8605,
+      "step": 43500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.817606211811301e-05,
+      "loss": 1.8621,
+      "step": 43600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.817063373155977e-05,
+      "loss": 1.8613,
+      "step": 43700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.816520534500654e-05,
+      "loss": 1.839,
+      "step": 43800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.81597769584533e-05,
+      "loss": 1.8462,
+      "step": 43900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.815434857190006e-05,
+      "loss": 1.7864,
+      "step": 44000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.814892018534684e-05,
+      "loss": 1.8452,
+      "step": 44100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.81434917987936e-05,
+      "loss": 1.8469,
+      "step": 44200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.813806341224036e-05,
+      "loss": 1.8697,
+      "step": 44300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.813263502568713e-05,
+      "loss": 1.8287,
+      "step": 44400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.812720663913389e-05,
+      "loss": 1.8486,
+      "step": 44500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8121778252580655e-05,
+      "loss": 1.8107,
+      "step": 44600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.811634986602742e-05,
+      "loss": 1.8509,
+      "step": 44700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8110921479474185e-05,
+      "loss": 1.8531,
+      "step": 44800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.810549309292095e-05,
+      "loss": 1.8266,
+      "step": 44900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8100064706367716e-05,
+      "loss": 1.8125,
+      "step": 45000
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 1.6042814254760742,
+      "eval_runtime": 18.8207,
+      "eval_samples_per_second": 531.33,
+      "eval_steps_per_second": 16.631,
+      "step": 45000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8094636319814484e-05,
+      "loss": 1.8248,
+      "step": 45100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8089207933261246e-05,
+      "loss": 1.83,
+      "step": 45200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8083779546708015e-05,
+      "loss": 1.8339,
+      "step": 45300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.807835116015478e-05,
+      "loss": 1.8315,
+      "step": 45400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.807292277360154e-05,
+      "loss": 1.8314,
+      "step": 45500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.806749438704831e-05,
+      "loss": 1.8101,
+      "step": 45600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.806206600049507e-05,
+      "loss": 1.822,
+      "step": 45700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.805663761394184e-05,
+      "loss": 1.8244,
+      "step": 45800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.80512092273886e-05,
+      "loss": 1.8312,
+      "step": 45900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.804578084083536e-05,
+      "loss": 1.8351,
+      "step": 46000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.804035245428213e-05,
+      "loss": 1.8355,
+      "step": 46100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.80349240677289e-05,
+      "loss": 1.8204,
+      "step": 46200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.802949568117566e-05,
+      "loss": 1.8721,
+      "step": 46300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.802406729462243e-05,
+      "loss": 1.8235,
+      "step": 46400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.801863890806919e-05,
+      "loss": 1.8558,
+      "step": 46500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.801321052151595e-05,
+      "loss": 1.8399,
+      "step": 46600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.800778213496272e-05,
+      "loss": 1.8175,
+      "step": 46700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8002353748409484e-05,
+      "loss": 1.825,
+      "step": 46800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.7996925361856246e-05,
+      "loss": 1.8156,
+      "step": 46900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.7991496975303014e-05,
+      "loss": 1.8063,
+      "step": 47000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.7986068588749776e-05,
+      "loss": 1.8077,
+      "step": 47100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.7980640202196545e-05,
+      "loss": 1.8135,
+      "step": 47200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.7975211815643313e-05,
+      "loss": 1.8466,
+      "step": 47300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.7969783429090075e-05,
+      "loss": 1.8319,
+      "step": 47400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.796435504253684e-05,
+      "loss": 1.8166,
+      "step": 47500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.7958926655983606e-05,
+      "loss": 1.8097,
+      "step": 47600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.795349826943037e-05,
+      "loss": 1.8231,
+      "step": 47700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7948069882877136e-05,
+      "loss": 1.8371,
+      "step": 47800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.79426414963239e-05,
+      "loss": 1.8519,
+      "step": 47900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.793721310977066e-05,
+      "loss": 1.8226,
+      "step": 48000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.793178472321743e-05,
+      "loss": 1.8081,
+      "step": 48100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.792635633666419e-05,
+      "loss": 1.7755,
+      "step": 48200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.792092795011096e-05,
+      "loss": 1.8338,
+      "step": 48300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.791549956355773e-05,
+      "loss": 1.8139,
+      "step": 48400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.791007117700449e-05,
+      "loss": 1.8214,
+      "step": 48500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.790464279045125e-05,
+      "loss": 1.8017,
+      "step": 48600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.789921440389802e-05,
+      "loss": 1.8379,
+      "step": 48700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.789378601734478e-05,
+      "loss": 1.8305,
+      "step": 48800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7888357630791544e-05,
+      "loss": 1.8496,
+      "step": 48900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.788292924423831e-05,
+      "loss": 1.8071,
+      "step": 49000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7877500857685075e-05,
+      "loss": 1.8217,
+      "step": 49100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.787207247113184e-05,
+      "loss": 1.7922,
+      "step": 49200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.786664408457861e-05,
+      "loss": 1.8252,
+      "step": 49300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7861215698025374e-05,
+      "loss": 1.8218,
+      "step": 49400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7855787311472136e-05,
+      "loss": 1.8188,
+      "step": 49500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7850358924918904e-05,
+      "loss": 1.8211,
+      "step": 49600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7844930538365666e-05,
+      "loss": 1.8032,
+      "step": 49700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7839502151812435e-05,
+      "loss": 1.7968,
+      "step": 49800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.78340737652592e-05,
+      "loss": 1.8099,
+      "step": 49900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.782864537870596e-05,
+      "loss": 1.8188,
+      "step": 50000
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 1.5851926803588867,
+      "eval_runtime": 18.9094,
+      "eval_samples_per_second": 528.838,
+      "eval_steps_per_second": 16.553,
+      "step": 50000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.782321699215273e-05,
+      "loss": 1.8253,
+      "step": 50100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.781778860559949e-05,
+      "loss": 1.7936,
+      "step": 50200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.781236021904625e-05,
+      "loss": 1.8353,
+      "step": 50300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7806931832493026e-05,
+      "loss": 1.7939,
+      "step": 50400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.780150344593979e-05,
+      "loss": 1.8055,
+      "step": 50500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.779607505938655e-05,
+      "loss": 1.8032,
+      "step": 50600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.779064667283332e-05,
+      "loss": 1.8171,
+      "step": 50700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.778521828628008e-05,
+      "loss": 1.8282,
+      "step": 50800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.777978989972684e-05,
+      "loss": 1.8412,
+      "step": 50900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.777436151317361e-05,
+      "loss": 1.8437,
+      "step": 51000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.776893312662037e-05,
+      "loss": 1.8513,
+      "step": 51100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.776350474006714e-05,
+      "loss": 1.7993,
+      "step": 51200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7758076353513904e-05,
+      "loss": 1.7976,
+      "step": 51300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.775264796696067e-05,
+      "loss": 1.8163,
+      "step": 51400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7747219580407434e-05,
+      "loss": 1.8247,
+      "step": 51500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.77417911938542e-05,
+      "loss": 1.7752,
+      "step": 51600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7736362807300965e-05,
+      "loss": 1.8202,
+      "step": 51700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.773093442074773e-05,
+      "loss": 1.8208,
+      "step": 51800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7725506034194495e-05,
+      "loss": 1.8177,
+      "step": 51900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.772007764764126e-05,
+      "loss": 1.8148,
+      "step": 52000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7714649261088026e-05,
+      "loss": 1.8204,
+      "step": 52100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.770922087453479e-05,
+      "loss": 1.8226,
+      "step": 52200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.770379248798155e-05,
+      "loss": 1.8235,
+      "step": 52300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.769836410142832e-05,
+      "loss": 1.7702,
+      "step": 52400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.769293571487509e-05,
+      "loss": 1.8013,
+      "step": 52500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.768750732832185e-05,
+      "loss": 1.801,
+      "step": 52600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.768207894176862e-05,
+      "loss": 1.8202,
+      "step": 52700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.767665055521538e-05,
+      "loss": 1.7901,
+      "step": 52800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.767122216866214e-05,
+      "loss": 1.8395,
+      "step": 52900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.766579378210891e-05,
+      "loss": 1.7996,
+      "step": 53000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.766036539555567e-05,
+      "loss": 1.7787,
+      "step": 53100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.765493700900244e-05,
+      "loss": 1.7861,
+      "step": 53200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.76495086224492e-05,
+      "loss": 1.8164,
+      "step": 53300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7644080235895964e-05,
+      "loss": 1.8221,
+      "step": 53400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.763865184934273e-05,
+      "loss": 1.7885,
+      "step": 53500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.76332234627895e-05,
+      "loss": 1.8154,
+      "step": 53600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.762779507623626e-05,
+      "loss": 1.7901,
+      "step": 53700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.762236668968303e-05,
+      "loss": 1.8313,
+      "step": 53800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7616938303129794e-05,
+      "loss": 1.801,
+      "step": 53900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7611509916576556e-05,
+      "loss": 1.8021,
+      "step": 54000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7606081530023324e-05,
+      "loss": 1.8145,
+      "step": 54100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7600653143470086e-05,
+      "loss": 1.7976,
+      "step": 54200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.759522475691685e-05,
+      "loss": 1.7932,
+      "step": 54300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.758979637036362e-05,
+      "loss": 1.8452,
+      "step": 54400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.758436798381038e-05,
+      "loss": 1.7923,
+      "step": 54500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.757893959725715e-05,
+      "loss": 1.7934,
+      "step": 54600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7573511210703916e-05,
+      "loss": 1.8072,
+      "step": 54700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.756808282415068e-05,
+      "loss": 1.8279,
+      "step": 54800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.756265443759744e-05,
+      "loss": 1.8289,
+      "step": 54900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.755722605104421e-05,
+      "loss": 1.812,
+      "step": 55000
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 1.5801465511322021,
+      "eval_runtime": 18.8482,
+      "eval_samples_per_second": 530.554,
+      "eval_steps_per_second": 16.606,
+      "step": 55000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.755179766449097e-05,
+      "loss": 1.7898,
+      "step": 55100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.754636927793774e-05,
+      "loss": 1.7933,
+      "step": 55200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.75409408913845e-05,
+      "loss": 1.7846,
+      "step": 55300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.753551250483126e-05,
+      "loss": 1.7918,
+      "step": 55400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.753008411827803e-05,
+      "loss": 1.798,
+      "step": 55500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.75246557317248e-05,
+      "loss": 1.799,
+      "step": 55600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.751922734517156e-05,
+      "loss": 1.8078,
+      "step": 55700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.751379895861833e-05,
+      "loss": 1.808,
+      "step": 55800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.750837057206509e-05,
+      "loss": 1.7562,
+      "step": 55900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7502942185511854e-05,
+      "loss": 1.8129,
+      "step": 56000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.749751379895862e-05,
+      "loss": 1.8013,
+      "step": 56100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7492085412405385e-05,
+      "loss": 1.8138,
+      "step": 56200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7486657025852146e-05,
+      "loss": 1.8014,
+      "step": 56300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7481228639298915e-05,
+      "loss": 1.8381,
+      "step": 56400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.747580025274568e-05,
+      "loss": 1.7954,
+      "step": 56500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7470371866192446e-05,
+      "loss": 1.7694,
+      "step": 56600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7464943479639214e-05,
+      "loss": 1.7909,
+      "step": 56700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7459515093085976e-05,
+      "loss": 1.8301,
+      "step": 56800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.745408670653274e-05,
+      "loss": 1.7995,
+      "step": 56900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.744865831997951e-05,
+      "loss": 1.8072,
+      "step": 57000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.744322993342627e-05,
+      "loss": 1.784,
+      "step": 57100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.743780154687304e-05,
+      "loss": 1.8026,
+      "step": 57200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.74323731603198e-05,
+      "loss": 1.7927,
+      "step": 57300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.742694477376656e-05,
+      "loss": 1.7959,
+      "step": 57400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.742151638721333e-05,
+      "loss": 1.8272,
+      "step": 57500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.741608800066009e-05,
+      "loss": 1.7942,
+      "step": 57600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.741065961410686e-05,
+      "loss": 1.8021,
+      "step": 57700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.740523122755363e-05,
+      "loss": 1.8274,
+      "step": 57800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.739980284100039e-05,
+      "loss": 1.7891,
+      "step": 57900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.739437445444715e-05,
+      "loss": 1.7906,
+      "step": 58000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.738894606789392e-05,
+      "loss": 1.7785,
+      "step": 58100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.738351768134068e-05,
+      "loss": 1.7844,
+      "step": 58200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7378089294787445e-05,
+      "loss": 1.7928,
+      "step": 58300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7372660908234214e-05,
+      "loss": 1.7681,
+      "step": 58400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7367232521680975e-05,
+      "loss": 1.7946,
+      "step": 58500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7361804135127744e-05,
+      "loss": 1.7686,
+      "step": 58600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7356375748574506e-05,
+      "loss": 1.7791,
+      "step": 58700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7350947362021275e-05,
+      "loss": 1.8047,
+      "step": 58800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7345518975468037e-05,
+      "loss": 1.7986,
+      "step": 58900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7340090588914805e-05,
+      "loss": 1.8107,
+      "step": 59000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.733466220236157e-05,
+      "loss": 1.7675,
+      "step": 59100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7329233815808336e-05,
+      "loss": 1.7916,
+      "step": 59200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.73238054292551e-05,
+      "loss": 1.7797,
+      "step": 59300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.731837704270186e-05,
+      "loss": 1.7827,
+      "step": 59400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.731294865614863e-05,
+      "loss": 1.7766,
+      "step": 59500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.730752026959539e-05,
+      "loss": 1.7927,
+      "step": 59600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.730209188304215e-05,
+      "loss": 1.7896,
+      "step": 59700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.729666349648893e-05,
+      "loss": 1.7775,
+      "step": 59800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.729123510993569e-05,
+      "loss": 1.7838,
+      "step": 59900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.728580672338245e-05,
+      "loss": 1.7729,
+      "step": 60000
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 1.5760066509246826,
+      "eval_runtime": 18.8895,
+      "eval_samples_per_second": 529.396,
+      "eval_steps_per_second": 16.57,
+      "step": 60000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.728037833682922e-05,
+      "loss": 1.822,
+      "step": 60100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.727494995027598e-05,
+      "loss": 1.7779,
+      "step": 60200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7269521563722743e-05,
+      "loss": 1.7766,
+      "step": 60300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.726409317716951e-05,
+      "loss": 1.7906,
+      "step": 60400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7258664790616274e-05,
+      "loss": 1.8065,
+      "step": 60500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.725323640406304e-05,
+      "loss": 1.7614,
+      "step": 60600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7247808017509805e-05,
+      "loss": 1.7844,
+      "step": 60700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.724237963095657e-05,
+      "loss": 1.7864,
+      "step": 60800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7236951244403335e-05,
+      "loss": 1.7761,
+      "step": 60900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7231522857850104e-05,
+      "loss": 1.7707,
+      "step": 61000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7226094471296866e-05,
+      "loss": 1.8049,
+      "step": 61100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7220666084743634e-05,
+      "loss": 1.786,
+      "step": 61200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7215237698190396e-05,
+      "loss": 1.8102,
+      "step": 61300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.720980931163716e-05,
+      "loss": 1.7937,
+      "step": 61400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7204380925083927e-05,
+      "loss": 1.7523,
+      "step": 61500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.719895253853069e-05,
+      "loss": 1.7976,
+      "step": 61600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.719352415197745e-05,
+      "loss": 1.7951,
+      "step": 61700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.718809576542422e-05,
+      "loss": 1.767,
+      "step": 61800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.718266737887099e-05,
+      "loss": 1.7567,
+      "step": 61900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.717723899231775e-05,
+      "loss": 1.8087,
+      "step": 62000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.717181060576452e-05,
+      "loss": 1.8043,
+      "step": 62100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.716638221921128e-05,
+      "loss": 1.7823,
+      "step": 62200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.716095383265804e-05,
+      "loss": 1.7519,
+      "step": 62300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.715552544610481e-05,
+      "loss": 1.7821,
+      "step": 62400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.715009705955157e-05,
+      "loss": 1.7673,
+      "step": 62500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.714466867299834e-05,
+      "loss": 1.7676,
+      "step": 62600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.71392402864451e-05,
+      "loss": 1.7674,
+      "step": 62700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7133811899891865e-05,
+      "loss": 1.7856,
+      "step": 62800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7128383513338634e-05,
+      "loss": 1.791,
+      "step": 62900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.71229551267854e-05,
+      "loss": 1.7697,
+      "step": 63000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7117526740232164e-05,
+      "loss": 1.7897,
+      "step": 63100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.711209835367893e-05,
+      "loss": 1.7579,
+      "step": 63200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7106669967125695e-05,
+      "loss": 1.7752,
+      "step": 63300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7101241580572456e-05,
+      "loss": 1.772,
+      "step": 63400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7095813194019225e-05,
+      "loss": 1.7648,
+      "step": 63500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.709038480746599e-05,
+      "loss": 1.7699,
+      "step": 63600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.708495642091275e-05,
+      "loss": 1.7708,
+      "step": 63700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.707952803435952e-05,
+      "loss": 1.7878,
+      "step": 63800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.707409964780628e-05,
+      "loss": 1.805,
+      "step": 63900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.706867126125305e-05,
+      "loss": 1.774,
+      "step": 64000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.706324287469982e-05,
+      "loss": 1.7729,
+      "step": 64100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.705781448814658e-05,
+      "loss": 1.759,
+      "step": 64200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.705238610159334e-05,
+      "loss": 1.7649,
+      "step": 64300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.704695771504011e-05,
+      "loss": 1.7887,
+      "step": 64400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.704152932848687e-05,
+      "loss": 1.8055,
+      "step": 64500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.703610094193364e-05,
+      "loss": 1.7912,
+      "step": 64600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.70306725553804e-05,
+      "loss": 1.759,
+      "step": 64700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.702524416882716e-05,
+      "loss": 1.7841,
+      "step": 64800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.701981578227393e-05,
+      "loss": 1.7759,
+      "step": 64900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.70143873957207e-05,
+      "loss": 1.7306,
+      "step": 65000
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 1.5630521774291992,
+      "eval_runtime": 18.8835,
+      "eval_samples_per_second": 529.564,
+      "eval_steps_per_second": 16.575,
+      "step": 65000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.700895900916746e-05,
+      "loss": 1.7537,
+      "step": 65100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.700353062261423e-05,
+      "loss": 1.7783,
+      "step": 65200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.699810223606099e-05,
+      "loss": 1.7818,
+      "step": 65300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.6992673849507755e-05,
+      "loss": 1.7749,
+      "step": 65400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.6987245462954524e-05,
+      "loss": 1.7662,
+      "step": 65500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.6981817076401285e-05,
+      "loss": 1.7677,
+      "step": 65600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.697638868984805e-05,
+      "loss": 1.7521,
+      "step": 65700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.6970960303294816e-05,
+      "loss": 1.7751,
+      "step": 65800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.696553191674158e-05,
+      "loss": 1.7846,
+      "step": 65900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.696010353018834e-05,
+      "loss": 1.7987,
+      "step": 66000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.6954675143635115e-05,
+      "loss": 1.7691,
+      "step": 66100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.694924675708188e-05,
+      "loss": 1.7955,
+      "step": 66200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.694381837052864e-05,
+      "loss": 1.8008,
+      "step": 66300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.693838998397541e-05,
+      "loss": 1.77,
+      "step": 66400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.693296159742217e-05,
+      "loss": 1.7612,
+      "step": 66500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.692753321086894e-05,
+      "loss": 1.7797,
+      "step": 66600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.69221048243157e-05,
+      "loss": 1.8038,
+      "step": 66700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.691667643776246e-05,
+      "loss": 1.7792,
+      "step": 66800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.691124805120923e-05,
+      "loss": 1.7614,
+      "step": 66900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.690581966465599e-05,
+      "loss": 1.7548,
+      "step": 67000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.690039127810276e-05,
+      "loss": 1.773,
+      "step": 67100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.689496289154953e-05,
+      "loss": 1.7859,
+      "step": 67200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.688953450499629e-05,
+      "loss": 1.7366,
+      "step": 67300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.6884106118443053e-05,
+      "loss": 1.7685,
+      "step": 67400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.687867773188982e-05,
+      "loss": 1.7628,
+      "step": 67500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.6873249345336584e-05,
+      "loss": 1.7561,
+      "step": 67600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.6867820958783346e-05,
+      "loss": 1.8026,
+      "step": 67700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.6862392572230114e-05,
+      "loss": 1.7847,
+      "step": 67800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.6856964185676876e-05,
+      "loss": 1.7646,
+      "step": 67900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.685153579912364e-05,
+      "loss": 1.7906,
+      "step": 68000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.684610741257041e-05,
+      "loss": 1.7476,
+      "step": 68100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.6840679026017176e-05,
+      "loss": 1.7534,
+      "step": 68200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.683525063946394e-05,
+      "loss": 1.7639,
+      "step": 68300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.6829822252910706e-05,
+      "loss": 1.7499,
+      "step": 68400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.682439386635747e-05,
+      "loss": 1.7743,
+      "step": 68500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.6818965479804237e-05,
+      "loss": 1.7483,
+      "step": 68600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.6813537093251e-05,
+      "loss": 1.7807,
+      "step": 68700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.680810870669776e-05,
+      "loss": 1.7912,
+      "step": 68800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.680268032014453e-05,
+      "loss": 1.7632,
+      "step": 68900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.679725193359129e-05,
+      "loss": 1.762,
+      "step": 69000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.679182354703805e-05,
+      "loss": 1.7498,
+      "step": 69100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.678639516048483e-05,
+      "loss": 1.7597,
+      "step": 69200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.678096677393159e-05,
+      "loss": 1.7851,
+      "step": 69300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.677553838737835e-05,
+      "loss": 1.7746,
+      "step": 69400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.677011000082512e-05,
+      "loss": 1.7738,
+      "step": 69500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.676468161427188e-05,
+      "loss": 1.7909,
+      "step": 69600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.6759253227718644e-05,
+      "loss": 1.768,
+      "step": 69700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.675382484116541e-05,
+      "loss": 1.7782,
+      "step": 69800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.6748396454612175e-05,
+      "loss": 1.7663,
+      "step": 69900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.674296806805894e-05,
+      "loss": 1.7202,
+      "step": 70000
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 1.5640525817871094,
+      "eval_runtime": 18.9125,
+      "eval_samples_per_second": 528.752,
+      "eval_steps_per_second": 16.55,
+      "step": 70000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.6737539681505705e-05,
+      "loss": 1.7697,
+      "step": 70100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.673211129495247e-05,
+      "loss": 1.7581,
+      "step": 70200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.6726682908399236e-05,
+      "loss": 1.7291,
+      "step": 70300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.6721254521846005e-05,
+      "loss": 1.7576,
+      "step": 70400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.6715826135292766e-05,
+      "loss": 1.7384,
+      "step": 70500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.671039774873953e-05,
+      "loss": 1.7676,
+      "step": 70600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.67049693621863e-05,
+      "loss": 1.7576,
+      "step": 70700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.669954097563306e-05,
+      "loss": 1.7404,
+      "step": 70800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.669411258907983e-05,
+      "loss": 1.7635,
+      "step": 70900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.668868420252659e-05,
+      "loss": 1.7898,
+      "step": 71000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.668325581597335e-05,
+      "loss": 1.7269,
+      "step": 71100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.667782742942012e-05,
+      "loss": 1.756,
+      "step": 71200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.667239904286689e-05,
+      "loss": 1.7796,
+      "step": 71300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.666697065631365e-05,
+      "loss": 1.7805,
+      "step": 71400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.666154226976042e-05,
+      "loss": 1.7324,
+      "step": 71500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.665611388320718e-05,
+      "loss": 1.7412,
+      "step": 71600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.665068549665394e-05,
+      "loss": 1.7363,
+      "step": 71700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.664525711010071e-05,
+      "loss": 1.7615,
+      "step": 71800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.663982872354747e-05,
+      "loss": 1.7356,
+      "step": 71900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6634400336994235e-05,
+      "loss": 1.7364,
+      "step": 72000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6628971950441004e-05,
+      "loss": 1.7587,
+      "step": 72100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6623543563887766e-05,
+      "loss": 1.742,
+      "step": 72200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6618115177334534e-05,
+      "loss": 1.7618,
+      "step": 72300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.66126867907813e-05,
+      "loss": 1.7374,
+      "step": 72400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6607258404228065e-05,
+      "loss": 1.7804,
+      "step": 72500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.660183001767483e-05,
+      "loss": 1.7173,
+      "step": 72600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6596401631121595e-05,
+      "loss": 1.7459,
+      "step": 72700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.659097324456836e-05,
+      "loss": 1.7441,
+      "step": 72800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6585544858015126e-05,
+      "loss": 1.7598,
+      "step": 72900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.658011647146189e-05,
+      "loss": 1.7476,
+      "step": 73000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.657468808490865e-05,
+      "loss": 1.7596,
+      "step": 73100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.656925969835542e-05,
+      "loss": 1.7438,
+      "step": 73200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.656383131180218e-05,
+      "loss": 1.7752,
+      "step": 73300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.655840292524895e-05,
+      "loss": 1.7214,
+      "step": 73400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.655297453869572e-05,
+      "loss": 1.7437,
+      "step": 73500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.654754615214248e-05,
+      "loss": 1.7531,
+      "step": 73600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.654211776558924e-05,
+      "loss": 1.746,
+      "step": 73700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.653668937903601e-05,
+      "loss": 1.748,
+      "step": 73800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.653126099248277e-05,
+      "loss": 1.7357,
+      "step": 73900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6525832605929534e-05,
+      "loss": 1.7325,
+      "step": 74000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.65204042193763e-05,
+      "loss": 1.7155,
+      "step": 74100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6514975832823064e-05,
+      "loss": 1.7436,
+      "step": 74200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.650954744626983e-05,
+      "loss": 1.7245,
+      "step": 74300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6504119059716595e-05,
+      "loss": 1.7332,
+      "step": 74400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6498690673163363e-05,
+      "loss": 1.7614,
+      "step": 74500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6493262286610125e-05,
+      "loss": 1.7565,
+      "step": 74600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6487833900056894e-05,
+      "loss": 1.7641,
+      "step": 74700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6482405513503656e-05,
+      "loss": 1.7415,
+      "step": 74800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6476977126950424e-05,
+      "loss": 1.7563,
+      "step": 74900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6471548740397186e-05,
+      "loss": 1.7745,
+      "step": 75000
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 1.5477814674377441,
+      "eval_runtime": 18.8766,
+      "eval_samples_per_second": 529.758,
+      "eval_steps_per_second": 16.581,
+      "step": 75000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.646612035384395e-05,
+      "loss": 1.7393,
+      "step": 75100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.646069196729072e-05,
+      "loss": 1.752,
+      "step": 75200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.645526358073748e-05,
+      "loss": 1.7421,
+      "step": 75300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.644983519418424e-05,
+      "loss": 1.7443,
+      "step": 75400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6444406807631016e-05,
+      "loss": 1.7661,
+      "step": 75500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.643897842107778e-05,
+      "loss": 1.7527,
+      "step": 75600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.643355003452454e-05,
+      "loss": 1.7628,
+      "step": 75700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.642812164797131e-05,
+      "loss": 1.752,
+      "step": 75800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.642269326141807e-05,
+      "loss": 1.7392,
+      "step": 75900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.641726487486483e-05,
+      "loss": 1.7515,
+      "step": 76000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.64118364883116e-05,
+      "loss": 1.7412,
+      "step": 76100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.640640810175836e-05,
+      "loss": 1.7418,
+      "step": 76200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.640097971520513e-05,
+      "loss": 1.7248,
+      "step": 76300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.639555132865189e-05,
+      "loss": 1.7437,
+      "step": 76400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6390122942098655e-05,
+      "loss": 1.7433,
+      "step": 76500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6384694555545424e-05,
+      "loss": 1.7226,
+      "step": 76600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.637926616899219e-05,
+      "loss": 1.7266,
+      "step": 76700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6373837782438954e-05,
+      "loss": 1.7332,
+      "step": 76800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.636840939588572e-05,
+      "loss": 1.7427,
+      "step": 76900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6362981009332485e-05,
+      "loss": 1.7679,
+      "step": 77000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.635755262277925e-05,
+      "loss": 1.7527,
+      "step": 77100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6352124236226015e-05,
+      "loss": 1.7443,
+      "step": 77200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.634669584967278e-05,
+      "loss": 1.7738,
+      "step": 77300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.634126746311954e-05,
+      "loss": 1.735,
+      "step": 77400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.633583907656631e-05,
+      "loss": 1.7555,
+      "step": 77500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6330410690013076e-05,
+      "loss": 1.7362,
+      "step": 77600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.632498230345984e-05,
+      "loss": 1.7026,
+      "step": 77700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.631955391690661e-05,
+      "loss": 1.7306,
+      "step": 77800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.631412553035337e-05,
+      "loss": 1.7717,
+      "step": 77900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.630869714380013e-05,
+      "loss": 1.7467,
+      "step": 78000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.63032687572469e-05,
+      "loss": 1.7282,
+      "step": 78100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.629784037069366e-05,
+      "loss": 1.7238,
+      "step": 78200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.629241198414043e-05,
+      "loss": 1.7432,
+      "step": 78300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.628698359758719e-05,
+      "loss": 1.7318,
+      "step": 78400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6281555211033954e-05,
+      "loss": 1.7544,
+      "step": 78500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.627612682448072e-05,
+      "loss": 1.7452,
+      "step": 78600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.627069843792749e-05,
+      "loss": 1.757,
+      "step": 78700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.626527005137425e-05,
+      "loss": 1.754,
+      "step": 78800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.625984166482102e-05,
+      "loss": 1.7073,
+      "step": 78900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.625441327826778e-05,
+      "loss": 1.7104,
+      "step": 79000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6248984891714545e-05,
+      "loss": 1.7339,
+      "step": 79100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6243556505161314e-05,
+      "loss": 1.722,
+      "step": 79200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6238128118608076e-05,
+      "loss": 1.7456,
+      "step": 79300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.623269973205484e-05,
+      "loss": 1.7273,
+      "step": 79400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6227271345501606e-05,
+      "loss": 1.7432,
+      "step": 79500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.622184295894837e-05,
+      "loss": 1.76,
+      "step": 79600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.621641457239514e-05,
+      "loss": 1.7446,
+      "step": 79700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6210986185841905e-05,
+      "loss": 1.7589,
+      "step": 79800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.620555779928867e-05,
+      "loss": 1.7662,
+      "step": 79900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.620012941273543e-05,
+      "loss": 1.7185,
+      "step": 80000
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 1.5425916910171509,
+      "eval_runtime": 18.916,
+      "eval_samples_per_second": 528.654,
+      "eval_steps_per_second": 16.547,
+      "step": 80000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.61947010261822e-05,
+      "loss": 1.7206,
+      "step": 80100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.618927263962896e-05,
+      "loss": 1.7431,
+      "step": 80200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.618384425307573e-05,
+      "loss": 1.7112,
+      "step": 80300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.617841586652249e-05,
+      "loss": 1.7242,
+      "step": 80400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.617298747996925e-05,
+      "loss": 1.7106,
+      "step": 80500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.616755909341602e-05,
+      "loss": 1.7363,
+      "step": 80600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.616213070686279e-05,
+      "loss": 1.7411,
+      "step": 80700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.615670232030955e-05,
+      "loss": 1.7663,
+      "step": 80800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.615127393375632e-05,
+      "loss": 1.7612,
+      "step": 80900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.614584554720308e-05,
+      "loss": 1.7457,
+      "step": 81000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6140417160649844e-05,
+      "loss": 1.7319,
+      "step": 81100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.613498877409661e-05,
+      "loss": 1.744,
+      "step": 81200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6129560387543374e-05,
+      "loss": 1.7076,
+      "step": 81300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6124132000990136e-05,
+      "loss": 1.7594,
+      "step": 81400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6118703614436905e-05,
+      "loss": 1.7443,
+      "step": 81500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6113275227883667e-05,
+      "loss": 1.7158,
+      "step": 81600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6107846841330435e-05,
+      "loss": 1.7315,
+      "step": 81700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6102418454777204e-05,
+      "loss": 1.756,
+      "step": 81800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6096990068223966e-05,
+      "loss": 1.7466,
+      "step": 81900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.609156168167073e-05,
+      "loss": 1.7233,
+      "step": 82000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6086133295117496e-05,
+      "loss": 1.736,
+      "step": 82100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.608070490856426e-05,
+      "loss": 1.7329,
+      "step": 82200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.607527652201103e-05,
+      "loss": 1.7311,
+      "step": 82300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.606984813545779e-05,
+      "loss": 1.7318,
+      "step": 82400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.606441974890455e-05,
+      "loss": 1.7485,
+      "step": 82500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.605899136235132e-05,
+      "loss": 1.6999,
+      "step": 82600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.605356297579808e-05,
+      "loss": 1.7136,
+      "step": 82700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.604813458924485e-05,
+      "loss": 1.724,
+      "step": 82800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.604270620269162e-05,
+      "loss": 1.7513,
+      "step": 82900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.603727781613838e-05,
+      "loss": 1.7232,
+      "step": 83000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.603184942958514e-05,
+      "loss": 1.7535,
+      "step": 83100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.602642104303191e-05,
+      "loss": 1.7338,
+      "step": 83200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.602099265647867e-05,
+      "loss": 1.7305,
+      "step": 83300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6015564269925435e-05,
+      "loss": 1.7295,
+      "step": 83400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.60101358833722e-05,
+      "loss": 1.7555,
+      "step": 83500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6004707496818965e-05,
+      "loss": 1.7299,
+      "step": 83600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5999279110265734e-05,
+      "loss": 1.7132,
+      "step": 83700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5993850723712496e-05,
+      "loss": 1.7587,
+      "step": 83800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5988422337159264e-05,
+      "loss": 1.7486,
+      "step": 83900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5982993950606026e-05,
+      "loss": 1.6919,
+      "step": 84000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5977565564052795e-05,
+      "loss": 1.7116,
+      "step": 84100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.597213717749956e-05,
+      "loss": 1.7302,
+      "step": 84200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5966708790946325e-05,
+      "loss": 1.7044,
+      "step": 84300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.596128040439309e-05,
+      "loss": 1.7475,
+      "step": 84400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.595585201783985e-05,
+      "loss": 1.7357,
+      "step": 84500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.595042363128662e-05,
+      "loss": 1.7234,
+      "step": 84600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.594499524473338e-05,
+      "loss": 1.7307,
+      "step": 84700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.593956685818014e-05,
+      "loss": 1.7138,
+      "step": 84800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.593413847162692e-05,
+      "loss": 1.7437,
+      "step": 84900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.592871008507368e-05,
+      "loss": 1.761,
+      "step": 85000
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 1.5349645614624023,
+      "eval_runtime": 18.9426,
+      "eval_samples_per_second": 527.911,
+      "eval_steps_per_second": 16.524,
+      "step": 85000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.592328169852044e-05,
+      "loss": 1.7331,
+      "step": 85100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.591785331196721e-05,
+      "loss": 1.7209,
+      "step": 85200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.591242492541397e-05,
+      "loss": 1.7032,
+      "step": 85300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.590699653886073e-05,
+      "loss": 1.7575,
+      "step": 85400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.59015681523075e-05,
+      "loss": 1.7372,
+      "step": 85500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5896139765754264e-05,
+      "loss": 1.7554,
+      "step": 85600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.589071137920103e-05,
+      "loss": 1.7196,
+      "step": 85700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5885282992647794e-05,
+      "loss": 1.6947,
+      "step": 85800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5879854606094556e-05,
+      "loss": 1.7231,
+      "step": 85900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5874426219541325e-05,
+      "loss": 1.7123,
+      "step": 86000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.586899783298809e-05,
+      "loss": 1.7224,
+      "step": 86100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5863569446434855e-05,
+      "loss": 1.7402,
+      "step": 86200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5858141059881624e-05,
+      "loss": 1.7189,
+      "step": 86300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5852712673328386e-05,
+      "loss": 1.7363,
+      "step": 86400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.584728428677515e-05,
+      "loss": 1.7376,
+      "step": 86500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5841855900221916e-05,
+      "loss": 1.739,
+      "step": 86600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.583642751366868e-05,
+      "loss": 1.7794,
+      "step": 86700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.583099912711544e-05,
+      "loss": 1.7424,
+      "step": 86800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.582557074056221e-05,
+      "loss": 1.7139,
+      "step": 86900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.582014235400898e-05,
+      "loss": 1.7696,
+      "step": 87000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.581471396745574e-05,
+      "loss": 1.7081,
+      "step": 87100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.580928558090251e-05,
+      "loss": 1.7361,
+      "step": 87200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.580385719434927e-05,
+      "loss": 1.7095,
+      "step": 87300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.579842880779603e-05,
+      "loss": 1.7303,
+      "step": 87400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.57930004212428e-05,
+      "loss": 1.6847,
+      "step": 87500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.578757203468956e-05,
+      "loss": 1.7079,
+      "step": 87600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.578214364813633e-05,
+      "loss": 1.7344,
+      "step": 87700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.577671526158309e-05,
+      "loss": 1.731,
+      "step": 87800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.5771286875029854e-05,
+      "loss": 1.7397,
+      "step": 87900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.576585848847662e-05,
+      "loss": 1.7308,
+      "step": 88000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.576043010192339e-05,
+      "loss": 1.7609,
+      "step": 88100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.5755001715370154e-05,
+      "loss": 1.7123,
+      "step": 88200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.574957332881692e-05,
+      "loss": 1.7093,
+      "step": 88300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.5744144942263684e-05,
+      "loss": 1.7262,
+      "step": 88400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.5738716555710446e-05,
+      "loss": 1.7057,
+      "step": 88500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.5733288169157215e-05,
+      "loss": 1.7473,
+      "step": 88600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.5727859782603977e-05,
+      "loss": 1.7331,
+      "step": 88700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.572243139605074e-05,
+      "loss": 1.7549,
+      "step": 88800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.571700300949751e-05,
+      "loss": 1.7222,
+      "step": 88900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.571157462294427e-05,
+      "loss": 1.6896,
+      "step": 89000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.570614623639104e-05,
+      "loss": 1.716,
+      "step": 89100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.5700717849837806e-05,
+      "loss": 1.7272,
+      "step": 89200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.569528946328457e-05,
+      "loss": 1.692,
+      "step": 89300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.568986107673133e-05,
+      "loss": 1.7422,
+      "step": 89400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.56844326901781e-05,
+      "loss": 1.7287,
+      "step": 89500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.567900430362486e-05,
+      "loss": 1.7368,
+      "step": 89600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.567357591707163e-05,
+      "loss": 1.6947,
+      "step": 89700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.566814753051839e-05,
+      "loss": 1.7064,
+      "step": 89800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.566271914396515e-05,
+      "loss": 1.7035,
+      "step": 89900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.565729075741192e-05,
+      "loss": 1.7442,
+      "step": 90000
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 1.5332447290420532,
+      "eval_runtime": 18.9917,
+      "eval_samples_per_second": 526.545,
+      "eval_steps_per_second": 16.481,
+      "step": 90000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5651862370858683e-05,
+      "loss": 1.7316,
+      "step": 90100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.564643398430545e-05,
+      "loss": 1.6982,
+      "step": 90200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.564100559775222e-05,
+      "loss": 1.7471,
+      "step": 90300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.563557721119898e-05,
+      "loss": 1.7449,
+      "step": 90400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5630148824645745e-05,
+      "loss": 1.7299,
+      "step": 90500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.562472043809251e-05,
+      "loss": 1.7059,
+      "step": 90600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5619292051539275e-05,
+      "loss": 1.7734,
+      "step": 90700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.561386366498604e-05,
+      "loss": 1.7181,
+      "step": 90800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5608435278432806e-05,
+      "loss": 1.7107,
+      "step": 90900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.560300689187957e-05,
+      "loss": 1.7182,
+      "step": 91000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.559757850532633e-05,
+      "loss": 1.6972,
+      "step": 91100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5592150118773105e-05,
+      "loss": 1.7136,
+      "step": 91200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.558672173221987e-05,
+      "loss": 1.71,
+      "step": 91300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.558129334566663e-05,
+      "loss": 1.7111,
+      "step": 91400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.55758649591134e-05,
+      "loss": 1.6984,
+      "step": 91500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.557043657256016e-05,
+      "loss": 1.7014,
+      "step": 91600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.556500818600693e-05,
+      "loss": 1.7238,
+      "step": 91700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.555957979945369e-05,
+      "loss": 1.7368,
+      "step": 91800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.555415141290045e-05,
+      "loss": 1.7427,
+      "step": 91900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.554872302634722e-05,
+      "loss": 1.7477,
+      "step": 92000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.554329463979398e-05,
+      "loss": 1.707,
+      "step": 92100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5537866253240744e-05,
+      "loss": 1.6978,
+      "step": 92200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.553243786668752e-05,
+      "loss": 1.7144,
+      "step": 92300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.552700948013428e-05,
+      "loss": 1.702,
+      "step": 92400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.552158109358104e-05,
+      "loss": 1.7267,
+      "step": 92500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.551615270702781e-05,
+      "loss": 1.7618,
+      "step": 92600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5510724320474574e-05,
+      "loss": 1.7011,
+      "step": 92700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5505295933921335e-05,
+      "loss": 1.6948,
+      "step": 92800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5499867547368104e-05,
+      "loss": 1.7423,
+      "step": 92900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5494439160814866e-05,
+      "loss": 1.7146,
+      "step": 93000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.548901077426163e-05,
+      "loss": 1.6756,
+      "step": 93100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5483582387708396e-05,
+      "loss": 1.7365,
+      "step": 93200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5478154001155165e-05,
+      "loss": 1.7188,
+      "step": 93300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.547272561460193e-05,
+      "loss": 1.7,
+      "step": 93400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5467297228048696e-05,
+      "loss": 1.7037,
+      "step": 93500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.546186884149546e-05,
+      "loss": 1.7251,
+      "step": 93600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5456440454942226e-05,
+      "loss": 1.7172,
+      "step": 93700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.545101206838899e-05,
+      "loss": 1.707,
+      "step": 93800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.544558368183575e-05,
+      "loss": 1.7124,
+      "step": 93900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.544015529528252e-05,
+      "loss": 1.7222,
+      "step": 94000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.543472690872928e-05,
+      "loss": 1.7044,
+      "step": 94100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.542929852217604e-05,
+      "loss": 1.7188,
+      "step": 94200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.542387013562281e-05,
+      "loss": 1.6926,
+      "step": 94300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.541844174906958e-05,
+      "loss": 1.733,
+      "step": 94400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.541301336251634e-05,
+      "loss": 1.7195,
+      "step": 94500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.540758497596311e-05,
+      "loss": 1.7094,
+      "step": 94600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.540215658940987e-05,
+      "loss": 1.6916,
+      "step": 94700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5396728202856634e-05,
+      "loss": 1.7094,
+      "step": 94800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.53912998163034e-05,
+      "loss": 1.7307,
+      "step": 94900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5385871429750164e-05,
+      "loss": 1.7251,
+      "step": 95000
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 1.524511694908142,
+      "eval_runtime": 19.0193,
+      "eval_samples_per_second": 525.783,
+      "eval_steps_per_second": 16.457,
+      "step": 95000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5380443043196926e-05,
+      "loss": 1.7106,
+      "step": 95100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5375014656643695e-05,
+      "loss": 1.713,
+      "step": 95200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.536958627009046e-05,
+      "loss": 1.7098,
+      "step": 95300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5364157883537225e-05,
+      "loss": 1.6978,
+      "step": 95400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5358729496983994e-05,
+      "loss": 1.7265,
+      "step": 95500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5353301110430756e-05,
+      "loss": 1.7404,
+      "step": 95600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5347872723877525e-05,
+      "loss": 1.7086,
+      "step": 95700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5342444337324287e-05,
+      "loss": 1.6896,
+      "step": 95800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.533701595077105e-05,
+      "loss": 1.7227,
+      "step": 95900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.533158756421782e-05,
+      "loss": 1.7086,
+      "step": 96000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.532615917766458e-05,
+      "loss": 1.6899,
+      "step": 96100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.532073079111134e-05,
+      "loss": 1.7089,
+      "step": 96200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.531530240455811e-05,
+      "loss": 1.6741,
+      "step": 96300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.530987401800487e-05,
+      "loss": 1.6886,
+      "step": 96400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.530444563145164e-05,
+      "loss": 1.7238,
+      "step": 96500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.529901724489841e-05,
+      "loss": 1.7371,
+      "step": 96600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.529358885834517e-05,
+      "loss": 1.7138,
+      "step": 96700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.528816047179193e-05,
+      "loss": 1.7232,
+      "step": 96800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.52827320852387e-05,
+      "loss": 1.7497,
+      "step": 96900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.527730369868546e-05,
+      "loss": 1.7202,
+      "step": 97000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5271875312132225e-05,
+      "loss": 1.7008,
+      "step": 97100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5266446925578993e-05,
+      "loss": 1.6832,
+      "step": 97200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5261018539025755e-05,
+      "loss": 1.6893,
+      "step": 97300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5255590152472524e-05,
+      "loss": 1.7067,
+      "step": 97400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.525016176591929e-05,
+      "loss": 1.7119,
+      "step": 97500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5244733379366054e-05,
+      "loss": 1.7267,
+      "step": 97600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.523930499281282e-05,
+      "loss": 1.6911,
+      "step": 97700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5233876606259585e-05,
+      "loss": 1.7164,
+      "step": 97800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.522844821970635e-05,
+      "loss": 1.6748,
+      "step": 97900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5223019833153116e-05,
+      "loss": 1.7339,
+      "step": 98000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.521759144659988e-05,
+      "loss": 1.692,
+      "step": 98100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.521216306004664e-05,
+      "loss": 1.6869,
+      "step": 98200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.520673467349341e-05,
+      "loss": 1.7191,
+      "step": 98300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.520130628694017e-05,
+      "loss": 1.7228,
+      "step": 98400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.519587790038693e-05,
+      "loss": 1.7141,
+      "step": 98500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.519044951383371e-05,
+      "loss": 1.7032,
+      "step": 98600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.518502112728047e-05,
+      "loss": 1.7326,
+      "step": 98700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.517959274072723e-05,
+      "loss": 1.7107,
+      "step": 98800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5174164354174e-05,
+      "loss": 1.6987,
+      "step": 98900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.516873596762076e-05,
+      "loss": 1.697,
+      "step": 99000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.516330758106752e-05,
+      "loss": 1.7505,
+      "step": 99100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.515787919451429e-05,
+      "loss": 1.7094,
+      "step": 99200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5152450807961054e-05,
+      "loss": 1.6878,
+      "step": 99300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.514702242140782e-05,
+      "loss": 1.6845,
+      "step": 99400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5141594034854584e-05,
+      "loss": 1.6908,
+      "step": 99500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.513616564830135e-05,
+      "loss": 1.7112,
+      "step": 99600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.513073726174812e-05,
+      "loss": 1.7154,
+      "step": 99700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5125308875194884e-05,
+      "loss": 1.7267,
+      "step": 99800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5119880488641645e-05,
+      "loss": 1.7002,
+      "step": 99900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5114452102088414e-05,
+      "loss": 1.7055,
+      "step": 100000
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 1.520733118057251,
+      "eval_runtime": 19.0032,
+      "eval_samples_per_second": 526.228,
+      "eval_steps_per_second": 16.471,
+      "step": 100000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5109023715535176e-05,
+      "loss": 1.7008,
+      "step": 100100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.510359532898194e-05,
+      "loss": 1.728,
+      "step": 100200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5098166942428706e-05,
+      "loss": 1.7455,
+      "step": 100300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.509273855587547e-05,
+      "loss": 1.7056,
+      "step": 100400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.508731016932223e-05,
+      "loss": 1.722,
+      "step": 100500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5081881782769e-05,
+      "loss": 1.7009,
+      "step": 100600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.507645339621577e-05,
+      "loss": 1.6849,
+      "step": 100700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.507102500966253e-05,
+      "loss": 1.7089,
+      "step": 100800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.50655966231093e-05,
+      "loss": 1.7269,
+      "step": 100900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.506016823655606e-05,
+      "loss": 1.6955,
+      "step": 101000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.505473985000282e-05,
+      "loss": 1.7307,
+      "step": 101100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.504931146344959e-05,
+      "loss": 1.6969,
+      "step": 101200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.504388307689635e-05,
+      "loss": 1.7104,
+      "step": 101300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.503845469034312e-05,
+      "loss": 1.7175,
+      "step": 101400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.503302630378988e-05,
+      "loss": 1.7054,
+      "step": 101500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5027597917236645e-05,
+      "loss": 1.6907,
+      "step": 101600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.502216953068342e-05,
+      "loss": 1.7202,
+      "step": 101700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.501674114413018e-05,
+      "loss": 1.7127,
+      "step": 101800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5011312757576944e-05,
+      "loss": 1.7052,
+      "step": 101900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.500588437102371e-05,
+      "loss": 1.7044,
+      "step": 102000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5000455984470474e-05,
+      "loss": 1.7022,
+      "step": 102100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4995027597917236e-05,
+      "loss": 1.7027,
+      "step": 102200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4989599211364005e-05,
+      "loss": 1.6933,
+      "step": 102300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.498417082481077e-05,
+      "loss": 1.7638,
+      "step": 102400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.497874243825753e-05,
+      "loss": 1.7242,
+      "step": 102500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.49733140517043e-05,
+      "loss": 1.7152,
+      "step": 102600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4967885665151066e-05,
+      "loss": 1.7038,
+      "step": 102700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.496245727859783e-05,
+      "loss": 1.7027,
+      "step": 102800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4957028892044597e-05,
+      "loss": 1.7215,
+      "step": 102900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.495160050549136e-05,
+      "loss": 1.6996,
+      "step": 103000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.494617211893812e-05,
+      "loss": 1.6896,
+      "step": 103100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.494074373238489e-05,
+      "loss": 1.6978,
+      "step": 103200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.493531534583165e-05,
+      "loss": 1.6943,
+      "step": 103300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.492988695927842e-05,
+      "loss": 1.7214,
+      "step": 103400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.492445857272518e-05,
+      "loss": 1.717,
+      "step": 103500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.491903018617194e-05,
+      "loss": 1.6582,
+      "step": 103600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.491360179961871e-05,
+      "loss": 1.6872,
+      "step": 103700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.490817341306548e-05,
+      "loss": 1.7238,
+      "step": 103800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.490274502651224e-05,
+      "loss": 1.6819,
+      "step": 103900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.489731663995901e-05,
+      "loss": 1.6898,
+      "step": 104000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.489188825340577e-05,
+      "loss": 1.6868,
+      "step": 104100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.4886459866852535e-05,
+      "loss": 1.6888,
+      "step": 104200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.4881031480299303e-05,
+      "loss": 1.7046,
+      "step": 104300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.4875603093746065e-05,
+      "loss": 1.6846,
+      "step": 104400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.487017470719283e-05,
+      "loss": 1.6848,
+      "step": 104500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.4864746320639596e-05,
+      "loss": 1.6822,
+      "step": 104600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.485931793408636e-05,
+      "loss": 1.6829,
+      "step": 104700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.4853889547533126e-05,
+      "loss": 1.7006,
+      "step": 104800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.4848461160979895e-05,
+      "loss": 1.6972,
+      "step": 104900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.484303277442666e-05,
+      "loss": 1.7391,
+      "step": 105000
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 1.5192019939422607,
+      "eval_runtime": 18.9573,
+      "eval_samples_per_second": 527.5,
+      "eval_steps_per_second": 16.511,
+      "step": 105000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.483760438787342e-05,
+      "loss": 1.6914,
+      "step": 105100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.483217600132019e-05,
+      "loss": 1.6868,
+      "step": 105200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.482674761476695e-05,
+      "loss": 1.679,
+      "step": 105300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.482131922821372e-05,
+      "loss": 1.6947,
+      "step": 105400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.481589084166048e-05,
+      "loss": 1.683,
+      "step": 105500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.481046245510724e-05,
+      "loss": 1.7186,
+      "step": 105600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.480503406855401e-05,
+      "loss": 1.7041,
+      "step": 105700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.479960568200077e-05,
+      "loss": 1.6946,
+      "step": 105800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.479417729544754e-05,
+      "loss": 1.6408,
+      "step": 105900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.478874890889431e-05,
+      "loss": 1.6982,
+      "step": 106000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.478332052234107e-05,
+      "loss": 1.6851,
+      "step": 106100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.477789213578783e-05,
+      "loss": 1.7197,
+      "step": 106200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.47724637492346e-05,
+      "loss": 1.6963,
+      "step": 106300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.4767035362681364e-05,
+      "loss": 1.7035,
+      "step": 106400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.4761606976128126e-05,
+      "loss": 1.684,
+      "step": 106500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.4756178589574894e-05,
+      "loss": 1.7079,
+      "step": 106600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.4750750203021656e-05,
+      "loss": 1.6791,
+      "step": 106700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.4745321816468425e-05,
+      "loss": 1.6869,
+      "step": 106800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.4739893429915193e-05,
+      "loss": 1.6882,
+      "step": 106900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.4734465043361955e-05,
+      "loss": 1.6805,
+      "step": 107000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.472903665680872e-05,
+      "loss": 1.6941,
+      "step": 107100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.4723608270255486e-05,
+      "loss": 1.7022,
+      "step": 107200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.471817988370225e-05,
+      "loss": 1.6859,
+      "step": 107300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.4712751497149016e-05,
+      "loss": 1.6924,
+      "step": 107400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.470732311059578e-05,
+      "loss": 1.6669,
+      "step": 107500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.470189472404254e-05,
+      "loss": 1.6708,
+      "step": 107600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.469646633748931e-05,
+      "loss": 1.7214,
+      "step": 107700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.469103795093607e-05,
+      "loss": 1.6922,
+      "step": 107800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.468560956438283e-05,
+      "loss": 1.7017,
+      "step": 107900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.468018117782961e-05,
+      "loss": 1.6998,
+      "step": 108000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.467475279127637e-05,
+      "loss": 1.703,
+      "step": 108100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.466932440472313e-05,
+      "loss": 1.6755,
+      "step": 108200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.46638960181699e-05,
+      "loss": 1.7187,
+      "step": 108300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.465846763161666e-05,
+      "loss": 1.6764,
+      "step": 108400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.4653039245063424e-05,
+      "loss": 1.7016,
+      "step": 108500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.464761085851019e-05,
+      "loss": 1.7061,
+      "step": 108600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.4642182471956955e-05,
+      "loss": 1.6835,
+      "step": 108700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.463675408540372e-05,
+      "loss": 1.6994,
+      "step": 108800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.4631325698850485e-05,
+      "loss": 1.6901,
+      "step": 108900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.4625897312297254e-05,
+      "loss": 1.7016,
+      "step": 109000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.4620468925744016e-05,
+      "loss": 1.6906,
+      "step": 109100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.4615040539190784e-05,
+      "loss": 1.6993,
+      "step": 109200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.4609612152637546e-05,
+      "loss": 1.7195,
+      "step": 109300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.4604183766084315e-05,
+      "loss": 1.6942,
+      "step": 109400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.459875537953108e-05,
+      "loss": 1.7097,
+      "step": 109500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.459332699297784e-05,
+      "loss": 1.669,
+      "step": 109600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.458789860642461e-05,
+      "loss": 1.7243,
+      "step": 109700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.458247021987137e-05,
+      "loss": 1.6813,
+      "step": 109800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.457704183331813e-05,
+      "loss": 1.7064,
+      "step": 109900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.45716134467649e-05,
+      "loss": 1.7051,
+      "step": 110000
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 1.51393461227417,
+      "eval_runtime": 19.0208,
+      "eval_samples_per_second": 525.74,
+      "eval_steps_per_second": 16.456,
+      "step": 110000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.456618506021167e-05,
+      "loss": 1.6981,
+      "step": 110100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.456075667365843e-05,
+      "loss": 1.7,
+      "step": 110200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.45553282871052e-05,
+      "loss": 1.6962,
+      "step": 110300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.454989990055196e-05,
+      "loss": 1.6814,
+      "step": 110400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.454447151399872e-05,
+      "loss": 1.6765,
+      "step": 110500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.453904312744549e-05,
+      "loss": 1.6811,
+      "step": 110600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.453361474089225e-05,
+      "loss": 1.6918,
+      "step": 110700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.452818635433902e-05,
+      "loss": 1.7026,
+      "step": 110800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.4522757967785784e-05,
+      "loss": 1.6863,
+      "step": 110900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.4517329581232546e-05,
+      "loss": 1.6976,
+      "step": 111000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.4511901194679314e-05,
+      "loss": 1.715,
+      "step": 111100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.450647280812608e-05,
+      "loss": 1.689,
+      "step": 111200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.4501044421572845e-05,
+      "loss": 1.679,
+      "step": 111300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.4495616035019613e-05,
+      "loss": 1.6749,
+      "step": 111400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.4490187648466375e-05,
+      "loss": 1.6885,
+      "step": 111500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.448475926191314e-05,
+      "loss": 1.6888,
+      "step": 111600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.4479330875359906e-05,
+      "loss": 1.6944,
+      "step": 111700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.447390248880667e-05,
+      "loss": 1.6997,
+      "step": 111800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.446847410225343e-05,
+      "loss": 1.685,
+      "step": 111900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.44630457157002e-05,
+      "loss": 1.6695,
+      "step": 112000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.445761732914696e-05,
+      "loss": 1.6916,
+      "step": 112100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.445218894259373e-05,
+      "loss": 1.7029,
+      "step": 112200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.44467605560405e-05,
+      "loss": 1.7237,
+      "step": 112300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.444133216948726e-05,
+      "loss": 1.7238,
+      "step": 112400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.443590378293402e-05,
+      "loss": 1.6702,
+      "step": 112500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.443047539638079e-05,
+      "loss": 1.6817,
+      "step": 112600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.442504700982755e-05,
+      "loss": 1.7053,
+      "step": 112700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.441961862327432e-05,
+      "loss": 1.6739,
+      "step": 112800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.441419023672108e-05,
+      "loss": 1.682,
+      "step": 112900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.4408761850167844e-05,
+      "loss": 1.6826,
+      "step": 113000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.440333346361461e-05,
+      "loss": 1.7159,
+      "step": 113100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.439790507706138e-05,
+      "loss": 1.6906,
+      "step": 113200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.439247669050814e-05,
+      "loss": 1.6761,
+      "step": 113300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.438704830395491e-05,
+      "loss": 1.7117,
+      "step": 113400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.4381619917401674e-05,
+      "loss": 1.6905,
+      "step": 113500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.4376191530848436e-05,
+      "loss": 1.6703,
+      "step": 113600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.4370763144295204e-05,
+      "loss": 1.6829,
+      "step": 113700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.4365334757741966e-05,
+      "loss": 1.7074,
+      "step": 113800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.435990637118873e-05,
+      "loss": 1.6749,
+      "step": 113900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.43544779846355e-05,
+      "loss": 1.6903,
+      "step": 114000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.434904959808226e-05,
+      "loss": 1.6965,
+      "step": 114100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.434362121152903e-05,
+      "loss": 1.707,
+      "step": 114200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.4338192824975796e-05,
+      "loss": 1.7036,
+      "step": 114300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.433276443842256e-05,
+      "loss": 1.6914,
+      "step": 114400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.432733605186932e-05,
+      "loss": 1.7129,
+      "step": 114500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.432190766531609e-05,
+      "loss": 1.6971,
+      "step": 114600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.431647927876285e-05,
+      "loss": 1.69,
+      "step": 114700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.431105089220962e-05,
+      "loss": 1.6701,
+      "step": 114800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.430562250565638e-05,
+      "loss": 1.7008,
+      "step": 114900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.430019411910314e-05,
+      "loss": 1.7034,
+      "step": 115000
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 1.5167127847671509,
+      "eval_runtime": 19.0213,
+      "eval_samples_per_second": 525.726,
+      "eval_steps_per_second": 16.455,
+      "step": 115000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.429476573254991e-05,
+      "loss": 1.6797,
+      "step": 115100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.428933734599667e-05,
+      "loss": 1.6997,
+      "step": 115200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.428390895944344e-05,
+      "loss": 1.6886,
+      "step": 115300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.427848057289021e-05,
+      "loss": 1.7028,
+      "step": 115400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.427305218633697e-05,
+      "loss": 1.675,
+      "step": 115500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4267623799783734e-05,
+      "loss": 1.6623,
+      "step": 115600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.42621954132305e-05,
+      "loss": 1.6872,
+      "step": 115700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4256767026677265e-05,
+      "loss": 1.6599,
+      "step": 115800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4251338640124027e-05,
+      "loss": 1.6994,
+      "step": 115900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4245910253570795e-05,
+      "loss": 1.6918,
+      "step": 116000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.424048186701756e-05,
+      "loss": 1.6871,
+      "step": 116100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4235053480464326e-05,
+      "loss": 1.7014,
+      "step": 116200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.422962509391109e-05,
+      "loss": 1.6834,
+      "step": 116300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4224196707357856e-05,
+      "loss": 1.6943,
+      "step": 116400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.421876832080462e-05,
+      "loss": 1.6789,
+      "step": 116500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.421333993425139e-05,
+      "loss": 1.718,
+      "step": 116600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.420791154769815e-05,
+      "loss": 1.7048,
+      "step": 116700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.420248316114492e-05,
+      "loss": 1.7218,
+      "step": 116800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.419705477459168e-05,
+      "loss": 1.6819,
+      "step": 116900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.419162638803844e-05,
+      "loss": 1.7022,
+      "step": 117000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.418619800148521e-05,
+      "loss": 1.6735,
+      "step": 117100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.418076961493197e-05,
+      "loss": 1.7054,
+      "step": 117200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4175341228378733e-05,
+      "loss": 1.6838,
+      "step": 117300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.416991284182551e-05,
+      "loss": 1.7075,
+      "step": 117400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.416448445527227e-05,
+      "loss": 1.6719,
+      "step": 117500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.415905606871903e-05,
+      "loss": 1.693,
+      "step": 117600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.41536276821658e-05,
+      "loss": 1.678,
+      "step": 117700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.414819929561256e-05,
+      "loss": 1.723,
+      "step": 117800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4142770909059325e-05,
+      "loss": 1.7097,
+      "step": 117900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4137342522506094e-05,
+      "loss": 1.6379,
+      "step": 118000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4131914135952856e-05,
+      "loss": 1.6527,
+      "step": 118100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4126485749399624e-05,
+      "loss": 1.6692,
+      "step": 118200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4121057362846386e-05,
+      "loss": 1.6807,
+      "step": 118300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.411562897629315e-05,
+      "loss": 1.6781,
+      "step": 118400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4110200589739917e-05,
+      "loss": 1.6679,
+      "step": 118500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4104772203186685e-05,
+      "loss": 1.6798,
+      "step": 118600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.409934381663345e-05,
+      "loss": 1.6614,
+      "step": 118700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4093915430080216e-05,
+      "loss": 1.7208,
+      "step": 118800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.408848704352698e-05,
+      "loss": 1.6992,
+      "step": 118900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.408305865697374e-05,
+      "loss": 1.7189,
+      "step": 119000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.407763027042051e-05,
+      "loss": 1.6979,
+      "step": 119100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.407220188386727e-05,
+      "loss": 1.6903,
+      "step": 119200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.406677349731403e-05,
+      "loss": 1.679,
+      "step": 119300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.40613451107608e-05,
+      "loss": 1.6904,
+      "step": 119400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.405591672420757e-05,
+      "loss": 1.6697,
+      "step": 119500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.405048833765433e-05,
+      "loss": 1.6896,
+      "step": 119600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.40450599511011e-05,
+      "loss": 1.6877,
+      "step": 119700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.403963156454786e-05,
+      "loss": 1.6599,
+      "step": 119800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4034203177994623e-05,
+      "loss": 1.7118,
+      "step": 119900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.402877479144139e-05,
+      "loss": 1.6659,
+      "step": 120000
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 1.5134057998657227,
+      "eval_runtime": 19.0004,
+      "eval_samples_per_second": 526.305,
+      "eval_steps_per_second": 16.473,
+      "step": 120000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4023346404888154e-05,
+      "loss": 1.6814,
+      "step": 120100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.401791801833492e-05,
+      "loss": 1.6708,
+      "step": 120200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4012489631781685e-05,
+      "loss": 1.6479,
+      "step": 120300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4007061245228446e-05,
+      "loss": 1.6741,
+      "step": 120400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4001632858675215e-05,
+      "loss": 1.6662,
+      "step": 120500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.3996204472121984e-05,
+      "loss": 1.6919,
+      "step": 120600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.3990776085568746e-05,
+      "loss": 1.6835,
+      "step": 120700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.3985347699015514e-05,
+      "loss": 1.6708,
+      "step": 120800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.3979919312462276e-05,
+      "loss": 1.6714,
+      "step": 120900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.397449092590904e-05,
+      "loss": 1.674,
+      "step": 121000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.396906253935581e-05,
+      "loss": 1.6925,
+      "step": 121100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.396363415280257e-05,
+      "loss": 1.654,
+      "step": 121200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.395820576624933e-05,
+      "loss": 1.6583,
+      "step": 121300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.39527773796961e-05,
+      "loss": 1.6943,
+      "step": 121400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.394734899314286e-05,
+      "loss": 1.6652,
+      "step": 121500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.394192060658963e-05,
+      "loss": 1.6484,
+      "step": 121600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.39364922200364e-05,
+      "loss": 1.6899,
+      "step": 121700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.393106383348316e-05,
+      "loss": 1.6845,
+      "step": 121800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.392563544692992e-05,
+      "loss": 1.6607,
+      "step": 121900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.392020706037669e-05,
+      "loss": 1.6697,
+      "step": 122000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.391477867382345e-05,
+      "loss": 1.6999,
+      "step": 122100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.390935028727022e-05,
+      "loss": 1.6962,
+      "step": 122200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.390392190071698e-05,
+      "loss": 1.6608,
+      "step": 122300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.3898493514163745e-05,
+      "loss": 1.7008,
+      "step": 122400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.3893065127610514e-05,
+      "loss": 1.6752,
+      "step": 122500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.388763674105728e-05,
+      "loss": 1.6751,
+      "step": 122600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.3882208354504044e-05,
+      "loss": 1.6752,
+      "step": 122700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.387677996795081e-05,
+      "loss": 1.6605,
+      "step": 122800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.3871351581397575e-05,
+      "loss": 1.6843,
+      "step": 122900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.3865923194844336e-05,
+      "loss": 1.6849,
+      "step": 123000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.3860494808291105e-05,
+      "loss": 1.6412,
+      "step": 123100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.385506642173787e-05,
+      "loss": 1.6645,
+      "step": 123200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.384963803518463e-05,
+      "loss": 1.6803,
+      "step": 123300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.38442096486314e-05,
+      "loss": 1.6622,
+      "step": 123400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.383878126207816e-05,
+      "loss": 1.6673,
+      "step": 123500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.383335287552492e-05,
+      "loss": 1.6334,
+      "step": 123600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.38279244889717e-05,
+      "loss": 1.6778,
+      "step": 123700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.382249610241846e-05,
+      "loss": 1.6805,
+      "step": 123800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.381706771586522e-05,
+      "loss": 1.711,
+      "step": 123900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.381163932931199e-05,
+      "loss": 1.6893,
+      "step": 124000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.380621094275875e-05,
+      "loss": 1.6537,
+      "step": 124100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.380078255620552e-05,
+      "loss": 1.6938,
+      "step": 124200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.379535416965228e-05,
+      "loss": 1.6666,
+      "step": 124300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.3789925783099043e-05,
+      "loss": 1.709,
+      "step": 124400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.378449739654581e-05,
+      "loss": 1.6937,
+      "step": 124500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3779069009992574e-05,
+      "loss": 1.7111,
+      "step": 124600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.377364062343934e-05,
+      "loss": 1.6885,
+      "step": 124700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.376821223688611e-05,
+      "loss": 1.6691,
+      "step": 124800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.376278385033287e-05,
+      "loss": 1.6659,
+      "step": 124900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3757355463779635e-05,
+      "loss": 1.6814,
+      "step": 125000
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 1.5067857503890991,
+      "eval_runtime": 18.9814,
+      "eval_samples_per_second": 526.832,
+      "eval_steps_per_second": 16.49,
+      "step": 125000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3751927077226404e-05,
+      "loss": 1.7118,
+      "step": 125100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3746498690673166e-05,
+      "loss": 1.6906,
+      "step": 125200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.374107030411993e-05,
+      "loss": 1.6594,
+      "step": 125300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3735641917566696e-05,
+      "loss": 1.6768,
+      "step": 125400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.373021353101346e-05,
+      "loss": 1.6496,
+      "step": 125500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.372478514446022e-05,
+      "loss": 1.6796,
+      "step": 125600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.371935675790699e-05,
+      "loss": 1.6844,
+      "step": 125700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.371392837135376e-05,
+      "loss": 1.6712,
+      "step": 125800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.370849998480052e-05,
+      "loss": 1.6566,
+      "step": 125900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.370307159824729e-05,
+      "loss": 1.653,
+      "step": 126000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.369764321169405e-05,
+      "loss": 1.711,
+      "step": 126100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.369221482514082e-05,
+      "loss": 1.6782,
+      "step": 126200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.368678643858758e-05,
+      "loss": 1.6618,
+      "step": 126300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.368135805203434e-05,
+      "loss": 1.6439,
+      "step": 126400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.367592966548111e-05,
+      "loss": 1.6435,
+      "step": 126500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.367050127892787e-05,
+      "loss": 1.6772,
+      "step": 126600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3665072892374634e-05,
+      "loss": 1.6647,
+      "step": 126700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.365964450582141e-05,
+      "loss": 1.6984,
+      "step": 126800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.365421611926817e-05,
+      "loss": 1.652,
+      "step": 126900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3648787732714933e-05,
+      "loss": 1.6726,
+      "step": 127000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.36433593461617e-05,
+      "loss": 1.6786,
+      "step": 127100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3637930959608464e-05,
+      "loss": 1.6714,
+      "step": 127200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3632502573055226e-05,
+      "loss": 1.6693,
+      "step": 127300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3627074186501995e-05,
+      "loss": 1.6929,
+      "step": 127400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3621645799948756e-05,
+      "loss": 1.6755,
+      "step": 127500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.361621741339552e-05,
+      "loss": 1.6561,
+      "step": 127600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.361078902684229e-05,
+      "loss": 1.6771,
+      "step": 127700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.360536064028905e-05,
+      "loss": 1.682,
+      "step": 127800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.359993225373582e-05,
+      "loss": 1.6525,
+      "step": 127900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3594503867182586e-05,
+      "loss": 1.6523,
+      "step": 128000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.358907548062935e-05,
+      "loss": 1.6703,
+      "step": 128100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.358364709407612e-05,
+      "loss": 1.659,
+      "step": 128200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.357821870752288e-05,
+      "loss": 1.6595,
+      "step": 128300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.357279032096964e-05,
+      "loss": 1.655,
+      "step": 128400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.356736193441641e-05,
+      "loss": 1.674,
+      "step": 128500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.356193354786317e-05,
+      "loss": 1.6901,
+      "step": 128600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.355650516130993e-05,
+      "loss": 1.681,
+      "step": 128700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.35510767747567e-05,
+      "loss": 1.6741,
+      "step": 128800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.354564838820347e-05,
+      "loss": 1.6859,
+      "step": 128900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.354022000165023e-05,
+      "loss": 1.6399,
+      "step": 129000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3534791615097e-05,
+      "loss": 1.6625,
+      "step": 129100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.352936322854376e-05,
+      "loss": 1.6884,
+      "step": 129200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3523934841990524e-05,
+      "loss": 1.6917,
+      "step": 129300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.351850645543729e-05,
+      "loss": 1.6867,
+      "step": 129400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3513078068884055e-05,
+      "loss": 1.6759,
+      "step": 129500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.350764968233082e-05,
+      "loss": 1.6718,
+      "step": 129600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3502221295777585e-05,
+      "loss": 1.6435,
+      "step": 129700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.349679290922435e-05,
+      "loss": 1.6757,
+      "step": 129800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3491364522671116e-05,
+      "loss": 1.6973,
+      "step": 129900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3485936136117885e-05,
+      "loss": 1.6855,
+      "step": 130000
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 1.5044703483581543,
+      "eval_runtime": 18.9872,
+      "eval_samples_per_second": 526.672,
+      "eval_steps_per_second": 16.485,
+      "step": 130000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3480507749564646e-05,
+      "loss": 1.6843,
+      "step": 130100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.347507936301141e-05,
+      "loss": 1.6709,
+      "step": 130200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.346965097645818e-05,
+      "loss": 1.6656,
+      "step": 130300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.346422258990494e-05,
+      "loss": 1.6635,
+      "step": 130400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.345879420335171e-05,
+      "loss": 1.6454,
+      "step": 130500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.345336581679847e-05,
+      "loss": 1.678,
+      "step": 130600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.344793743024523e-05,
+      "loss": 1.6965,
+      "step": 130700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3442509043692e-05,
+      "loss": 1.6447,
+      "step": 130800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.343708065713876e-05,
+      "loss": 1.6605,
+      "step": 130900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.343165227058553e-05,
+      "loss": 1.6782,
+      "step": 131000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.34262238840323e-05,
+      "loss": 1.6535,
+      "step": 131100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.342079549747906e-05,
+      "loss": 1.6603,
+      "step": 131200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.341536711092582e-05,
+      "loss": 1.6508,
+      "step": 131300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.340993872437259e-05,
+      "loss": 1.6625,
+      "step": 131400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.340451033781935e-05,
+      "loss": 1.6584,
+      "step": 131500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3399081951266115e-05,
+      "loss": 1.6942,
+      "step": 131600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3393653564712884e-05,
+      "loss": 1.6679,
+      "step": 131700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3388225178159646e-05,
+      "loss": 1.6651,
+      "step": 131800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3382796791606414e-05,
+      "loss": 1.6969,
+      "step": 131900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3377368405053176e-05,
+      "loss": 1.7259,
+      "step": 132000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3371940018499945e-05,
+      "loss": 1.6257,
+      "step": 132100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.336651163194671e-05,
+      "loss": 1.6517,
+      "step": 132200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3361083245393475e-05,
+      "loss": 1.7008,
+      "step": 132300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.335565485884024e-05,
+      "loss": 1.6519,
+      "step": 132400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3350226472287006e-05,
+      "loss": 1.6914,
+      "step": 132500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.334479808573377e-05,
+      "loss": 1.6574,
+      "step": 132600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.333936969918053e-05,
+      "loss": 1.6667,
+      "step": 132700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.33339413126273e-05,
+      "loss": 1.6928,
+      "step": 132800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.332851292607406e-05,
+      "loss": 1.677,
+      "step": 132900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.332308453952082e-05,
+      "loss": 1.6511,
+      "step": 133000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.33176561529676e-05,
+      "loss": 1.6605,
+      "step": 133100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.331222776641436e-05,
+      "loss": 1.6317,
+      "step": 133200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.330679937986112e-05,
+      "loss": 1.6925,
+      "step": 133300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.330137099330789e-05,
+      "loss": 1.6553,
+      "step": 133400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.329594260675465e-05,
+      "loss": 1.6675,
+      "step": 133500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3290514220201414e-05,
+      "loss": 1.6711,
+      "step": 133600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.328508583364818e-05,
+      "loss": 1.64,
+      "step": 133700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3279657447094944e-05,
+      "loss": 1.6866,
+      "step": 133800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.327422906054171e-05,
+      "loss": 1.6831,
+      "step": 133900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.3268800673988475e-05,
+      "loss": 1.6413,
+      "step": 134000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.326337228743524e-05,
+      "loss": 1.6605,
+      "step": 134100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.3257943900882005e-05,
+      "loss": 1.7032,
+      "step": 134200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.3252515514328774e-05,
+      "loss": 1.6694,
+      "step": 134300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.3247087127775536e-05,
+      "loss": 1.6462,
+      "step": 134400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.3241658741222304e-05,
+      "loss": 1.6618,
+      "step": 134500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.3236230354669066e-05,
+      "loss": 1.6611,
+      "step": 134600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.323080196811583e-05,
+      "loss": 1.6743,
+      "step": 134700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.32253735815626e-05,
+      "loss": 1.7014,
+      "step": 134800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.321994519500936e-05,
+      "loss": 1.6307,
+      "step": 134900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.321451680845612e-05,
+      "loss": 1.7163,
+      "step": 135000
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 1.502817153930664,
+      "eval_runtime": 19.0069,
+      "eval_samples_per_second": 526.124,
+      "eval_steps_per_second": 16.468,
+      "step": 135000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.320908842190289e-05,
+      "loss": 1.6829,
+      "step": 135100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.320366003534966e-05,
+      "loss": 1.6813,
+      "step": 135200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.319823164879642e-05,
+      "loss": 1.7002,
+      "step": 135300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.319280326224319e-05,
+      "loss": 1.69,
+      "step": 135400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.318737487568995e-05,
+      "loss": 1.6889,
+      "step": 135500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.318194648913671e-05,
+      "loss": 1.6668,
+      "step": 135600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.317651810258348e-05,
+      "loss": 1.6814,
+      "step": 135700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.317108971603024e-05,
+      "loss": 1.6247,
+      "step": 135800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.316566132947701e-05,
+      "loss": 1.6818,
+      "step": 135900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.316023294292377e-05,
+      "loss": 1.696,
+      "step": 136000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.3154804556370535e-05,
+      "loss": 1.6665,
+      "step": 136100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3149376169817304e-05,
+      "loss": 1.6671,
+      "step": 136200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.314394778326407e-05,
+      "loss": 1.6745,
+      "step": 136300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3138519396710834e-05,
+      "loss": 1.6701,
+      "step": 136400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.31330910101576e-05,
+      "loss": 1.6677,
+      "step": 136500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3127662623604365e-05,
+      "loss": 1.6759,
+      "step": 136600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.312223423705113e-05,
+      "loss": 1.6808,
+      "step": 136700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3116805850497895e-05,
+      "loss": 1.6809,
+      "step": 136800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.311137746394466e-05,
+      "loss": 1.6222,
+      "step": 136900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.310594907739142e-05,
+      "loss": 1.698,
+      "step": 137000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.310052069083819e-05,
+      "loss": 1.6752,
+      "step": 137100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.309509230428495e-05,
+      "loss": 1.7055,
+      "step": 137200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.308966391773172e-05,
+      "loss": 1.6652,
+      "step": 137300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.308423553117849e-05,
+      "loss": 1.6686,
+      "step": 137400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.307880714462525e-05,
+      "loss": 1.6582,
+      "step": 137500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.307337875807201e-05,
+      "loss": 1.6555,
+      "step": 137600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.306795037151878e-05,
+      "loss": 1.6501,
+      "step": 137700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.306252198496554e-05,
+      "loss": 1.6653,
+      "step": 137800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.305709359841231e-05,
+      "loss": 1.6581,
+      "step": 137900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.305166521185907e-05,
+      "loss": 1.6755,
+      "step": 138000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3046236825305834e-05,
+      "loss": 1.6837,
+      "step": 138100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.30408084387526e-05,
+      "loss": 1.6873,
+      "step": 138200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3035380052199364e-05,
+      "loss": 1.6475,
+      "step": 138300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.302995166564613e-05,
+      "loss": 1.6719,
+      "step": 138400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.30245232790929e-05,
+      "loss": 1.664,
+      "step": 138500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.301909489253966e-05,
+      "loss": 1.6426,
+      "step": 138600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.3013666505986425e-05,
+      "loss": 1.6444,
+      "step": 138700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.3008238119433194e-05,
+      "loss": 1.6721,
+      "step": 138800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.3002809732879956e-05,
+      "loss": 1.6697,
+      "step": 138900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.299738134632672e-05,
+      "loss": 1.6546,
+      "step": 139000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.2991952959773486e-05,
+      "loss": 1.6909,
+      "step": 139100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.298652457322025e-05,
+      "loss": 1.6854,
+      "step": 139200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.298109618666702e-05,
+      "loss": 1.6527,
+      "step": 139300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.2975667800113785e-05,
+      "loss": 1.6853,
+      "step": 139400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.297023941356055e-05,
+      "loss": 1.6556,
+      "step": 139500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.296481102700731e-05,
+      "loss": 1.6656,
+      "step": 139600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.295938264045408e-05,
+      "loss": 1.6581,
+      "step": 139700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.295395425390084e-05,
+      "loss": 1.6422,
+      "step": 139800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.294852586734761e-05,
+      "loss": 1.6744,
+      "step": 139900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.294309748079437e-05,
+      "loss": 1.6645,
+      "step": 140000
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 1.4998124837875366,
+      "eval_runtime": 18.9623,
+      "eval_samples_per_second": 527.362,
+      "eval_steps_per_second": 16.506,
+      "step": 140000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.293766909424113e-05,
+      "loss": 1.7061,
+      "step": 140100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.29322407076879e-05,
+      "loss": 1.6883,
+      "step": 140200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.292681232113466e-05,
+      "loss": 1.6674,
+      "step": 140300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.2921383934581425e-05,
+      "loss": 1.6763,
+      "step": 140400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.29159555480282e-05,
+      "loss": 1.6551,
+      "step": 140500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.291052716147496e-05,
+      "loss": 1.6787,
+      "step": 140600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.2905098774921724e-05,
+      "loss": 1.6661,
+      "step": 140700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.289967038836849e-05,
+      "loss": 1.6769,
+      "step": 140800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.2894242001815254e-05,
+      "loss": 1.6581,
+      "step": 140900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.2888813615262016e-05,
+      "loss": 1.6592,
+      "step": 141000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.2883385228708785e-05,
+      "loss": 1.6744,
+      "step": 141100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.287795684215555e-05,
+      "loss": 1.6752,
+      "step": 141200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.2872528455602315e-05,
+      "loss": 1.6546,
+      "step": 141300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.286710006904908e-05,
+      "loss": 1.6599,
+      "step": 141400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.2861671682495846e-05,
+      "loss": 1.6713,
+      "step": 141500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.285624329594261e-05,
+      "loss": 1.6276,
+      "step": 141600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.2850814909389376e-05,
+      "loss": 1.6647,
+      "step": 141700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.284538652283614e-05,
+      "loss": 1.6548,
+      "step": 141800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.283995813628291e-05,
+      "loss": 1.6565,
+      "step": 141900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.283452974972967e-05,
+      "loss": 1.6679,
+      "step": 142000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.282910136317643e-05,
+      "loss": 1.6429,
+      "step": 142100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.28236729766232e-05,
+      "loss": 1.6365,
+      "step": 142200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.281824459006996e-05,
+      "loss": 1.6318,
+      "step": 142300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.281281620351672e-05,
+      "loss": 1.6821,
+      "step": 142400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.28073878169635e-05,
+      "loss": 1.658,
+      "step": 142500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.280195943041026e-05,
+      "loss": 1.6852,
+      "step": 142600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.279653104385702e-05,
+      "loss": 1.6617,
+      "step": 142700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.279110265730379e-05,
+      "loss": 1.699,
+      "step": 142800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.278567427075055e-05,
+      "loss": 1.6443,
+      "step": 142900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.2780245884197315e-05,
+      "loss": 1.6357,
+      "step": 143000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.277481749764408e-05,
+      "loss": 1.6719,
+      "step": 143100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.2769389111090845e-05,
+      "loss": 1.6529,
+      "step": 143200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.2763960724537614e-05,
+      "loss": 1.6443,
+      "step": 143300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.2758532337984376e-05,
+      "loss": 1.6534,
+      "step": 143400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.275310395143114e-05,
+      "loss": 1.6673,
+      "step": 143500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.2747675564877906e-05,
+      "loss": 1.6496,
+      "step": 143600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.2742247178324675e-05,
+      "loss": 1.6808,
+      "step": 143700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.273681879177144e-05,
+      "loss": 1.6599,
+      "step": 143800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.2731390405218205e-05,
+      "loss": 1.6626,
+      "step": 143900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.272596201866497e-05,
+      "loss": 1.6623,
+      "step": 144000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.272053363211173e-05,
+      "loss": 1.6576,
+      "step": 144100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.27151052455585e-05,
+      "loss": 1.6479,
+      "step": 144200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.270967685900526e-05,
+      "loss": 1.6492,
+      "step": 144300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.270424847245202e-05,
+      "loss": 1.6916,
+      "step": 144400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.269882008589879e-05,
+      "loss": 1.6518,
+      "step": 144500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.269339169934556e-05,
+      "loss": 1.6815,
+      "step": 144600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.268796331279232e-05,
+      "loss": 1.6319,
+      "step": 144700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.268253492623909e-05,
+      "loss": 1.668,
+      "step": 144800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.267710653968585e-05,
+      "loss": 1.6594,
+      "step": 144900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.267167815313261e-05,
+      "loss": 1.6918,
+      "step": 145000
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 1.4964494705200195,
+      "eval_runtime": 19.0245,
+      "eval_samples_per_second": 525.638,
+      "eval_steps_per_second": 16.452,
+      "step": 145000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.266624976657938e-05,
+      "loss": 1.6603,
+      "step": 145100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.2660821380026144e-05,
+      "loss": 1.6859,
+      "step": 145200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.265539299347291e-05,
+      "loss": 1.6378,
+      "step": 145300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.2649964606919674e-05,
+      "loss": 1.6768,
+      "step": 145400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.2644536220366436e-05,
+      "loss": 1.6845,
+      "step": 145500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.2639107833813205e-05,
+      "loss": 1.6757,
+      "step": 145600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.263367944725997e-05,
+      "loss": 1.6575,
+      "step": 145700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.2628251060706735e-05,
+      "loss": 1.6798,
+      "step": 145800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.2622822674153504e-05,
+      "loss": 1.6504,
+      "step": 145900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.2617394287600266e-05,
+      "loss": 1.6527,
+      "step": 146000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.261196590104703e-05,
+      "loss": 1.6972,
+      "step": 146100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.2606537514493796e-05,
+      "loss": 1.6573,
+      "step": 146200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.260110912794056e-05,
+      "loss": 1.6811,
+      "step": 146300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.259568074138732e-05,
+      "loss": 1.6494,
+      "step": 146400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.259025235483409e-05,
+      "loss": 1.6271,
+      "step": 146500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.258482396828085e-05,
+      "loss": 1.6904,
+      "step": 146600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.257939558172762e-05,
+      "loss": 1.6779,
+      "step": 146700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.257396719517439e-05,
+      "loss": 1.6485,
+      "step": 146800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.256853880862115e-05,
+      "loss": 1.6511,
+      "step": 146900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.256311042206791e-05,
+      "loss": 1.6497,
+      "step": 147000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.255768203551468e-05,
+      "loss": 1.6858,
+      "step": 147100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.255225364896144e-05,
+      "loss": 1.6702,
+      "step": 147200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.254682526240821e-05,
+      "loss": 1.6352,
+      "step": 147300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.254139687585497e-05,
+      "loss": 1.6741,
+      "step": 147400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.2535968489301735e-05,
+      "loss": 1.6842,
+      "step": 147500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.25305401027485e-05,
+      "loss": 1.6406,
+      "step": 147600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.2525111716195265e-05,
+      "loss": 1.6622,
+      "step": 147700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.2519683329642034e-05,
+      "loss": 1.636,
+      "step": 147800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.25142549430888e-05,
+      "loss": 1.6487,
+      "step": 147900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.2508826556535564e-05,
+      "loss": 1.6737,
+      "step": 148000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.2503398169982326e-05,
+      "loss": 1.6372,
+      "step": 148100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.2497969783429095e-05,
+      "loss": 1.6472,
+      "step": 148200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.2492541396875857e-05,
+      "loss": 1.6508,
+      "step": 148300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.248711301032262e-05,
+      "loss": 1.6693,
+      "step": 148400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.248168462376939e-05,
+      "loss": 1.6808,
+      "step": 148500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.247625623721615e-05,
+      "loss": 1.6758,
+      "step": 148600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.247082785066291e-05,
+      "loss": 1.6726,
+      "step": 148700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.2465399464109686e-05,
+      "loss": 1.6293,
+      "step": 148800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.245997107755645e-05,
+      "loss": 1.6371,
+      "step": 148900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.245454269100321e-05,
+      "loss": 1.6726,
+      "step": 149000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.244911430444998e-05,
+      "loss": 1.6672,
+      "step": 149100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.244368591789674e-05,
+      "loss": 1.6564,
+      "step": 149200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.243825753134351e-05,
+      "loss": 1.6373,
+      "step": 149300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.243282914479027e-05,
+      "loss": 1.6795,
+      "step": 149400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.242740075823703e-05,
+      "loss": 1.6624,
+      "step": 149500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.24219723716838e-05,
+      "loss": 1.6491,
+      "step": 149600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.2416543985130564e-05,
+      "loss": 1.6289,
+      "step": 149700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.2411115598577325e-05,
+      "loss": 1.6284,
+      "step": 149800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.24056872120241e-05,
+      "loss": 1.6781,
+      "step": 149900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.240025882547086e-05,
+      "loss": 1.6385,
+      "step": 150000
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 1.498610019683838,
+      "eval_runtime": 19.0453,
+      "eval_samples_per_second": 525.065,
+      "eval_steps_per_second": 16.435,
+      "step": 150000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.2394830438917625e-05,
+      "loss": 1.663,
+      "step": 150100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.238940205236439e-05,
+      "loss": 1.6227,
+      "step": 150200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.2383973665811155e-05,
+      "loss": 1.6561,
+      "step": 150300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.237854527925792e-05,
+      "loss": 1.6525,
+      "step": 150400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.2373116892704686e-05,
+      "loss": 1.6423,
+      "step": 150500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.236768850615145e-05,
+      "loss": 1.6528,
+      "step": 150600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.236226011959821e-05,
+      "loss": 1.6514,
+      "step": 150700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.235683173304498e-05,
+      "loss": 1.6747,
+      "step": 150800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.235140334649175e-05,
+      "loss": 1.6646,
+      "step": 150900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.234597495993851e-05,
+      "loss": 1.6572,
+      "step": 151000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.234054657338528e-05,
+      "loss": 1.6396,
+      "step": 151100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.233511818683204e-05,
+      "loss": 1.647,
+      "step": 151200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.232968980027881e-05,
+      "loss": 1.6633,
+      "step": 151300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.232426141372557e-05,
+      "loss": 1.6479,
+      "step": 151400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.231883302717233e-05,
+      "loss": 1.6554,
+      "step": 151500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.23134046406191e-05,
+      "loss": 1.6536,
+      "step": 151600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.230797625406586e-05,
+      "loss": 1.6546,
+      "step": 151700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.2302547867512624e-05,
+      "loss": 1.6994,
+      "step": 151800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.229711948095939e-05,
+      "loss": 1.6134,
+      "step": 151900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.229169109440616e-05,
+      "loss": 1.6861,
+      "step": 152000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.228626270785292e-05,
+      "loss": 1.6588,
+      "step": 152100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.228083432129969e-05,
+      "loss": 1.6333,
+      "step": 152200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.2275405934746454e-05,
+      "loss": 1.6712,
+      "step": 152300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.2269977548193215e-05,
+      "loss": 1.6621,
+      "step": 152400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.2264549161639984e-05,
+      "loss": 1.6447,
+      "step": 152500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.2259120775086746e-05,
+      "loss": 1.6746,
+      "step": 152600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.225369238853351e-05,
+      "loss": 1.6465,
+      "step": 152700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.2248264001980277e-05,
+      "loss": 1.6743,
+      "step": 152800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.224283561542704e-05,
+      "loss": 1.6732,
+      "step": 152900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.223740722887381e-05,
+      "loss": 1.6312,
+      "step": 153000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.2231978842320576e-05,
+      "loss": 1.6696,
+      "step": 153100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.222655045576734e-05,
+      "loss": 1.6439,
+      "step": 153200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.2221122069214106e-05,
+      "loss": 1.6474,
+      "step": 153300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.221569368266087e-05,
+      "loss": 1.6472,
+      "step": 153400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.221026529610763e-05,
+      "loss": 1.6635,
+      "step": 153500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.22048369095544e-05,
+      "loss": 1.6512,
+      "step": 153600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.219940852300116e-05,
+      "loss": 1.6748,
+      "step": 153700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.219398013644792e-05,
+      "loss": 1.666,
+      "step": 153800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.218855174989469e-05,
+      "loss": 1.6338,
+      "step": 153900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.218312336334145e-05,
+      "loss": 1.6643,
+      "step": 154000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.217769497678822e-05,
+      "loss": 1.6548,
+      "step": 154100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.217226659023499e-05,
+      "loss": 1.6619,
+      "step": 154200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.216683820368175e-05,
+      "loss": 1.6543,
+      "step": 154300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.2161409817128514e-05,
+      "loss": 1.6247,
+      "step": 154400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.215598143057528e-05,
+      "loss": 1.6504,
+      "step": 154500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.2150553044022044e-05,
+      "loss": 1.6754,
+      "step": 154600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.2145124657468806e-05,
+      "loss": 1.649,
+      "step": 154700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.2139696270915575e-05,
+      "loss": 1.6665,
+      "step": 154800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.213426788436234e-05,
+      "loss": 1.6553,
+      "step": 154900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.2128839497809106e-05,
+      "loss": 1.632,
+      "step": 155000
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 1.4961014986038208,
+      "eval_runtime": 18.9397,
+      "eval_samples_per_second": 527.991,
+      "eval_steps_per_second": 16.526,
+      "step": 155000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.2123411111255874e-05,
+      "loss": 1.6508,
+      "step": 155100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.2117982724702636e-05,
+      "loss": 1.6373,
+      "step": 155200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.2112554338149405e-05,
+      "loss": 1.6201,
+      "step": 155300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.2107125951596167e-05,
+      "loss": 1.6426,
+      "step": 155400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.210169756504293e-05,
+      "loss": 1.678,
+      "step": 155500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.20962691784897e-05,
+      "loss": 1.65,
+      "step": 155600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.209084079193646e-05,
+      "loss": 1.6561,
+      "step": 155700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.208541240538322e-05,
+      "loss": 1.6604,
+      "step": 155800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.207998401882999e-05,
+      "loss": 1.6548,
+      "step": 155900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.207455563227675e-05,
+      "loss": 1.6122,
+      "step": 156000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.206912724572351e-05,
+      "loss": 1.6578,
+      "step": 156100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.206369885917029e-05,
+      "loss": 1.6357,
+      "step": 156200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.205827047261705e-05,
+      "loss": 1.6642,
+      "step": 156300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.205284208606381e-05,
+      "loss": 1.6514,
+      "step": 156400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.204741369951058e-05,
+      "loss": 1.6546,
+      "step": 156500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.204198531295734e-05,
+      "loss": 1.6795,
+      "step": 156600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.2036556926404105e-05,
+      "loss": 1.6423,
+      "step": 156700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.2031128539850873e-05,
+      "loss": 1.6395,
+      "step": 156800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.2025700153297635e-05,
+      "loss": 1.6905,
+      "step": 156900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.2020271766744404e-05,
+      "loss": 1.6481,
+      "step": 157000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.2014843380191166e-05,
+      "loss": 1.6364,
+      "step": 157100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.2009414993637935e-05,
+      "loss": 1.6389,
+      "step": 157200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.20039866070847e-05,
+      "loss": 1.6198,
+      "step": 157300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.1998558220531465e-05,
+      "loss": 1.6397,
+      "step": 157400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.199312983397823e-05,
+      "loss": 1.6326,
+      "step": 157500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.1987701447424996e-05,
+      "loss": 1.6583,
+      "step": 157600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.198227306087176e-05,
+      "loss": 1.6271,
+      "step": 157700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.197684467431852e-05,
+      "loss": 1.6297,
+      "step": 157800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.197141628776529e-05,
+      "loss": 1.653,
+      "step": 157900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.196598790121205e-05,
+      "loss": 1.6225,
+      "step": 158000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.196055951465881e-05,
+      "loss": 1.6523,
+      "step": 158100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.195513112810558e-05,
+      "loss": 1.6518,
+      "step": 158200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.194970274155235e-05,
+      "loss": 1.6458,
+      "step": 158300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.194427435499911e-05,
+      "loss": 1.6243,
+      "step": 158400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.193884596844588e-05,
+      "loss": 1.644,
+      "step": 158500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.193341758189264e-05,
+      "loss": 1.6883,
+      "step": 158600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.19279891953394e-05,
+      "loss": 1.6521,
+      "step": 158700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.192256080878617e-05,
+      "loss": 1.7046,
+      "step": 158800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.1917132422232934e-05,
+      "loss": 1.663,
+      "step": 158900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.19117040356797e-05,
+      "loss": 1.6656,
+      "step": 159000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.1906275649126464e-05,
+      "loss": 1.6674,
+      "step": 159100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.1900847262573226e-05,
+      "loss": 1.6277,
+      "step": 159200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.189541887602e-05,
+      "loss": 1.6436,
+      "step": 159300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.1889990489466764e-05,
+      "loss": 1.6566,
+      "step": 159400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.1884562102913525e-05,
+      "loss": 1.6512,
+      "step": 159500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.1879133716360294e-05,
+      "loss": 1.6525,
+      "step": 159600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.1873705329807056e-05,
+      "loss": 1.643,
+      "step": 159700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.186827694325382e-05,
+      "loss": 1.6227,
+      "step": 159800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.1862848556700586e-05,
+      "loss": 1.6451,
+      "step": 159900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.185742017014735e-05,
+      "loss": 1.6474,
+      "step": 160000
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 1.4892033338546753,
+      "eval_runtime": 18.9542,
+      "eval_samples_per_second": 527.588,
+      "eval_steps_per_second": 16.513,
+      "step": 160000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.185199178359411e-05,
+      "loss": 1.653,
+      "step": 160100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.184656339704088e-05,
+      "loss": 1.6536,
+      "step": 160200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.184113501048764e-05,
+      "loss": 1.6498,
+      "step": 160300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.183570662393441e-05,
+      "loss": 1.6401,
+      "step": 160400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.183027823738118e-05,
+      "loss": 1.6488,
+      "step": 160500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.182484985082794e-05,
+      "loss": 1.6423,
+      "step": 160600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.18194214642747e-05,
+      "loss": 1.636,
+      "step": 160700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.181399307772147e-05,
+      "loss": 1.6501,
+      "step": 160800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.180856469116823e-05,
+      "loss": 1.6302,
+      "step": 160900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.1803136304615e-05,
+      "loss": 1.6565,
+      "step": 161000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.179770791806176e-05,
+      "loss": 1.636,
+      "step": 161100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.1792279531508525e-05,
+      "loss": 1.666,
+      "step": 161200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.1786851144955293e-05,
+      "loss": 1.6687,
+      "step": 161300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.178142275840206e-05,
+      "loss": 1.6333,
+      "step": 161400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.1775994371848824e-05,
+      "loss": 1.645,
+      "step": 161500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.177056598529559e-05,
+      "loss": 1.6549,
+      "step": 161600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.1765137598742354e-05,
+      "loss": 1.656,
+      "step": 161700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1759709212189116e-05,
+      "loss": 1.6529,
+      "step": 161800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1754280825635885e-05,
+      "loss": 1.6249,
+      "step": 161900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.174885243908265e-05,
+      "loss": 1.644,
+      "step": 162000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.174342405252941e-05,
+      "loss": 1.6583,
+      "step": 162100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.173799566597618e-05,
+      "loss": 1.6601,
+      "step": 162200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.173256727942294e-05,
+      "loss": 1.6653,
+      "step": 162300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.172713889286971e-05,
+      "loss": 1.6278,
+      "step": 162400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1721710506316477e-05,
+      "loss": 1.7011,
+      "step": 162500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.171628211976324e-05,
+      "loss": 1.6512,
+      "step": 162600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.171085373321e-05,
+      "loss": 1.6867,
+      "step": 162700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.170542534665677e-05,
+      "loss": 1.6322,
+      "step": 162800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.169999696010353e-05,
+      "loss": 1.6367,
+      "step": 162900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.16945685735503e-05,
+      "loss": 1.6477,
+      "step": 163000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.168914018699706e-05,
+      "loss": 1.6592,
+      "step": 163100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.168371180044382e-05,
+      "loss": 1.6355,
+      "step": 163200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.167828341389059e-05,
+      "loss": 1.6156,
+      "step": 163300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1672855027337354e-05,
+      "loss": 1.6602,
+      "step": 163400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.166742664078412e-05,
+      "loss": 1.6826,
+      "step": 163500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.166199825423089e-05,
+      "loss": 1.6549,
+      "step": 163600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.165656986767765e-05,
+      "loss": 1.6715,
+      "step": 163700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1651141481124415e-05,
+      "loss": 1.7004,
+      "step": 163800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1645713094571183e-05,
+      "loss": 1.6407,
+      "step": 163900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1640284708017945e-05,
+      "loss": 1.6596,
+      "step": 164000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.163485632146471e-05,
+      "loss": 1.6159,
+      "step": 164100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1629427934911476e-05,
+      "loss": 1.6305,
+      "step": 164200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.162399954835824e-05,
+      "loss": 1.6307,
+      "step": 164300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1618571161805006e-05,
+      "loss": 1.6507,
+      "step": 164400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1613142775251775e-05,
+      "loss": 1.668,
+      "step": 164500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.160771438869854e-05,
+      "loss": 1.6264,
+      "step": 164600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.16022860021453e-05,
+      "loss": 1.6534,
+      "step": 164700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.159685761559207e-05,
+      "loss": 1.655,
+      "step": 164800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.159142922903883e-05,
+      "loss": 1.6359,
+      "step": 164900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.15860008424856e-05,
+      "loss": 1.656,
+      "step": 165000
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 1.4892749786376953,
+      "eval_runtime": 19.0388,
+      "eval_samples_per_second": 525.242,
+      "eval_steps_per_second": 16.44,
+      "step": 165000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.158057245593236e-05,
+      "loss": 1.6642,
+      "step": 165100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.157514406937912e-05,
+      "loss": 1.6051,
+      "step": 165200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156971568282589e-05,
+      "loss": 1.6373,
+      "step": 165300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156428729627265e-05,
+      "loss": 1.6431,
+      "step": 165400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1558858909719414e-05,
+      "loss": 1.6662,
+      "step": 165500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.155343052316619e-05,
+      "loss": 1.6377,
+      "step": 165600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.154800213661295e-05,
+      "loss": 1.6392,
+      "step": 165700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.154257375005971e-05,
+      "loss": 1.6564,
+      "step": 165800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.153714536350648e-05,
+      "loss": 1.6458,
+      "step": 165900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1531716976953244e-05,
+      "loss": 1.6254,
+      "step": 166000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1526288590400006e-05,
+      "loss": 1.6604,
+      "step": 166100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1520860203846774e-05,
+      "loss": 1.6246,
+      "step": 166200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1515431817293536e-05,
+      "loss": 1.6278,
+      "step": 166300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1510003430740305e-05,
+      "loss": 1.6363,
+      "step": 166400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.150457504418707e-05,
+      "loss": 1.6379,
+      "step": 166500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1499146657633835e-05,
+      "loss": 1.6476,
+      "step": 166600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.14937182710806e-05,
+      "loss": 1.6306,
+      "step": 166700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1488289884527366e-05,
+      "loss": 1.6411,
+      "step": 166800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.148286149797413e-05,
+      "loss": 1.6529,
+      "step": 166900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1477433111420896e-05,
+      "loss": 1.6336,
+      "step": 167000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.147200472486766e-05,
+      "loss": 1.641,
+      "step": 167100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.146657633831442e-05,
+      "loss": 1.6318,
+      "step": 167200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.146114795176119e-05,
+      "loss": 1.6243,
+      "step": 167300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.145571956520795e-05,
+      "loss": 1.6485,
+      "step": 167400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.145029117865471e-05,
+      "loss": 1.6424,
+      "step": 167500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.144486279210148e-05,
+      "loss": 1.6495,
+      "step": 167600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.143943440554825e-05,
+      "loss": 1.6417,
+      "step": 167700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.143400601899501e-05,
+      "loss": 1.6229,
+      "step": 167800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.142857763244178e-05,
+      "loss": 1.6565,
+      "step": 167900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.142314924588854e-05,
+      "loss": 1.6324,
+      "step": 168000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1417720859335304e-05,
+      "loss": 1.6411,
+      "step": 168100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.141229247278207e-05,
+      "loss": 1.6343,
+      "step": 168200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1406864086228835e-05,
+      "loss": 1.6355,
+      "step": 168300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.14014356996756e-05,
+      "loss": 1.646,
+      "step": 168400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1396007313122365e-05,
+      "loss": 1.6457,
+      "step": 168500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.139057892656913e-05,
+      "loss": 1.6516,
+      "step": 168600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.1385150540015896e-05,
+      "loss": 1.6479,
+      "step": 168700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1379722153462664e-05,
+      "loss": 1.6242,
+      "step": 168800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1374293766909426e-05,
+      "loss": 1.6569,
+      "step": 168900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1368865380356195e-05,
+      "loss": 1.6481,
+      "step": 169000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.136343699380296e-05,
+      "loss": 1.6365,
+      "step": 169100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.135800860724972e-05,
+      "loss": 1.6197,
+      "step": 169200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.135258022069649e-05,
+      "loss": 1.6592,
+      "step": 169300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.134715183414325e-05,
+      "loss": 1.6261,
+      "step": 169400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.134172344759001e-05,
+      "loss": 1.6491,
+      "step": 169500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.133629506103678e-05,
+      "loss": 1.6227,
+      "step": 169600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.133086667448354e-05,
+      "loss": 1.6486,
+      "step": 169700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.132543828793031e-05,
+      "loss": 1.6376,
+      "step": 169800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.132000990137708e-05,
+      "loss": 1.6393,
+      "step": 169900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.131458151482384e-05,
+      "loss": 1.6392,
+      "step": 170000
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 1.4881998300552368,
+      "eval_runtime": 19.0174,
+      "eval_samples_per_second": 525.834,
+      "eval_steps_per_second": 16.459,
+      "step": 170000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.13091531282706e-05,
+      "loss": 1.6026,
+      "step": 170100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.130372474171737e-05,
+      "loss": 1.6467,
+      "step": 170200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.129829635516413e-05,
+      "loss": 1.6615,
+      "step": 170300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.12928679686109e-05,
+      "loss": 1.6277,
+      "step": 170400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1287439582057664e-05,
+      "loss": 1.6062,
+      "step": 170500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1282011195504426e-05,
+      "loss": 1.6005,
+      "step": 170600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1276582808951194e-05,
+      "loss": 1.6392,
+      "step": 170700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.127115442239796e-05,
+      "loss": 1.635,
+      "step": 170800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1265726035844725e-05,
+      "loss": 1.6457,
+      "step": 170900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1260297649291493e-05,
+      "loss": 1.6484,
+      "step": 171000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1254869262738255e-05,
+      "loss": 1.6292,
+      "step": 171100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.124944087618502e-05,
+      "loss": 1.6596,
+      "step": 171200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1244012489631786e-05,
+      "loss": 1.6474,
+      "step": 171300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.123858410307855e-05,
+      "loss": 1.6271,
+      "step": 171400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.123315571652531e-05,
+      "loss": 1.6191,
+      "step": 171500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.122772732997208e-05,
+      "loss": 1.6352,
+      "step": 171600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.122229894341884e-05,
+      "loss": 1.6396,
+      "step": 171700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.121687055686561e-05,
+      "loss": 1.6545,
+      "step": 171800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.121144217031238e-05,
+      "loss": 1.6158,
+      "step": 171900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.120601378375914e-05,
+      "loss": 1.6239,
+      "step": 172000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.12005853972059e-05,
+      "loss": 1.6401,
+      "step": 172100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.119515701065267e-05,
+      "loss": 1.6423,
+      "step": 172200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.118972862409943e-05,
+      "loss": 1.633,
+      "step": 172300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.11843002375462e-05,
+      "loss": 1.6451,
+      "step": 172400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.117887185099296e-05,
+      "loss": 1.6551,
+      "step": 172500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1173443464439724e-05,
+      "loss": 1.6391,
+      "step": 172600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.116801507788649e-05,
+      "loss": 1.6313,
+      "step": 172700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1162586691333255e-05,
+      "loss": 1.6167,
+      "step": 172800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.115715830478002e-05,
+      "loss": 1.6307,
+      "step": 172900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.115172991822679e-05,
+      "loss": 1.6271,
+      "step": 173000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1146301531673554e-05,
+      "loss": 1.6487,
+      "step": 173100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1140873145120316e-05,
+      "loss": 1.6376,
+      "step": 173200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1135444758567084e-05,
+      "loss": 1.6223,
+      "step": 173300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1130016372013846e-05,
+      "loss": 1.6545,
+      "step": 173400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.112458798546061e-05,
+      "loss": 1.6432,
+      "step": 173500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.111915959890738e-05,
+      "loss": 1.6115,
+      "step": 173600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.111373121235414e-05,
+      "loss": 1.6354,
+      "step": 173700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.110830282580091e-05,
+      "loss": 1.6091,
+      "step": 173800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.110287443924767e-05,
+      "loss": 1.6004,
+      "step": 173900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.109744605269444e-05,
+      "loss": 1.6446,
+      "step": 174000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.10920176661412e-05,
+      "loss": 1.6052,
+      "step": 174100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.108658927958797e-05,
+      "loss": 1.6133,
+      "step": 174200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.108116089303473e-05,
+      "loss": 1.6266,
+      "step": 174300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.10757325064815e-05,
+      "loss": 1.6683,
+      "step": 174400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.107030411992826e-05,
+      "loss": 1.6587,
+      "step": 174500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.106487573337502e-05,
+      "loss": 1.649,
+      "step": 174600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.105944734682179e-05,
+      "loss": 1.6482,
+      "step": 174700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.105401896026855e-05,
+      "loss": 1.6345,
+      "step": 174800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.1048590573715315e-05,
+      "loss": 1.6236,
+      "step": 174900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.104316218716209e-05,
+      "loss": 1.6541,
+      "step": 175000
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 1.4871829748153687,
+      "eval_runtime": 19.0226,
+      "eval_samples_per_second": 525.691,
+      "eval_steps_per_second": 16.454,
+      "step": 175000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.103773380060885e-05,
+      "loss": 1.6537,
+      "step": 175100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.1032305414055614e-05,
+      "loss": 1.6186,
+      "step": 175200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.102687702750238e-05,
+      "loss": 1.6422,
+      "step": 175300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.1021448640949145e-05,
+      "loss": 1.6535,
+      "step": 175400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.1016020254395907e-05,
+      "loss": 1.654,
+      "step": 175500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.1010591867842675e-05,
+      "loss": 1.6388,
+      "step": 175600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.100516348128944e-05,
+      "loss": 1.6391,
+      "step": 175700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.0999735094736206e-05,
+      "loss": 1.6144,
+      "step": 175800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.099430670818297e-05,
+      "loss": 1.6391,
+      "step": 175900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.098887832162973e-05,
+      "loss": 1.6552,
+      "step": 176000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.09834499350765e-05,
+      "loss": 1.6019,
+      "step": 176100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.097802154852327e-05,
+      "loss": 1.6177,
+      "step": 176200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.097259316197003e-05,
+      "loss": 1.6369,
+      "step": 176300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.09671647754168e-05,
+      "loss": 1.6127,
+      "step": 176400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.096173638886356e-05,
+      "loss": 1.6506,
+      "step": 176500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.095630800231032e-05,
+      "loss": 1.6262,
+      "step": 176600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.095087961575709e-05,
+      "loss": 1.6192,
+      "step": 176700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.094545122920385e-05,
+      "loss": 1.6,
+      "step": 176800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.0940022842650613e-05,
+      "loss": 1.6306,
+      "step": 176900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.093459445609738e-05,
+      "loss": 1.639,
+      "step": 177000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.092916606954415e-05,
+      "loss": 1.64,
+      "step": 177100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.092373768299091e-05,
+      "loss": 1.6938,
+      "step": 177200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.091830929643768e-05,
+      "loss": 1.6229,
+      "step": 177300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.091288090988444e-05,
+      "loss": 1.6599,
+      "step": 177400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.0907452523331205e-05,
+      "loss": 1.6133,
+      "step": 177500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.0902024136777974e-05,
+      "loss": 1.6335,
+      "step": 177600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.0896595750224736e-05,
+      "loss": 1.6426,
+      "step": 177700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.0891167363671504e-05,
+      "loss": 1.6161,
+      "step": 177800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.0885738977118266e-05,
+      "loss": 1.6421,
+      "step": 177900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.088031059056503e-05,
+      "loss": 1.6161,
+      "step": 178000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.08748822040118e-05,
+      "loss": 1.6326,
+      "step": 178100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0869453817458565e-05,
+      "loss": 1.6143,
+      "step": 178200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.086402543090533e-05,
+      "loss": 1.6332,
+      "step": 178300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0858597044352096e-05,
+      "loss": 1.6198,
+      "step": 178400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.085316865779886e-05,
+      "loss": 1.6098,
+      "step": 178500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.084774027124562e-05,
+      "loss": 1.6328,
+      "step": 178600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.084231188469239e-05,
+      "loss": 1.6164,
+      "step": 178700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.083688349813915e-05,
+      "loss": 1.6256,
+      "step": 178800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.083145511158591e-05,
+      "loss": 1.6203,
+      "step": 178900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.082602672503268e-05,
+      "loss": 1.6494,
+      "step": 179000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.082059833847944e-05,
+      "loss": 1.645,
+      "step": 179100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.081516995192621e-05,
+      "loss": 1.6414,
+      "step": 179200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.080974156537298e-05,
+      "loss": 1.6364,
+      "step": 179300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.080431317881974e-05,
+      "loss": 1.6388,
+      "step": 179400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0798884792266504e-05,
+      "loss": 1.6376,
+      "step": 179500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.079345640571327e-05,
+      "loss": 1.6193,
+      "step": 179600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0788028019160034e-05,
+      "loss": 1.6593,
+      "step": 179700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.07825996326068e-05,
+      "loss": 1.6466,
+      "step": 179800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0777171246053565e-05,
+      "loss": 1.6292,
+      "step": 179900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0771742859500326e-05,
+      "loss": 1.6344,
+      "step": 180000
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 1.4833685159683228,
+      "eval_runtime": 19.0127,
+      "eval_samples_per_second": 525.965,
+      "eval_steps_per_second": 16.463,
+      "step": 180000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0766314472947095e-05,
+      "loss": 1.6491,
+      "step": 180100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.076088608639386e-05,
+      "loss": 1.6105,
+      "step": 180200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0755457699840626e-05,
+      "loss": 1.6402,
+      "step": 180300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0750029313287394e-05,
+      "loss": 1.632,
+      "step": 180400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0744600926734156e-05,
+      "loss": 1.6187,
+      "step": 180500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.073917254018092e-05,
+      "loss": 1.6083,
+      "step": 180600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.073374415362769e-05,
+      "loss": 1.6105,
+      "step": 180700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.072831576707445e-05,
+      "loss": 1.6104,
+      "step": 180800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.072288738052121e-05,
+      "loss": 1.6171,
+      "step": 180900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.071745899396798e-05,
+      "loss": 1.6525,
+      "step": 181000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.071203060741474e-05,
+      "loss": 1.6322,
+      "step": 181100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.07066022208615e-05,
+      "loss": 1.6336,
+      "step": 181200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.070117383430828e-05,
+      "loss": 1.6341,
+      "step": 181300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.069574544775504e-05,
+      "loss": 1.6351,
+      "step": 181400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.06903170612018e-05,
+      "loss": 1.6252,
+      "step": 181500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.068488867464857e-05,
+      "loss": 1.6479,
+      "step": 181600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.067946028809533e-05,
+      "loss": 1.655,
+      "step": 181700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.06740319015421e-05,
+      "loss": 1.6359,
+      "step": 181800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.066860351498886e-05,
+      "loss": 1.6313,
+      "step": 181900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0663175128435625e-05,
+      "loss": 1.6219,
+      "step": 182000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0657746741882394e-05,
+      "loss": 1.6426,
+      "step": 182100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0652318355329155e-05,
+      "loss": 1.6112,
+      "step": 182200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0646889968775924e-05,
+      "loss": 1.6419,
+      "step": 182300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.064146158222269e-05,
+      "loss": 1.6644,
+      "step": 182400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0636033195669455e-05,
+      "loss": 1.6382,
+      "step": 182500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0630604809116217e-05,
+      "loss": 1.6431,
+      "step": 182600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0625176422562985e-05,
+      "loss": 1.6237,
+      "step": 182700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.061974803600975e-05,
+      "loss": 1.66,
+      "step": 182800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.061431964945651e-05,
+      "loss": 1.6497,
+      "step": 182900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.060889126290328e-05,
+      "loss": 1.6327,
+      "step": 183000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.060346287635004e-05,
+      "loss": 1.6126,
+      "step": 183100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.05980344897968e-05,
+      "loss": 1.6265,
+      "step": 183200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.059260610324357e-05,
+      "loss": 1.6381,
+      "step": 183300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.058717771669034e-05,
+      "loss": 1.6149,
+      "step": 183400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.05817493301371e-05,
+      "loss": 1.6469,
+      "step": 183500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.057632094358387e-05,
+      "loss": 1.6346,
+      "step": 183600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.057089255703063e-05,
+      "loss": 1.6221,
+      "step": 183700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.05654641704774e-05,
+      "loss": 1.6388,
+      "step": 183800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.056003578392416e-05,
+      "loss": 1.6235,
+      "step": 183900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.0554607397370923e-05,
+      "loss": 1.6216,
+      "step": 184000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.054917901081769e-05,
+      "loss": 1.6219,
+      "step": 184100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.0543750624264454e-05,
+      "loss": 1.6149,
+      "step": 184200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.0538322237711216e-05,
+      "loss": 1.6441,
+      "step": 184300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.053289385115799e-05,
+      "loss": 1.6382,
+      "step": 184400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.052746546460475e-05,
+      "loss": 1.6103,
+      "step": 184500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.0522037078051515e-05,
+      "loss": 1.6291,
+      "step": 184600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.0516608691498284e-05,
+      "loss": 1.6306,
+      "step": 184700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.0511180304945046e-05,
+      "loss": 1.6277,
+      "step": 184800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.050575191839181e-05,
+      "loss": 1.6136,
+      "step": 184900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.0500323531838576e-05,
+      "loss": 1.6362,
+      "step": 185000
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 1.480057954788208,
+      "eval_runtime": 19.051,
+      "eval_samples_per_second": 524.905,
+      "eval_steps_per_second": 16.43,
+      "step": 185000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.049489514528534e-05,
+      "loss": 1.6264,
+      "step": 185100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.04894667587321e-05,
+      "loss": 1.6282,
+      "step": 185200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.048403837217887e-05,
+      "loss": 1.6327,
+      "step": 185300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.047860998562563e-05,
+      "loss": 1.6176,
+      "step": 185400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.04731815990724e-05,
+      "loss": 1.6385,
+      "step": 185500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.046775321251917e-05,
+      "loss": 1.6421,
+      "step": 185600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.046232482596593e-05,
+      "loss": 1.6373,
+      "step": 185700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.04568964394127e-05,
+      "loss": 1.6184,
+      "step": 185800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.045146805285946e-05,
+      "loss": 1.6273,
+      "step": 185900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.044603966630622e-05,
+      "loss": 1.6186,
+      "step": 186000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.044061127975299e-05,
+      "loss": 1.6068,
+      "step": 186100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.043518289319975e-05,
+      "loss": 1.6195,
+      "step": 186200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0429754506646514e-05,
+      "loss": 1.6136,
+      "step": 186300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.042432612009328e-05,
+      "loss": 1.6372,
+      "step": 186400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.041889773354005e-05,
+      "loss": 1.6169,
+      "step": 186500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0413469346986814e-05,
+      "loss": 1.631,
+      "step": 186600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.040804096043358e-05,
+      "loss": 1.631,
+      "step": 186700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0402612573880344e-05,
+      "loss": 1.6475,
+      "step": 186800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0397184187327106e-05,
+      "loss": 1.6576,
+      "step": 186900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0391755800773875e-05,
+      "loss": 1.6185,
+      "step": 187000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0386327414220636e-05,
+      "loss": 1.6293,
+      "step": 187100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.03808990276674e-05,
+      "loss": 1.6254,
+      "step": 187200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.037547064111417e-05,
+      "loss": 1.6058,
+      "step": 187300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.037004225456093e-05,
+      "loss": 1.5944,
+      "step": 187400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.03646138680077e-05,
+      "loss": 1.6126,
+      "step": 187500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.0359185481454466e-05,
+      "loss": 1.6438,
+      "step": 187600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.035375709490123e-05,
+      "loss": 1.6302,
+      "step": 187700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.034832870834799e-05,
+      "loss": 1.6346,
+      "step": 187800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.034290032179476e-05,
+      "loss": 1.6221,
+      "step": 187900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.033747193524152e-05,
+      "loss": 1.6381,
+      "step": 188000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.033204354868829e-05,
+      "loss": 1.607,
+      "step": 188100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.032661516213505e-05,
+      "loss": 1.5969,
+      "step": 188200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.032118677558181e-05,
+      "loss": 1.6342,
+      "step": 188300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.031575838902858e-05,
+      "loss": 1.6447,
+      "step": 188400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.031033000247534e-05,
+      "loss": 1.6283,
+      "step": 188500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.030490161592211e-05,
+      "loss": 1.6406,
+      "step": 188600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.029947322936888e-05,
+      "loss": 1.6354,
+      "step": 188700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.029404484281564e-05,
+      "loss": 1.5996,
+      "step": 188800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.0288616456262404e-05,
+      "loss": 1.6525,
+      "step": 188900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.028318806970917e-05,
+      "loss": 1.6675,
+      "step": 189000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.0277759683155935e-05,
+      "loss": 1.6242,
+      "step": 189100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.02723312966027e-05,
+      "loss": 1.6512,
+      "step": 189200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.0266902910049465e-05,
+      "loss": 1.6472,
+      "step": 189300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.026147452349623e-05,
+      "loss": 1.6366,
+      "step": 189400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.0256046136942996e-05,
+      "loss": 1.6389,
+      "step": 189500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.025061775038976e-05,
+      "loss": 1.6271,
+      "step": 189600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.0245189363836527e-05,
+      "loss": 1.6299,
+      "step": 189700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.023976097728329e-05,
+      "loss": 1.6095,
+      "step": 189800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.023433259073006e-05,
+      "loss": 1.6473,
+      "step": 189900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.022890420417682e-05,
+      "loss": 1.6325,
+      "step": 190000
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 1.4781873226165771,
+      "eval_runtime": 19.0438,
+      "eval_samples_per_second": 525.106,
+      "eval_steps_per_second": 16.436,
+      "step": 190000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.022347581762359e-05,
+      "loss": 1.6275,
+      "step": 190100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.021804743107035e-05,
+      "loss": 1.6262,
+      "step": 190200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.021261904451711e-05,
+      "loss": 1.588,
+      "step": 190300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.020719065796388e-05,
+      "loss": 1.6254,
+      "step": 190400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.020176227141064e-05,
+      "loss": 1.6347,
+      "step": 190500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0196333884857404e-05,
+      "loss": 1.6128,
+      "step": 190600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.019090549830418e-05,
+      "loss": 1.6265,
+      "step": 190700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.018547711175094e-05,
+      "loss": 1.6595,
+      "step": 190800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.01800487251977e-05,
+      "loss": 1.6308,
+      "step": 190900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.017462033864447e-05,
+      "loss": 1.6539,
+      "step": 191000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0169191952091233e-05,
+      "loss": 1.6196,
+      "step": 191100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0163763565537995e-05,
+      "loss": 1.6144,
+      "step": 191200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0158335178984764e-05,
+      "loss": 1.627,
+      "step": 191300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0152906792431526e-05,
+      "loss": 1.6389,
+      "step": 191400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0147478405878294e-05,
+      "loss": 1.5831,
+      "step": 191500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0142050019325056e-05,
+      "loss": 1.6268,
+      "step": 191600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.013662163277182e-05,
+      "loss": 1.634,
+      "step": 191700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.013119324621859e-05,
+      "loss": 1.635,
+      "step": 191800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0125764859665356e-05,
+      "loss": 1.6322,
+      "step": 191900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.012033647311212e-05,
+      "loss": 1.6223,
+      "step": 192000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0114908086558886e-05,
+      "loss": 1.6278,
+      "step": 192100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.010947970000565e-05,
+      "loss": 1.6351,
+      "step": 192200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.010405131345241e-05,
+      "loss": 1.5964,
+      "step": 192300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.009862292689918e-05,
+      "loss": 1.6657,
+      "step": 192400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.009319454034594e-05,
+      "loss": 1.6335,
+      "step": 192500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.00877661537927e-05,
+      "loss": 1.6092,
+      "step": 192600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.008233776723947e-05,
+      "loss": 1.6382,
+      "step": 192700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.007690938068624e-05,
+      "loss": 1.6323,
+      "step": 192800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0071480994133e-05,
+      "loss": 1.6095,
+      "step": 192900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.006605260757977e-05,
+      "loss": 1.6421,
+      "step": 193000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.006062422102653e-05,
+      "loss": 1.6308,
+      "step": 193100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0055195834473294e-05,
+      "loss": 1.6324,
+      "step": 193200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.004976744792006e-05,
+      "loss": 1.6081,
+      "step": 193300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0044339061366824e-05,
+      "loss": 1.6247,
+      "step": 193400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.003891067481359e-05,
+      "loss": 1.5981,
+      "step": 193500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0033482288260355e-05,
+      "loss": 1.6262,
+      "step": 193600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.002805390170712e-05,
+      "loss": 1.6318,
+      "step": 193700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0022625515153885e-05,
+      "loss": 1.6471,
+      "step": 193800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0017197128600654e-05,
+      "loss": 1.5975,
+      "step": 193900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0011768742047416e-05,
+      "loss": 1.5955,
+      "step": 194000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0006340355494185e-05,
+      "loss": 1.6424,
+      "step": 194100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0000911968940946e-05,
+      "loss": 1.6297,
+      "step": 194200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.999548358238771e-05,
+      "loss": 1.5919,
+      "step": 194300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.999005519583448e-05,
+      "loss": 1.6007,
+      "step": 194400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.998462680928124e-05,
+      "loss": 1.6243,
+      "step": 194500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9979198422728e-05,
+      "loss": 1.632,
+      "step": 194600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.997377003617477e-05,
+      "loss": 1.6295,
+      "step": 194700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.996834164962153e-05,
+      "loss": 1.6243,
+      "step": 194800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.99629132630683e-05,
+      "loss": 1.6342,
+      "step": 194900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.995748487651507e-05,
+      "loss": 1.6255,
+      "step": 195000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 1.4795633554458618,
+      "eval_runtime": 19.0393,
+      "eval_samples_per_second": 525.231,
+      "eval_steps_per_second": 16.44,
+      "step": 195000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.995205648996183e-05,
+      "loss": 1.6733,
+      "step": 195100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.994662810340859e-05,
+      "loss": 1.5987,
+      "step": 195200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.994119971685536e-05,
+      "loss": 1.6215,
+      "step": 195300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.993577133030212e-05,
+      "loss": 1.6399,
+      "step": 195400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.993034294374889e-05,
+      "loss": 1.6229,
+      "step": 195500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.992491455719565e-05,
+      "loss": 1.639,
+      "step": 195600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9919486170642415e-05,
+      "loss": 1.6256,
+      "step": 195700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9914057784089184e-05,
+      "loss": 1.6476,
+      "step": 195800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9908629397535946e-05,
+      "loss": 1.5958,
+      "step": 195900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9903201010982714e-05,
+      "loss": 1.6397,
+      "step": 196000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.989777262442948e-05,
+      "loss": 1.6466,
+      "step": 196100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9892344237876245e-05,
+      "loss": 1.6343,
+      "step": 196200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.988691585132301e-05,
+      "loss": 1.5967,
+      "step": 196300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9881487464769775e-05,
+      "loss": 1.6249,
+      "step": 196400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.987605907821654e-05,
+      "loss": 1.6111,
+      "step": 196500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.98706306916633e-05,
+      "loss": 1.6295,
+      "step": 196600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.986520230511007e-05,
+      "loss": 1.6362,
+      "step": 196700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.985977391855683e-05,
+      "loss": 1.6522,
+      "step": 196800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.98543455320036e-05,
+      "loss": 1.6481,
+      "step": 196900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.984891714545037e-05,
+      "loss": 1.6194,
+      "step": 197000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.984348875889713e-05,
+      "loss": 1.6086,
+      "step": 197100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.983806037234389e-05,
+      "loss": 1.6251,
+      "step": 197200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.983263198579066e-05,
+      "loss": 1.6097,
+      "step": 197300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.982720359923742e-05,
+      "loss": 1.6268,
+      "step": 197400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.982177521268419e-05,
+      "loss": 1.6386,
+      "step": 197500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.981634682613095e-05,
+      "loss": 1.6207,
+      "step": 197600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9810918439577714e-05,
+      "loss": 1.6248,
+      "step": 197700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.980549005302448e-05,
+      "loss": 1.6337,
+      "step": 197800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9800061666471244e-05,
+      "loss": 1.6195,
+      "step": 197900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9794633279918006e-05,
+      "loss": 1.6396,
+      "step": 198000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.978920489336478e-05,
+      "loss": 1.5919,
+      "step": 198100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.978377650681154e-05,
+      "loss": 1.6144,
+      "step": 198200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9778348120258305e-05,
+      "loss": 1.6001,
+      "step": 198300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9772919733705074e-05,
+      "loss": 1.6218,
+      "step": 198400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9767491347151836e-05,
+      "loss": 1.6068,
+      "step": 198500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.97620629605986e-05,
+      "loss": 1.6075,
+      "step": 198600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9756634574045366e-05,
+      "loss": 1.5883,
+      "step": 198700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.975120618749213e-05,
+      "loss": 1.5984,
+      "step": 198800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.97457778009389e-05,
+      "loss": 1.6311,
+      "step": 198900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.974034941438566e-05,
+      "loss": 1.6614,
+      "step": 199000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.973492102783243e-05,
+      "loss": 1.6541,
+      "step": 199100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.972949264127919e-05,
+      "loss": 1.6475,
+      "step": 199200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.972406425472596e-05,
+      "loss": 1.6138,
+      "step": 199300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.971863586817272e-05,
+      "loss": 1.6249,
+      "step": 199400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.971320748161949e-05,
+      "loss": 1.6141,
+      "step": 199500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.970777909506625e-05,
+      "loss": 1.6344,
+      "step": 199600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.970235070851301e-05,
+      "loss": 1.613,
+      "step": 199700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.969692232195978e-05,
+      "loss": 1.616,
+      "step": 199800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.969149393540654e-05,
+      "loss": 1.6162,
+      "step": 199900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9686065548853305e-05,
+      "loss": 1.6455,
+      "step": 200000
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 1.4754244089126587,
+      "eval_runtime": 18.9946,
+      "eval_samples_per_second": 526.464,
+      "eval_steps_per_second": 16.478,
+      "step": 200000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.968063716230007e-05,
+      "loss": 1.6311,
+      "step": 200100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.967520877574684e-05,
+      "loss": 1.6139,
+      "step": 200200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9669780389193604e-05,
+      "loss": 1.601,
+      "step": 200300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.966435200264037e-05,
+      "loss": 1.6321,
+      "step": 200400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9658923616087134e-05,
+      "loss": 1.6582,
+      "step": 200500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9653495229533896e-05,
+      "loss": 1.6293,
+      "step": 200600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9648066842980665e-05,
+      "loss": 1.6144,
+      "step": 200700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.964263845642743e-05,
+      "loss": 1.6154,
+      "step": 200800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9637210069874195e-05,
+      "loss": 1.6168,
+      "step": 200900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.963178168332096e-05,
+      "loss": 1.6595,
+      "step": 201000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.962635329676772e-05,
+      "loss": 1.6051,
+      "step": 201100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.962092491021449e-05,
+      "loss": 1.6103,
+      "step": 201200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9615496523661256e-05,
+      "loss": 1.6523,
+      "step": 201300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.961006813710802e-05,
+      "loss": 1.6216,
+      "step": 201400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.960463975055479e-05,
+      "loss": 1.6684,
+      "step": 201500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.959921136400155e-05,
+      "loss": 1.6098,
+      "step": 201600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.959378297744831e-05,
+      "loss": 1.6309,
+      "step": 201700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.958835459089508e-05,
+      "loss": 1.6329,
+      "step": 201800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.958292620434184e-05,
+      "loss": 1.6424,
+      "step": 201900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.95774978177886e-05,
+      "loss": 1.6526,
+      "step": 202000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.957206943123537e-05,
+      "loss": 1.6335,
+      "step": 202100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.956664104468214e-05,
+      "loss": 1.6238,
+      "step": 202200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.95612126581289e-05,
+      "loss": 1.6289,
+      "step": 202300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.955578427157567e-05,
+      "loss": 1.6357,
+      "step": 202400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.955035588502243e-05,
+      "loss": 1.6201,
+      "step": 202500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9544927498469195e-05,
+      "loss": 1.6048,
+      "step": 202600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.953949911191596e-05,
+      "loss": 1.6136,
+      "step": 202700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9534070725362725e-05,
+      "loss": 1.6227,
+      "step": 202800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9528642338809494e-05,
+      "loss": 1.6221,
+      "step": 202900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9523213952256256e-05,
+      "loss": 1.6349,
+      "step": 203000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.951778556570302e-05,
+      "loss": 1.6481,
+      "step": 203100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9512357179149786e-05,
+      "loss": 1.614,
+      "step": 203200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9506928792596555e-05,
+      "loss": 1.6304,
+      "step": 203300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.950150040604332e-05,
+      "loss": 1.6152,
+      "step": 203400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9496072019490085e-05,
+      "loss": 1.6251,
+      "step": 203500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.949064363293685e-05,
+      "loss": 1.6263,
+      "step": 203600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.948521524638361e-05,
+      "loss": 1.6257,
+      "step": 203700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.947978685983038e-05,
+      "loss": 1.6283,
+      "step": 203800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.947435847327714e-05,
+      "loss": 1.6304,
+      "step": 203900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.94689300867239e-05,
+      "loss": 1.6152,
+      "step": 204000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.946350170017067e-05,
+      "loss": 1.6025,
+      "step": 204100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.945807331361743e-05,
+      "loss": 1.63,
+      "step": 204200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.94526449270642e-05,
+      "loss": 1.611,
+      "step": 204300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.944721654051097e-05,
+      "loss": 1.6312,
+      "step": 204400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.944178815395773e-05,
+      "loss": 1.6026,
+      "step": 204500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.943635976740449e-05,
+      "loss": 1.6058,
+      "step": 204600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.943093138085126e-05,
+      "loss": 1.644,
+      "step": 204700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9425502994298024e-05,
+      "loss": 1.6163,
+      "step": 204800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.942007460774479e-05,
+      "loss": 1.6285,
+      "step": 204900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9414646221191554e-05,
+      "loss": 1.653,
+      "step": 205000
+    },
+    {
+      "epoch": 0.88,
+      "eval_loss": 1.4734227657318115,
+      "eval_runtime": 19.0238,
+      "eval_samples_per_second": 525.657,
+      "eval_steps_per_second": 16.453,
+      "step": 205000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9409217834638316e-05,
+      "loss": 1.6065,
+      "step": 205100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9403789448085085e-05,
+      "loss": 1.6388,
+      "step": 205200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9398361061531847e-05,
+      "loss": 1.6096,
+      "step": 205300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9392932674978615e-05,
+      "loss": 1.6194,
+      "step": 205400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9387504288425384e-05,
+      "loss": 1.6156,
+      "step": 205500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9382075901872146e-05,
+      "loss": 1.6191,
+      "step": 205600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.937664751531891e-05,
+      "loss": 1.6216,
+      "step": 205700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9371219128765676e-05,
+      "loss": 1.6099,
+      "step": 205800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.936579074221244e-05,
+      "loss": 1.6193,
+      "step": 205900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.93603623556592e-05,
+      "loss": 1.6379,
+      "step": 206000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.935493396910597e-05,
+      "loss": 1.5792,
+      "step": 206100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.934950558255273e-05,
+      "loss": 1.6132,
+      "step": 206200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.934407719599949e-05,
+      "loss": 1.598,
+      "step": 206300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.933864880944627e-05,
+      "loss": 1.656,
+      "step": 206400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.933322042289303e-05,
+      "loss": 1.6255,
+      "step": 206500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.932779203633979e-05,
+      "loss": 1.6058,
+      "step": 206600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.932236364978656e-05,
+      "loss": 1.6469,
+      "step": 206700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.931693526323332e-05,
+      "loss": 1.6283,
+      "step": 206800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.931150687668009e-05,
+      "loss": 1.6318,
+      "step": 206900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.930607849012685e-05,
+      "loss": 1.6243,
+      "step": 207000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9300650103573615e-05,
+      "loss": 1.6538,
+      "step": 207100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.929522171702038e-05,
+      "loss": 1.6271,
+      "step": 207200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9289793330467145e-05,
+      "loss": 1.6344,
+      "step": 207300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.928436494391391e-05,
+      "loss": 1.5985,
+      "step": 207400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.927893655736068e-05,
+      "loss": 1.5742,
+      "step": 207500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9273508170807444e-05,
+      "loss": 1.5863,
+      "step": 207600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9268079784254206e-05,
+      "loss": 1.6366,
+      "step": 207700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9262651397700975e-05,
+      "loss": 1.5802,
+      "step": 207800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.925722301114774e-05,
+      "loss": 1.6049,
+      "step": 207900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.92517946245945e-05,
+      "loss": 1.6003,
+      "step": 208000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.924636623804127e-05,
+      "loss": 1.6243,
+      "step": 208100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.924093785148803e-05,
+      "loss": 1.617,
+      "step": 208200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.923550946493479e-05,
+      "loss": 1.6359,
+      "step": 208300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.923008107838156e-05,
+      "loss": 1.622,
+      "step": 208400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.922465269182833e-05,
+      "loss": 1.5969,
+      "step": 208500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.921922430527509e-05,
+      "loss": 1.6187,
+      "step": 208600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.921379591872186e-05,
+      "loss": 1.6107,
+      "step": 208700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.920836753216862e-05,
+      "loss": 1.6074,
+      "step": 208800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.920293914561539e-05,
+      "loss": 1.644,
+      "step": 208900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.919751075906215e-05,
+      "loss": 1.5991,
+      "step": 209000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.919208237250891e-05,
+      "loss": 1.5961,
+      "step": 209100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.918665398595568e-05,
+      "loss": 1.6598,
+      "step": 209200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9181225599402444e-05,
+      "loss": 1.5988,
+      "step": 209300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9175797212849205e-05,
+      "loss": 1.6194,
+      "step": 209400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9170368826295974e-05,
+      "loss": 1.6176,
+      "step": 209500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.916494043974274e-05,
+      "loss": 1.6265,
+      "step": 209600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9159512053189505e-05,
+      "loss": 1.5909,
+      "step": 209700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.915408366663627e-05,
+      "loss": 1.626,
+      "step": 209800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9148655280083035e-05,
+      "loss": 1.5911,
+      "step": 209900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.91432268935298e-05,
+      "loss": 1.6366,
+      "step": 210000
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 1.4706979990005493,
+      "eval_runtime": 19.0444,
+      "eval_samples_per_second": 525.088,
+      "eval_steps_per_second": 16.435,
+      "step": 210000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9137798506976566e-05,
+      "loss": 1.6034,
+      "step": 210100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.913237012042333e-05,
+      "loss": 1.5996,
+      "step": 210200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.912694173387009e-05,
+      "loss": 1.6335,
+      "step": 210300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.912151334731686e-05,
+      "loss": 1.6271,
+      "step": 210400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.911608496076362e-05,
+      "loss": 1.6404,
+      "step": 210500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.911065657421039e-05,
+      "loss": 1.5918,
+      "step": 210600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.910522818765716e-05,
+      "loss": 1.6124,
+      "step": 210700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.909979980110392e-05,
+      "loss": 1.5966,
+      "step": 210800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.909437141455069e-05,
+      "loss": 1.6116,
+      "step": 210900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.908894302799745e-05,
+      "loss": 1.6332,
+      "step": 211000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.908351464144421e-05,
+      "loss": 1.6542,
+      "step": 211100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.907808625489098e-05,
+      "loss": 1.6089,
+      "step": 211200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.907265786833774e-05,
+      "loss": 1.6082,
+      "step": 211300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9067229481784504e-05,
+      "loss": 1.5931,
+      "step": 211400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.906180109523127e-05,
+      "loss": 1.6417,
+      "step": 211500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9056372708678034e-05,
+      "loss": 1.6244,
+      "step": 211600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.90509443221248e-05,
+      "loss": 1.5985,
+      "step": 211700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.904551593557157e-05,
+      "loss": 1.6141,
+      "step": 211800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9040087549018334e-05,
+      "loss": 1.6266,
+      "step": 211900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9034659162465096e-05,
+      "loss": 1.5968,
+      "step": 212000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9029285059777393e-05,
+      "loss": 1.6204,
+      "step": 212100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9023856673224155e-05,
+      "loss": 1.6088,
+      "step": 212200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9018428286670924e-05,
+      "loss": 1.6306,
+      "step": 212300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.901299990011769e-05,
+      "loss": 1.652,
+      "step": 212400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9007625797429984e-05,
+      "loss": 1.6179,
+      "step": 212500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.900219741087675e-05,
+      "loss": 1.6265,
+      "step": 212600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8996769024323514e-05,
+      "loss": 1.6164,
+      "step": 212700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8991340637770276e-05,
+      "loss": 1.6125,
+      "step": 212800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8985912251217045e-05,
+      "loss": 1.6091,
+      "step": 212900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8980483864663813e-05,
+      "loss": 1.5873,
+      "step": 213000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8975055478110575e-05,
+      "loss": 1.5956,
+      "step": 213100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8969627091557344e-05,
+      "loss": 1.6119,
+      "step": 213200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8964198705004106e-05,
+      "loss": 1.6209,
+      "step": 213300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8958770318450874e-05,
+      "loss": 1.6383,
+      "step": 213400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8953341931897636e-05,
+      "loss": 1.6231,
+      "step": 213500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.89479135453444e-05,
+      "loss": 1.6086,
+      "step": 213600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.894248515879117e-05,
+      "loss": 1.6127,
+      "step": 213700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.893705677223793e-05,
+      "loss": 1.6427,
+      "step": 213800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.893162838568469e-05,
+      "loss": 1.619,
+      "step": 213900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.892619999913146e-05,
+      "loss": 1.6003,
+      "step": 214000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.892077161257823e-05,
+      "loss": 1.6161,
+      "step": 214100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.891534322602499e-05,
+      "loss": 1.6027,
+      "step": 214200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.890991483947176e-05,
+      "loss": 1.6419,
+      "step": 214300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.890448645291852e-05,
+      "loss": 1.6429,
+      "step": 214400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.889905806636528e-05,
+      "loss": 1.6234,
+      "step": 214500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.889362967981205e-05,
+      "loss": 1.617,
+      "step": 214600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.888820129325881e-05,
+      "loss": 1.6263,
+      "step": 214700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8882772906705575e-05,
+      "loss": 1.6383,
+      "step": 214800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.887734452015234e-05,
+      "loss": 1.6223,
+      "step": 214900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8871916133599105e-05,
+      "loss": 1.5949,
+      "step": 215000
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 1.4699814319610596,
+      "eval_runtime": 19.0028,
+      "eval_samples_per_second": 526.238,
+      "eval_steps_per_second": 16.471,
+      "step": 215000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8866487747045874e-05,
+      "loss": 1.616,
+      "step": 215100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.886105936049264e-05,
+      "loss": 1.6109,
+      "step": 215200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8855630973939404e-05,
+      "loss": 1.6242,
+      "step": 215300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.885020258738617e-05,
+      "loss": 1.5882,
+      "step": 215400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8844774200832935e-05,
+      "loss": 1.6162,
+      "step": 215500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.88393458142797e-05,
+      "loss": 1.6346,
+      "step": 215600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8833917427726465e-05,
+      "loss": 1.6047,
+      "step": 215700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.882848904117323e-05,
+      "loss": 1.6118,
+      "step": 215800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.882306065461999e-05,
+      "loss": 1.5989,
+      "step": 215900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.881763226806676e-05,
+      "loss": 1.6242,
+      "step": 216000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8812258165379056e-05,
+      "loss": 1.6333,
+      "step": 216100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.880682977882582e-05,
+      "loss": 1.5985,
+      "step": 216200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8801401392272586e-05,
+      "loss": 1.5896,
+      "step": 216300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.879597300571935e-05,
+      "loss": 1.5957,
+      "step": 216400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.879054461916612e-05,
+      "loss": 1.5785,
+      "step": 216500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.878511623261288e-05,
+      "loss": 1.6345,
+      "step": 216600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.877968784605964e-05,
+      "loss": 1.5993,
+      "step": 216700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.877425945950641e-05,
+      "loss": 1.6124,
+      "step": 216800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.876883107295318e-05,
+      "loss": 1.6128,
+      "step": 216900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.876340268639994e-05,
+      "loss": 1.5962,
+      "step": 217000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.875797429984671e-05,
+      "loss": 1.5918,
+      "step": 217100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.875254591329347e-05,
+      "loss": 1.5932,
+      "step": 217200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.874711752674023e-05,
+      "loss": 1.5982,
+      "step": 217300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8741689140187e-05,
+      "loss": 1.6113,
+      "step": 217400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.873626075363376e-05,
+      "loss": 1.6173,
+      "step": 217500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8730832367080524e-05,
+      "loss": 1.6258,
+      "step": 217600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.872540398052729e-05,
+      "loss": 1.6248,
+      "step": 217700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8719975593974055e-05,
+      "loss": 1.608,
+      "step": 217800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8714547207420824e-05,
+      "loss": 1.6077,
+      "step": 217900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.870911882086759e-05,
+      "loss": 1.6182,
+      "step": 218000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8703690434314354e-05,
+      "loss": 1.5938,
+      "step": 218100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8698262047761116e-05,
+      "loss": 1.5984,
+      "step": 218200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8692833661207885e-05,
+      "loss": 1.6177,
+      "step": 218300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8687405274654647e-05,
+      "loss": 1.6058,
+      "step": 218400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8681976888101415e-05,
+      "loss": 1.5896,
+      "step": 218500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.867654850154818e-05,
+      "loss": 1.5958,
+      "step": 218600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.867112011499494e-05,
+      "loss": 1.6201,
+      "step": 218700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.866569172844171e-05,
+      "loss": 1.6015,
+      "step": 218800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.866026334188847e-05,
+      "loss": 1.6328,
+      "step": 218900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.865483495533524e-05,
+      "loss": 1.6127,
+      "step": 219000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.864940656878201e-05,
+      "loss": 1.6269,
+      "step": 219100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.864397818222877e-05,
+      "loss": 1.6029,
+      "step": 219200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.863854979567553e-05,
+      "loss": 1.6364,
+      "step": 219300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.86331214091223e-05,
+      "loss": 1.6451,
+      "step": 219400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.862769302256906e-05,
+      "loss": 1.5971,
+      "step": 219500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.862226463601582e-05,
+      "loss": 1.6272,
+      "step": 219600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.861683624946259e-05,
+      "loss": 1.5922,
+      "step": 219700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8611407862909353e-05,
+      "loss": 1.6224,
+      "step": 219800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.860597947635612e-05,
+      "loss": 1.6,
+      "step": 219900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.860055108980289e-05,
+      "loss": 1.5939,
+      "step": 220000
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 1.468553066253662,
+      "eval_runtime": 19.0907,
+      "eval_samples_per_second": 523.816,
+      "eval_steps_per_second": 16.395,
+      "step": 220000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.859517698711518e-05,
+      "loss": 1.5755,
+      "step": 220100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.858974860056195e-05,
+      "loss": 1.6354,
+      "step": 220200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.858432021400871e-05,
+      "loss": 1.6106,
+      "step": 220300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8578891827455474e-05,
+      "loss": 1.6072,
+      "step": 220400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.857351772476778e-05,
+      "loss": 1.6048,
+      "step": 220500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.856808933821454e-05,
+      "loss": 1.6102,
+      "step": 220600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.85626609516613e-05,
+      "loss": 1.5699,
+      "step": 220700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.855723256510807e-05,
+      "loss": 1.6345,
+      "step": 220800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.855180417855483e-05,
+      "loss": 1.5963,
+      "step": 220900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.85463757920016e-05,
+      "loss": 1.6111,
+      "step": 221000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8540947405448364e-05,
+      "loss": 1.6211,
+      "step": 221100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8535519018895126e-05,
+      "loss": 1.6116,
+      "step": 221200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8530090632341894e-05,
+      "loss": 1.5874,
+      "step": 221300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.852466224578866e-05,
+      "loss": 1.5935,
+      "step": 221400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8519233859235425e-05,
+      "loss": 1.6263,
+      "step": 221500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8513805472682194e-05,
+      "loss": 1.6139,
+      "step": 221600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8508377086128955e-05,
+      "loss": 1.5891,
+      "step": 221700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.850294869957572e-05,
+      "loss": 1.6065,
+      "step": 221800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8497520313022486e-05,
+      "loss": 1.6147,
+      "step": 221900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.849209192646925e-05,
+      "loss": 1.586,
+      "step": 222000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.848666353991601e-05,
+      "loss": 1.6067,
+      "step": 222100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8481289437228314e-05,
+      "loss": 1.6126,
+      "step": 222200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8475861050675076e-05,
+      "loss": 1.6568,
+      "step": 222300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8470432664121845e-05,
+      "loss": 1.6035,
+      "step": 222400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.846500427756861e-05,
+      "loss": 1.601,
+      "step": 222500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.845957589101537e-05,
+      "loss": 1.5908,
+      "step": 222600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.845414750446214e-05,
+      "loss": 1.6119,
+      "step": 222700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.84487191179089e-05,
+      "loss": 1.6209,
+      "step": 222800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.844329073135566e-05,
+      "loss": 1.6223,
+      "step": 222900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.843786234480243e-05,
+      "loss": 1.6028,
+      "step": 223000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.84324339582492e-05,
+      "loss": 1.5922,
+      "step": 223100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.842700557169596e-05,
+      "loss": 1.6253,
+      "step": 223200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.842157718514273e-05,
+      "loss": 1.586,
+      "step": 223300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.841614879858949e-05,
+      "loss": 1.6246,
+      "step": 223400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.841072041203625e-05,
+      "loss": 1.5854,
+      "step": 223500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.840529202548302e-05,
+      "loss": 1.6129,
+      "step": 223600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.839986363892978e-05,
+      "loss": 1.6101,
+      "step": 223700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8394435252376545e-05,
+      "loss": 1.5877,
+      "step": 223800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8389006865823314e-05,
+      "loss": 1.6158,
+      "step": 223900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8383578479270076e-05,
+      "loss": 1.6209,
+      "step": 224000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8378150092716844e-05,
+      "loss": 1.614,
+      "step": 224100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.837272170616361e-05,
+      "loss": 1.614,
+      "step": 224200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8367293319610375e-05,
+      "loss": 1.5698,
+      "step": 224300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8361864933057143e-05,
+      "loss": 1.6134,
+      "step": 224400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8356436546503905e-05,
+      "loss": 1.6126,
+      "step": 224500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.835100815995067e-05,
+      "loss": 1.6149,
+      "step": 224600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8345579773397436e-05,
+      "loss": 1.5913,
+      "step": 224700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.83401513868442e-05,
+      "loss": 1.6153,
+      "step": 224800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.833472300029096e-05,
+      "loss": 1.6078,
+      "step": 224900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.832929461373773e-05,
+      "loss": 1.6227,
+      "step": 225000
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 1.4662607908248901,
+      "eval_runtime": 19.0457,
+      "eval_samples_per_second": 525.054,
+      "eval_steps_per_second": 16.434,
+      "step": 225000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.832386622718449e-05,
+      "loss": 1.6016,
+      "step": 225100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.831843784063126e-05,
+      "loss": 1.6083,
+      "step": 225200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.831300945407803e-05,
+      "loss": 1.6064,
+      "step": 225300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.830758106752479e-05,
+      "loss": 1.5936,
+      "step": 225400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.830215268097155e-05,
+      "loss": 1.6125,
+      "step": 225500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.829672429441832e-05,
+      "loss": 1.6469,
+      "step": 225600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.829129590786508e-05,
+      "loss": 1.644,
+      "step": 225700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8285867521311844e-05,
+      "loss": 1.5836,
+      "step": 225800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.828043913475861e-05,
+      "loss": 1.6373,
+      "step": 225900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8275010748205374e-05,
+      "loss": 1.6034,
+      "step": 226000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.826958236165214e-05,
+      "loss": 1.6096,
+      "step": 226100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.826415397509891e-05,
+      "loss": 1.5963,
+      "step": 226200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.825872558854567e-05,
+      "loss": 1.597,
+      "step": 226300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8253297201992435e-05,
+      "loss": 1.582,
+      "step": 226400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8247868815439204e-05,
+      "loss": 1.5919,
+      "step": 226500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8242440428885966e-05,
+      "loss": 1.6027,
+      "step": 226600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8237012042332734e-05,
+      "loss": 1.6298,
+      "step": 226700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8231583655779496e-05,
+      "loss": 1.6083,
+      "step": 226800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.822615526922626e-05,
+      "loss": 1.5936,
+      "step": 226900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.822072688267303e-05,
+      "loss": 1.6154,
+      "step": 227000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.821529849611979e-05,
+      "loss": 1.6154,
+      "step": 227100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.820987010956655e-05,
+      "loss": 1.6051,
+      "step": 227200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8204441723013326e-05,
+      "loss": 1.625,
+      "step": 227300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.819901333646009e-05,
+      "loss": 1.6175,
+      "step": 227400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.819358494990685e-05,
+      "loss": 1.6182,
+      "step": 227500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.818815656335362e-05,
+      "loss": 1.5886,
+      "step": 227600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.818272817680038e-05,
+      "loss": 1.6225,
+      "step": 227700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.817729979024714e-05,
+      "loss": 1.6206,
+      "step": 227800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.817187140369391e-05,
+      "loss": 1.5953,
+      "step": 227900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.816644301714067e-05,
+      "loss": 1.6174,
+      "step": 228000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.816101463058744e-05,
+      "loss": 1.6002,
+      "step": 228100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.81555862440342e-05,
+      "loss": 1.6052,
+      "step": 228200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.815015785748097e-05,
+      "loss": 1.6123,
+      "step": 228300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8144729470927734e-05,
+      "loss": 1.6096,
+      "step": 228400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.81393010843745e-05,
+      "loss": 1.6018,
+      "step": 228500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8133872697821264e-05,
+      "loss": 1.5998,
+      "step": 228600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.812844431126803e-05,
+      "loss": 1.5882,
+      "step": 228700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8123015924714795e-05,
+      "loss": 1.6218,
+      "step": 228800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8117587538161557e-05,
+      "loss": 1.6005,
+      "step": 228900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8112159151608325e-05,
+      "loss": 1.587,
+      "step": 229000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.810673076505509e-05,
+      "loss": 1.5893,
+      "step": 229100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.810130237850185e-05,
+      "loss": 1.5836,
+      "step": 229200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.809587399194862e-05,
+      "loss": 1.595,
+      "step": 229300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8090445605395386e-05,
+      "loss": 1.6032,
+      "step": 229400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.808501721884215e-05,
+      "loss": 1.5978,
+      "step": 229500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.807958883228892e-05,
+      "loss": 1.5997,
+      "step": 229600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.807416044573568e-05,
+      "loss": 1.5946,
+      "step": 229700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.806873205918244e-05,
+      "loss": 1.5957,
+      "step": 229800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.806330367262921e-05,
+      "loss": 1.5728,
+      "step": 229900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.805787528607597e-05,
+      "loss": 1.6266,
+      "step": 230000
+    },
+    {
+      "epoch": 0.99,
+      "eval_loss": 1.4668673276901245,
+      "eval_runtime": 17.8239,
+      "eval_samples_per_second": 561.046,
+      "eval_steps_per_second": 17.561,
+      "step": 230000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.805244689952274e-05,
+      "loss": 1.6237,
+      "step": 230100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.80470185129695e-05,
+      "loss": 1.5906,
+      "step": 230200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8041590126416263e-05,
+      "loss": 1.5936,
+      "step": 230300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.803616173986303e-05,
+      "loss": 1.6093,
+      "step": 230400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.80307333533098e-05,
+      "loss": 1.6156,
+      "step": 230500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.802530496675656e-05,
+      "loss": 1.5972,
+      "step": 230600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.801987658020333e-05,
+      "loss": 1.6244,
+      "step": 230700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.801444819365009e-05,
+      "loss": 1.5729,
+      "step": 230800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8009019807096855e-05,
+      "loss": 1.5968,
+      "step": 230900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8003591420543624e-05,
+      "loss": 1.6167,
+      "step": 231000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7998163033990386e-05,
+      "loss": 1.6071,
+      "step": 231100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.799273464743715e-05,
+      "loss": 1.5794,
+      "step": 231200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7987306260883916e-05,
+      "loss": 1.6213,
+      "step": 231300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.798187787433068e-05,
+      "loss": 1.6047,
+      "step": 231400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7976449487777447e-05,
+      "loss": 1.6175,
+      "step": 231500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7971021101224215e-05,
+      "loss": 1.618,
+      "step": 231600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.796559271467098e-05,
+      "loss": 1.5891,
+      "step": 231700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.796016432811774e-05,
+      "loss": 1.5905,
+      "step": 231800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.795473594156451e-05,
+      "loss": 1.5845,
+      "step": 231900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.794930755501127e-05,
+      "loss": 1.5899,
+      "step": 232000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.794387916845804e-05,
+      "loss": 1.5894,
+      "step": 232100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.79384507819048e-05,
+      "loss": 1.6205,
+      "step": 232200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.793302239535156e-05,
+      "loss": 1.5876,
+      "step": 232300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.792759400879833e-05,
+      "loss": 1.5929,
+      "step": 232400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.79221656222451e-05,
+      "loss": 1.6052,
+      "step": 232500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.791673723569186e-05,
+      "loss": 1.5764,
+      "step": 232600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.791130884913863e-05,
+      "loss": 1.6319,
+      "step": 232700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.790588046258539e-05,
+      "loss": 1.6024,
+      "step": 232800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7900452076032154e-05,
+      "loss": 1.5373,
+      "step": 232900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.789502368947892e-05,
+      "loss": 1.5602,
+      "step": 233000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7889595302925684e-05,
+      "loss": 1.562,
+      "step": 233100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7884166916372446e-05,
+      "loss": 1.5622,
+      "step": 233200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7878738529819215e-05,
+      "loss": 1.5711,
+      "step": 233300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7873310143265976e-05,
+      "loss": 1.5624,
+      "step": 233400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7867881756712745e-05,
+      "loss": 1.622,
+      "step": 233500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7862453370159514e-05,
+      "loss": 1.5584,
+      "step": 233600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7857024983606276e-05,
+      "loss": 1.5626,
+      "step": 233700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.785159659705304e-05,
+      "loss": 1.5756,
+      "step": 233800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7846168210499806e-05,
+      "loss": 1.6049,
+      "step": 233900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.784073982394657e-05,
+      "loss": 1.5274,
+      "step": 234000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.783531143739334e-05,
+      "loss": 1.5675,
+      "step": 234100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.78298830508401e-05,
+      "loss": 1.5812,
+      "step": 234200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.782445466428686e-05,
+      "loss": 1.5798,
+      "step": 234300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.781902627773363e-05,
+      "loss": 1.5817,
+      "step": 234400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.781359789118039e-05,
+      "loss": 1.5589,
+      "step": 234500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.780816950462716e-05,
+      "loss": 1.5635,
+      "step": 234600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.780274111807393e-05,
+      "loss": 1.5495,
+      "step": 234700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.779731273152069e-05,
+      "loss": 1.5788,
+      "step": 234800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.779188434496745e-05,
+      "loss": 1.5325,
+      "step": 234900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.778645595841422e-05,
+      "loss": 1.582,
+      "step": 235000
+    },
+    {
+      "epoch": 1.01,
+      "eval_loss": 1.4644287824630737,
+      "eval_runtime": 17.7873,
+      "eval_samples_per_second": 562.198,
+      "eval_steps_per_second": 17.597,
+      "step": 235000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.778102757186098e-05,
+      "loss": 1.5531,
+      "step": 235100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7775599185307744e-05,
+      "loss": 1.5584,
+      "step": 235200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.777017079875451e-05,
+      "loss": 1.5617,
+      "step": 235300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7764742412201275e-05,
+      "loss": 1.5662,
+      "step": 235400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7759314025648044e-05,
+      "loss": 1.5365,
+      "step": 235500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7753885639094805e-05,
+      "loss": 1.5919,
+      "step": 235600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7748457252541574e-05,
+      "loss": 1.5485,
+      "step": 235700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7743028865988336e-05,
+      "loss": 1.5795,
+      "step": 235800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7737600479435105e-05,
+      "loss": 1.5378,
+      "step": 235900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7732172092881866e-05,
+      "loss": 1.5702,
+      "step": 236000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7726743706328635e-05,
+      "loss": 1.586,
+      "step": 236100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.77213153197754e-05,
+      "loss": 1.568,
+      "step": 236200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.771588693322216e-05,
+      "loss": 1.5939,
+      "step": 236300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.771045854666893e-05,
+      "loss": 1.5841,
+      "step": 236400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.770503016011569e-05,
+      "loss": 1.5841,
+      "step": 236500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.769960177356245e-05,
+      "loss": 1.543,
+      "step": 236600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.769417338700923e-05,
+      "loss": 1.5428,
+      "step": 236700
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.768874500045599e-05,
+      "loss": 1.5519,
+      "step": 236800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.768331661390275e-05,
+      "loss": 1.5362,
+      "step": 236900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.767788822734952e-05,
+      "loss": 1.543,
+      "step": 237000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.767245984079628e-05,
+      "loss": 1.5799,
+      "step": 237100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.766703145424304e-05,
+      "loss": 1.575,
+      "step": 237200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.766160306768981e-05,
+      "loss": 1.6244,
+      "step": 237300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7656174681136573e-05,
+      "loss": 1.5605,
+      "step": 237400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.765074629458334e-05,
+      "loss": 1.5672,
+      "step": 237500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7645317908030104e-05,
+      "loss": 1.5475,
+      "step": 237600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7639889521476866e-05,
+      "loss": 1.5619,
+      "step": 237700
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7634461134923634e-05,
+      "loss": 1.563,
+      "step": 237800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.76290327483704e-05,
+      "loss": 1.5412,
+      "step": 237900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7623604361817165e-05,
+      "loss": 1.5627,
+      "step": 238000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7618175975263934e-05,
+      "loss": 1.5718,
+      "step": 238100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7612747588710696e-05,
+      "loss": 1.5722,
+      "step": 238200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.760731920215746e-05,
+      "loss": 1.5315,
+      "step": 238300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7601890815604226e-05,
+      "loss": 1.5888,
+      "step": 238400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.759646242905099e-05,
+      "loss": 1.5922,
+      "step": 238500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.759103404249775e-05,
+      "loss": 1.5701,
+      "step": 238600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.758560565594452e-05,
+      "loss": 1.543,
+      "step": 238700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.758017726939129e-05,
+      "loss": 1.5407,
+      "step": 238800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.757474888283805e-05,
+      "loss": 1.5368,
+      "step": 238900
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.756932049628482e-05,
+      "loss": 1.5643,
+      "step": 239000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.756389210973158e-05,
+      "loss": 1.5682,
+      "step": 239100
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.755846372317834e-05,
+      "loss": 1.6079,
+      "step": 239200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.755303533662511e-05,
+      "loss": 1.5494,
+      "step": 239300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.754760695007187e-05,
+      "loss": 1.5696,
+      "step": 239400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.754217856351864e-05,
+      "loss": 1.589,
+      "step": 239500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.75367501769654e-05,
+      "loss": 1.5783,
+      "step": 239600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.7531321790412164e-05,
+      "loss": 1.5656,
+      "step": 239700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.752589340385893e-05,
+      "loss": 1.5555,
+      "step": 239800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.75204650173057e-05,
+      "loss": 1.5458,
+      "step": 239900
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.7515036630752463e-05,
+      "loss": 1.554,
+      "step": 240000
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 1.4660414457321167,
+      "eval_runtime": 17.7415,
+      "eval_samples_per_second": 563.65,
+      "eval_steps_per_second": 17.642,
+      "step": 240000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.750960824419923e-05,
+      "loss": 1.5507,
+      "step": 240100
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.7504179857645994e-05,
+      "loss": 1.576,
+      "step": 240200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.7498751471092756e-05,
+      "loss": 1.5729,
+      "step": 240300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.7493323084539525e-05,
+      "loss": 1.5668,
+      "step": 240400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.7487894697986286e-05,
+      "loss": 1.5422,
+      "step": 240500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.748246631143305e-05,
+      "loss": 1.5584,
+      "step": 240600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.747703792487982e-05,
+      "loss": 1.5586,
+      "step": 240700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.747160953832658e-05,
+      "loss": 1.581,
+      "step": 240800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.746618115177335e-05,
+      "loss": 1.5551,
+      "step": 240900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.7460752765220116e-05,
+      "loss": 1.5722,
+      "step": 241000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.745532437866688e-05,
+      "loss": 1.549,
+      "step": 241100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.744989599211364e-05,
+      "loss": 1.5682,
+      "step": 241200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.744446760556041e-05,
+      "loss": 1.5629,
+      "step": 241300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.743903921900717e-05,
+      "loss": 1.558,
+      "step": 241400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.743361083245394e-05,
+      "loss": 1.5554,
+      "step": 241500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.74281824459007e-05,
+      "loss": 1.5644,
+      "step": 241600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.742275405934746e-05,
+      "loss": 1.5892,
+      "step": 241700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.741732567279423e-05,
+      "loss": 1.5873,
+      "step": 241800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.741189728624099e-05,
+      "loss": 1.5837,
+      "step": 241900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.740646889968776e-05,
+      "loss": 1.5499,
+      "step": 242000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.740104051313453e-05,
+      "loss": 1.5734,
+      "step": 242100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.739561212658129e-05,
+      "loss": 1.5728,
+      "step": 242200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.7390183740028054e-05,
+      "loss": 1.5915,
+      "step": 242300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.738475535347482e-05,
+      "loss": 1.5816,
+      "step": 242400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.7379326966921585e-05,
+      "loss": 1.5696,
+      "step": 242500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.737389858036835e-05,
+      "loss": 1.5796,
+      "step": 242600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.7368470193815115e-05,
+      "loss": 1.5548,
+      "step": 242700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.736304180726188e-05,
+      "loss": 1.5453,
+      "step": 242800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.7357613420708646e-05,
+      "loss": 1.5668,
+      "step": 242900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.7352185034155415e-05,
+      "loss": 1.5536,
+      "step": 243000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.7346756647602176e-05,
+      "loss": 1.5793,
+      "step": 243100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.734132826104894e-05,
+      "loss": 1.571,
+      "step": 243200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.733589987449571e-05,
+      "loss": 1.5384,
+      "step": 243300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.733047148794247e-05,
+      "loss": 1.5789,
+      "step": 243400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.732504310138924e-05,
+      "loss": 1.5537,
+      "step": 243500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.7319614714836e-05,
+      "loss": 1.5811,
+      "step": 243600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.731418632828276e-05,
+      "loss": 1.5629,
+      "step": 243700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.730875794172953e-05,
+      "loss": 1.5489,
+      "step": 243800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.730332955517629e-05,
+      "loss": 1.5897,
+      "step": 243900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.729790116862306e-05,
+      "loss": 1.5541,
+      "step": 244000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.729247278206983e-05,
+      "loss": 1.5482,
+      "step": 244100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.728704439551659e-05,
+      "loss": 1.5748,
+      "step": 244200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.728161600896335e-05,
+      "loss": 1.5724,
+      "step": 244300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.727618762241012e-05,
+      "loss": 1.5728,
+      "step": 244400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.727075923585688e-05,
+      "loss": 1.5803,
+      "step": 244500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.7265330849303645e-05,
+      "loss": 1.5593,
+      "step": 244600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.7259902462750414e-05,
+      "loss": 1.5943,
+      "step": 244700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.7254474076197176e-05,
+      "loss": 1.5571,
+      "step": 244800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.724904568964394e-05,
+      "loss": 1.5679,
+      "step": 244900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.7243617303090706e-05,
+      "loss": 1.5831,
+      "step": 245000
+    },
+    {
+      "epoch": 1.05,
+      "eval_loss": 1.4603722095489502,
+      "eval_runtime": 17.8071,
+      "eval_samples_per_second": 561.575,
+      "eval_steps_per_second": 17.577,
+      "step": 245000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.7238188916537475e-05,
+      "loss": 1.5906,
+      "step": 245100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.723276052998424e-05,
+      "loss": 1.5713,
+      "step": 245200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.7227332143431005e-05,
+      "loss": 1.57,
+      "step": 245300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.722190375687777e-05,
+      "loss": 1.5683,
+      "step": 245400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.7216475370324536e-05,
+      "loss": 1.5457,
+      "step": 245500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.72110469837713e-05,
+      "loss": 1.5651,
+      "step": 245600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.720561859721806e-05,
+      "loss": 1.6014,
+      "step": 245700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.720019021066483e-05,
+      "loss": 1.5894,
+      "step": 245800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.719476182411159e-05,
+      "loss": 1.5541,
+      "step": 245900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.718933343755835e-05,
+      "loss": 1.5763,
+      "step": 246000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.718390505100513e-05,
+      "loss": 1.5676,
+      "step": 246100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.717847666445189e-05,
+      "loss": 1.6003,
+      "step": 246200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.717304827789865e-05,
+      "loss": 1.5466,
+      "step": 246300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.716761989134542e-05,
+      "loss": 1.5364,
+      "step": 246400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.716219150479218e-05,
+      "loss": 1.5713,
+      "step": 246500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.7156763118238944e-05,
+      "loss": 1.5586,
+      "step": 246600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.715133473168571e-05,
+      "loss": 1.5812,
+      "step": 246700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.7145906345132474e-05,
+      "loss": 1.6289,
+      "step": 246800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.7140477958579236e-05,
+      "loss": 1.5595,
+      "step": 246900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.7135049572026005e-05,
+      "loss": 1.5414,
+      "step": 247000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.712962118547277e-05,
+      "loss": 1.6069,
+      "step": 247100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.7124192798919535e-05,
+      "loss": 1.5832,
+      "step": 247200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.7118764412366304e-05,
+      "loss": 1.5645,
+      "step": 247300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.7113336025813066e-05,
+      "loss": 1.5702,
+      "step": 247400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.7107907639259834e-05,
+      "loss": 1.5632,
+      "step": 247500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.7102479252706596e-05,
+      "loss": 1.5915,
+      "step": 247600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.709705086615336e-05,
+      "loss": 1.5472,
+      "step": 247700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.709162247960013e-05,
+      "loss": 1.5432,
+      "step": 247800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.708619409304689e-05,
+      "loss": 1.5848,
+      "step": 247900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.708076570649365e-05,
+      "loss": 1.5771,
+      "step": 248000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.707533731994042e-05,
+      "loss": 1.5733,
+      "step": 248100
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.706990893338719e-05,
+      "loss": 1.564,
+      "step": 248200
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.706448054683395e-05,
+      "loss": 1.5416,
+      "step": 248300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.705905216028072e-05,
+      "loss": 1.5679,
+      "step": 248400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.705362377372748e-05,
+      "loss": 1.5614,
+      "step": 248500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.704819538717424e-05,
+      "loss": 1.5452,
+      "step": 248600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.704276700062101e-05,
+      "loss": 1.551,
+      "step": 248700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.703733861406777e-05,
+      "loss": 1.5544,
+      "step": 248800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.7031910227514535e-05,
+      "loss": 1.5753,
+      "step": 248900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.70264818409613e-05,
+      "loss": 1.5518,
+      "step": 249000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.7021053454408065e-05,
+      "loss": 1.581,
+      "step": 249100
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.7015625067854834e-05,
+      "loss": 1.556,
+      "step": 249200
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.70101966813016e-05,
+      "loss": 1.5366,
+      "step": 249300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.7004768294748364e-05,
+      "loss": 1.5959,
+      "step": 249400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.699933990819513e-05,
+      "loss": 1.5555,
+      "step": 249500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.6993911521641895e-05,
+      "loss": 1.5879,
+      "step": 249600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.698848313508866e-05,
+      "loss": 1.5679,
+      "step": 249700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.6983054748535425e-05,
+      "loss": 1.5627,
+      "step": 249800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.697762636198219e-05,
+      "loss": 1.581,
+      "step": 249900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.697219797542895e-05,
+      "loss": 1.5692,
+      "step": 250000
+    },
+    {
+      "epoch": 1.07,
+      "eval_loss": 1.4626483917236328,
+      "eval_runtime": 17.7967,
+      "eval_samples_per_second": 561.901,
+      "eval_steps_per_second": 17.588,
+      "step": 250000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.696676958887572e-05,
+      "loss": 1.5572,
+      "step": 250100
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.696134120232248e-05,
+      "loss": 1.5778,
+      "step": 250200
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.695591281576925e-05,
+      "loss": 1.5707,
+      "step": 250300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.695048442921602e-05,
+      "loss": 1.5474,
+      "step": 250400
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.694505604266278e-05,
+      "loss": 1.5661,
+      "step": 250500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.693962765610954e-05,
+      "loss": 1.5771,
+      "step": 250600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.693419926955631e-05,
+      "loss": 1.5655,
+      "step": 250700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.692877088300307e-05,
+      "loss": 1.565,
+      "step": 250800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.692334249644983e-05,
+      "loss": 1.553,
+      "step": 250900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.69179141098966e-05,
+      "loss": 1.5797,
+      "step": 251000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.6912485723343364e-05,
+      "loss": 1.5696,
+      "step": 251100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.690705733679013e-05,
+      "loss": 1.5366,
+      "step": 251200
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.6901628950236894e-05,
+      "loss": 1.5782,
+      "step": 251300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.689620056368366e-05,
+      "loss": 1.5443,
+      "step": 251400
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.689077217713043e-05,
+      "loss": 1.5377,
+      "step": 251500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.688534379057719e-05,
+      "loss": 1.5757,
+      "step": 251600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.6879915404023955e-05,
+      "loss": 1.5642,
+      "step": 251700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.6874487017470724e-05,
+      "loss": 1.5692,
+      "step": 251800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.6869058630917486e-05,
+      "loss": 1.5721,
+      "step": 251900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.686363024436425e-05,
+      "loss": 1.5283,
+      "step": 252000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.6858201857811016e-05,
+      "loss": 1.5645,
+      "step": 252100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.685277347125778e-05,
+      "loss": 1.5916,
+      "step": 252200
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.684734508470454e-05,
+      "loss": 1.556,
+      "step": 252300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.6841916698151315e-05,
+      "loss": 1.5695,
+      "step": 252400
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.683648831159808e-05,
+      "loss": 1.5723,
+      "step": 252500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.683105992504484e-05,
+      "loss": 1.576,
+      "step": 252600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.682563153849161e-05,
+      "loss": 1.5544,
+      "step": 252700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.682020315193837e-05,
+      "loss": 1.5653,
+      "step": 252800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.681477476538513e-05,
+      "loss": 1.5586,
+      "step": 252900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.68093463788319e-05,
+      "loss": 1.5375,
+      "step": 253000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.680391799227866e-05,
+      "loss": 1.5459,
+      "step": 253100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.679848960572543e-05,
+      "loss": 1.5852,
+      "step": 253200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.679306121917219e-05,
+      "loss": 1.5484,
+      "step": 253300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.6787632832618955e-05,
+      "loss": 1.58,
+      "step": 253400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.678220444606573e-05,
+      "loss": 1.5391,
+      "step": 253500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.677677605951249e-05,
+      "loss": 1.542,
+      "step": 253600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.6771347672959254e-05,
+      "loss": 1.5494,
+      "step": 253700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.676591928640602e-05,
+      "loss": 1.6048,
+      "step": 253800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.6760490899852784e-05,
+      "loss": 1.5955,
+      "step": 253900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.6755062513299546e-05,
+      "loss": 1.5669,
+      "step": 254000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.6749634126746315e-05,
+      "loss": 1.5435,
+      "step": 254100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.674420574019308e-05,
+      "loss": 1.5766,
+      "step": 254200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.673877735363984e-05,
+      "loss": 1.5561,
+      "step": 254300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.673334896708661e-05,
+      "loss": 1.565,
+      "step": 254400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.6727920580533376e-05,
+      "loss": 1.5592,
+      "step": 254500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.672249219398014e-05,
+      "loss": 1.5598,
+      "step": 254600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.6717063807426906e-05,
+      "loss": 1.5901,
+      "step": 254700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.671163542087367e-05,
+      "loss": 1.5766,
+      "step": 254800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.670620703432043e-05,
+      "loss": 1.5855,
+      "step": 254900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.67007786477672e-05,
+      "loss": 1.558,
+      "step": 255000
+    },
+    {
+      "epoch": 1.1,
+      "eval_loss": 1.464188814163208,
+      "eval_runtime": 17.8297,
+      "eval_samples_per_second": 560.862,
+      "eval_steps_per_second": 17.555,
+      "step": 255000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.669535026121396e-05,
+      "loss": 1.5314,
+      "step": 255100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.668992187466073e-05,
+      "loss": 1.5544,
+      "step": 255200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.668449348810749e-05,
+      "loss": 1.5317,
+      "step": 255300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.667906510155425e-05,
+      "loss": 1.5621,
+      "step": 255400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.667363671500102e-05,
+      "loss": 1.5577,
+      "step": 255500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.666820832844779e-05,
+      "loss": 1.6049,
+      "step": 255600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.666277994189455e-05,
+      "loss": 1.5713,
+      "step": 255700
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.665735155534132e-05,
+      "loss": 1.5628,
+      "step": 255800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.665192316878808e-05,
+      "loss": 1.5665,
+      "step": 255900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.6646494782234845e-05,
+      "loss": 1.5555,
+      "step": 256000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.664106639568161e-05,
+      "loss": 1.5425,
+      "step": 256100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.6635638009128375e-05,
+      "loss": 1.5902,
+      "step": 256200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.663020962257514e-05,
+      "loss": 1.5094,
+      "step": 256300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.6624781236021906e-05,
+      "loss": 1.5773,
+      "step": 256400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.661935284946867e-05,
+      "loss": 1.5582,
+      "step": 256500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.6613924462915436e-05,
+      "loss": 1.5661,
+      "step": 256600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.6608496076362205e-05,
+      "loss": 1.5714,
+      "step": 256700
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.660306768980897e-05,
+      "loss": 1.5805,
+      "step": 256800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.659763930325573e-05,
+      "loss": 1.6005,
+      "step": 256900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.65922109167025e-05,
+      "loss": 1.5594,
+      "step": 257000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.658678253014926e-05,
+      "loss": 1.564,
+      "step": 257100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.658135414359603e-05,
+      "loss": 1.5615,
+      "step": 257200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.657592575704279e-05,
+      "loss": 1.5556,
+      "step": 257300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.657049737048955e-05,
+      "loss": 1.5579,
+      "step": 257400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.656506898393632e-05,
+      "loss": 1.5476,
+      "step": 257500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.655964059738308e-05,
+      "loss": 1.5855,
+      "step": 257600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.655421221082985e-05,
+      "loss": 1.5781,
+      "step": 257700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.654878382427662e-05,
+      "loss": 1.5664,
+      "step": 257800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.654335543772338e-05,
+      "loss": 1.5789,
+      "step": 257900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.653792705117014e-05,
+      "loss": 1.5674,
+      "step": 258000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.653249866461691e-05,
+      "loss": 1.5452,
+      "step": 258100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.6527070278063674e-05,
+      "loss": 1.576,
+      "step": 258200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.6521641891510435e-05,
+      "loss": 1.5243,
+      "step": 258300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.6516213504957204e-05,
+      "loss": 1.5706,
+      "step": 258400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.6510785118403966e-05,
+      "loss": 1.5672,
+      "step": 258500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.6505356731850735e-05,
+      "loss": 1.5574,
+      "step": 258600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.64999283452975e-05,
+      "loss": 1.5636,
+      "step": 258700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.6494499958744265e-05,
+      "loss": 1.5683,
+      "step": 258800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.648907157219103e-05,
+      "loss": 1.6023,
+      "step": 258900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.6483643185637796e-05,
+      "loss": 1.6015,
+      "step": 259000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.647821479908456e-05,
+      "loss": 1.5853,
+      "step": 259100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.6472786412531326e-05,
+      "loss": 1.5587,
+      "step": 259200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.646735802597809e-05,
+      "loss": 1.5462,
+      "step": 259300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.646192963942485e-05,
+      "loss": 1.5808,
+      "step": 259400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.645650125287162e-05,
+      "loss": 1.5588,
+      "step": 259500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.645107286631838e-05,
+      "loss": 1.5724,
+      "step": 259600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.644564447976514e-05,
+      "loss": 1.5439,
+      "step": 259700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.644021609321192e-05,
+      "loss": 1.5489,
+      "step": 259800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.643478770665868e-05,
+      "loss": 1.5738,
+      "step": 259900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.642935932010544e-05,
+      "loss": 1.5925,
+      "step": 260000
+    },
+    {
+      "epoch": 1.12,
+      "eval_loss": 1.4609049558639526,
+      "eval_runtime": 17.8329,
+      "eval_samples_per_second": 560.761,
+      "eval_steps_per_second": 17.552,
+      "step": 260000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.642393093355221e-05,
+      "loss": 1.531,
+      "step": 260100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.641850254699897e-05,
+      "loss": 1.5514,
+      "step": 260200
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.6413074160445734e-05,
+      "loss": 1.5872,
+      "step": 260300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.64076457738925e-05,
+      "loss": 1.5746,
+      "step": 260400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.6402217387339265e-05,
+      "loss": 1.5698,
+      "step": 260500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.639678900078603e-05,
+      "loss": 1.5619,
+      "step": 260600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.6391360614232795e-05,
+      "loss": 1.5511,
+      "step": 260700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.6385932227679564e-05,
+      "loss": 1.562,
+      "step": 260800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.6380503841126326e-05,
+      "loss": 1.5527,
+      "step": 260900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.6375075454573094e-05,
+      "loss": 1.5525,
+      "step": 261000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.6369647068019856e-05,
+      "loss": 1.572,
+      "step": 261100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.6364218681466625e-05,
+      "loss": 1.6127,
+      "step": 261200
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.635879029491339e-05,
+      "loss": 1.5688,
+      "step": 261300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.635336190836015e-05,
+      "loss": 1.5462,
+      "step": 261400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.634793352180692e-05,
+      "loss": 1.5759,
+      "step": 261500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.634250513525368e-05,
+      "loss": 1.5499,
+      "step": 261600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.633707674870044e-05,
+      "loss": 1.5334,
+      "step": 261700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.633164836214721e-05,
+      "loss": 1.5306,
+      "step": 261800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.632621997559398e-05,
+      "loss": 1.5375,
+      "step": 261900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.632079158904074e-05,
+      "loss": 1.5449,
+      "step": 262000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.631536320248751e-05,
+      "loss": 1.5735,
+      "step": 262100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.630993481593427e-05,
+      "loss": 1.5231,
+      "step": 262200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.630450642938103e-05,
+      "loss": 1.5716,
+      "step": 262300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.62990780428278e-05,
+      "loss": 1.5735,
+      "step": 262400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.629364965627456e-05,
+      "loss": 1.5733,
+      "step": 262500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.628822126972133e-05,
+      "loss": 1.5416,
+      "step": 262600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.6282792883168094e-05,
+      "loss": 1.5918,
+      "step": 262700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.6277364496614855e-05,
+      "loss": 1.5342,
+      "step": 262800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.6271936110061624e-05,
+      "loss": 1.5777,
+      "step": 262900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.626650772350839e-05,
+      "loss": 1.5328,
+      "step": 263000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.6261079336955155e-05,
+      "loss": 1.5647,
+      "step": 263100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.625565095040192e-05,
+      "loss": 1.5391,
+      "step": 263200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.6250222563848685e-05,
+      "loss": 1.5861,
+      "step": 263300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.624479417729545e-05,
+      "loss": 1.5629,
+      "step": 263400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.6239365790742216e-05,
+      "loss": 1.5676,
+      "step": 263500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.623393740418898e-05,
+      "loss": 1.5394,
+      "step": 263600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.622850901763574e-05,
+      "loss": 1.5281,
+      "step": 263700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.622308063108251e-05,
+      "loss": 1.548,
+      "step": 263800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.621765224452928e-05,
+      "loss": 1.5544,
+      "step": 263900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.621222385797604e-05,
+      "loss": 1.5579,
+      "step": 264000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.620679547142281e-05,
+      "loss": 1.5662,
+      "step": 264100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.620136708486957e-05,
+      "loss": 1.5606,
+      "step": 264200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.619593869831633e-05,
+      "loss": 1.5445,
+      "step": 264300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.61905103117631e-05,
+      "loss": 1.5551,
+      "step": 264400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.618508192520986e-05,
+      "loss": 1.5423,
+      "step": 264500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.617965353865663e-05,
+      "loss": 1.5331,
+      "step": 264600
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.617422515210339e-05,
+      "loss": 1.5552,
+      "step": 264700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.6168796765550154e-05,
+      "loss": 1.5557,
+      "step": 264800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.616336837899692e-05,
+      "loss": 1.5547,
+      "step": 264900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.615793999244369e-05,
+      "loss": 1.5396,
+      "step": 265000
+    },
+    {
+      "epoch": 1.14,
+      "eval_loss": 1.4619945287704468,
+      "eval_runtime": 17.8247,
+      "eval_samples_per_second": 561.019,
+      "eval_steps_per_second": 17.56,
+      "step": 265000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.615251160589045e-05,
+      "loss": 1.5693,
+      "step": 265100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.614708321933722e-05,
+      "loss": 1.6009,
+      "step": 265200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.6141654832783984e-05,
+      "loss": 1.5776,
+      "step": 265300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.6136226446230745e-05,
+      "loss": 1.5691,
+      "step": 265400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.6130798059677514e-05,
+      "loss": 1.5786,
+      "step": 265500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.6125369673124276e-05,
+      "loss": 1.568,
+      "step": 265600
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.611994128657104e-05,
+      "loss": 1.5471,
+      "step": 265700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.6114512900017807e-05,
+      "loss": 1.5895,
+      "step": 265800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.610908451346457e-05,
+      "loss": 1.5691,
+      "step": 265900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.610365612691134e-05,
+      "loss": 1.5361,
+      "step": 266000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.6098227740358106e-05,
+      "loss": 1.5495,
+      "step": 266100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.609279935380487e-05,
+      "loss": 1.5553,
+      "step": 266200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.608737096725163e-05,
+      "loss": 1.5584,
+      "step": 266300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.60819425806984e-05,
+      "loss": 1.528,
+      "step": 266400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.607651419414516e-05,
+      "loss": 1.5457,
+      "step": 266500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.607108580759193e-05,
+      "loss": 1.5393,
+      "step": 266600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.606565742103869e-05,
+      "loss": 1.5813,
+      "step": 266700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.606022903448545e-05,
+      "loss": 1.5665,
+      "step": 266800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.605480064793222e-05,
+      "loss": 1.5375,
+      "step": 266900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.604937226137898e-05,
+      "loss": 1.5525,
+      "step": 267000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.604394387482575e-05,
+      "loss": 1.5531,
+      "step": 267100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.603851548827252e-05,
+      "loss": 1.5469,
+      "step": 267200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.603308710171928e-05,
+      "loss": 1.587,
+      "step": 267300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.6027658715166044e-05,
+      "loss": 1.5445,
+      "step": 267400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.602223032861281e-05,
+      "loss": 1.5907,
+      "step": 267500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.6016801942059574e-05,
+      "loss": 1.5455,
+      "step": 267600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.6011373555506336e-05,
+      "loss": 1.5386,
+      "step": 267700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.6005945168953105e-05,
+      "loss": 1.5539,
+      "step": 267800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.600051678239987e-05,
+      "loss": 1.5513,
+      "step": 267900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.5995088395846636e-05,
+      "loss": 1.5439,
+      "step": 268000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.5989660009293404e-05,
+      "loss": 1.5362,
+      "step": 268100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.5984231622740166e-05,
+      "loss": 1.554,
+      "step": 268200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.597880323618693e-05,
+      "loss": 1.5568,
+      "step": 268300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.5973374849633697e-05,
+      "loss": 1.5624,
+      "step": 268400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.596794646308046e-05,
+      "loss": 1.5489,
+      "step": 268500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.596251807652723e-05,
+      "loss": 1.5618,
+      "step": 268600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.595708968997399e-05,
+      "loss": 1.5626,
+      "step": 268700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.595166130342075e-05,
+      "loss": 1.5691,
+      "step": 268800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.594623291686752e-05,
+      "loss": 1.5809,
+      "step": 268900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.594080453031428e-05,
+      "loss": 1.5376,
+      "step": 269000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.593537614376104e-05,
+      "loss": 1.5724,
+      "step": 269100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.592994775720782e-05,
+      "loss": 1.5474,
+      "step": 269200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.592451937065458e-05,
+      "loss": 1.5703,
+      "step": 269300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.591909098410134e-05,
+      "loss": 1.5577,
+      "step": 269400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.591366259754811e-05,
+      "loss": 1.5545,
+      "step": 269500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.590823421099487e-05,
+      "loss": 1.5388,
+      "step": 269600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5902805824441635e-05,
+      "loss": 1.564,
+      "step": 269700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5897377437888403e-05,
+      "loss": 1.5424,
+      "step": 269800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5891949051335165e-05,
+      "loss": 1.5829,
+      "step": 269900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5886520664781934e-05,
+      "loss": 1.59,
+      "step": 270000
+    },
+    {
+      "epoch": 1.16,
+      "eval_loss": 1.4590508937835693,
+      "eval_runtime": 17.8197,
+      "eval_samples_per_second": 561.177,
+      "eval_steps_per_second": 17.565,
+      "step": 270000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5881092278228696e-05,
+      "loss": 1.5582,
+      "step": 270100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5875663891675465e-05,
+      "loss": 1.5847,
+      "step": 270200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5870235505122226e-05,
+      "loss": 1.5796,
+      "step": 270300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5864807118568995e-05,
+      "loss": 1.5329,
+      "step": 270400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.585937873201576e-05,
+      "loss": 1.5347,
+      "step": 270500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5853950345462526e-05,
+      "loss": 1.535,
+      "step": 270600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.584852195890929e-05,
+      "loss": 1.5335,
+      "step": 270700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.584309357235605e-05,
+      "loss": 1.5717,
+      "step": 270800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.583766518580282e-05,
+      "loss": 1.5496,
+      "step": 270900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.583223679924958e-05,
+      "loss": 1.5529,
+      "step": 271000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.582680841269634e-05,
+      "loss": 1.5607,
+      "step": 271100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.582138002614311e-05,
+      "loss": 1.5481,
+      "step": 271200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.581595163958988e-05,
+      "loss": 1.5442,
+      "step": 271300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.581052325303664e-05,
+      "loss": 1.5719,
+      "step": 271400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.580509486648341e-05,
+      "loss": 1.5845,
+      "step": 271500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.579966647993017e-05,
+      "loss": 1.5695,
+      "step": 271600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.579423809337693e-05,
+      "loss": 1.5496,
+      "step": 271700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.57888097068237e-05,
+      "loss": 1.5713,
+      "step": 271800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.5783381320270464e-05,
+      "loss": 1.5643,
+      "step": 271900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.577795293371723e-05,
+      "loss": 1.5655,
+      "step": 272000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.5772524547163994e-05,
+      "loss": 1.5866,
+      "step": 272100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.5767096160610756e-05,
+      "loss": 1.5959,
+      "step": 272200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.5761667774057525e-05,
+      "loss": 1.5499,
+      "step": 272300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.5756239387504294e-05,
+      "loss": 1.5568,
+      "step": 272400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.5750811000951055e-05,
+      "loss": 1.5488,
+      "step": 272500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.5745382614397824e-05,
+      "loss": 1.557,
+      "step": 272600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.5739954227844586e-05,
+      "loss": 1.5672,
+      "step": 272700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.573452584129135e-05,
+      "loss": 1.5689,
+      "step": 272800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.5729097454738116e-05,
+      "loss": 1.549,
+      "step": 272900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.572366906818488e-05,
+      "loss": 1.5771,
+      "step": 273000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.571824068163164e-05,
+      "loss": 1.5602,
+      "step": 273100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.571281229507841e-05,
+      "loss": 1.5542,
+      "step": 273200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.570738390852517e-05,
+      "loss": 1.5522,
+      "step": 273300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.570195552197194e-05,
+      "loss": 1.5752,
+      "step": 273400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.569652713541871e-05,
+      "loss": 1.5649,
+      "step": 273500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.569109874886547e-05,
+      "loss": 1.531,
+      "step": 273600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.568567036231223e-05,
+      "loss": 1.5486,
+      "step": 273700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.5680241975759e-05,
+      "loss": 1.5541,
+      "step": 273800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.567481358920576e-05,
+      "loss": 1.5403,
+      "step": 273900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.566938520265253e-05,
+      "loss": 1.556,
+      "step": 274000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.566395681609929e-05,
+      "loss": 1.573,
+      "step": 274100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.5658528429546055e-05,
+      "loss": 1.5342,
+      "step": 274200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.5653100042992823e-05,
+      "loss": 1.5544,
+      "step": 274300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.564767165643959e-05,
+      "loss": 1.5691,
+      "step": 274400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.5642243269886354e-05,
+      "loss": 1.5439,
+      "step": 274500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.563681488333312e-05,
+      "loss": 1.5634,
+      "step": 274600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.5631386496779884e-05,
+      "loss": 1.5517,
+      "step": 274700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.5625958110226646e-05,
+      "loss": 1.5376,
+      "step": 274800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.5620529723673415e-05,
+      "loss": 1.6075,
+      "step": 274900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.561510133712018e-05,
+      "loss": 1.5671,
+      "step": 275000
+    },
+    {
+      "epoch": 1.18,
+      "eval_loss": 1.4539235830307007,
+      "eval_runtime": 17.8159,
+      "eval_samples_per_second": 561.296,
+      "eval_steps_per_second": 17.569,
+      "step": 275000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.560967295056694e-05,
+      "loss": 1.5554,
+      "step": 275100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.560424456401371e-05,
+      "loss": 1.5644,
+      "step": 275200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.559881617746047e-05,
+      "loss": 1.5373,
+      "step": 275300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.559338779090723e-05,
+      "loss": 1.5611,
+      "step": 275400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.5587959404354007e-05,
+      "loss": 1.5701,
+      "step": 275500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.558253101780077e-05,
+      "loss": 1.5404,
+      "step": 275600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.557710263124753e-05,
+      "loss": 1.5647,
+      "step": 275700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.55716742446943e-05,
+      "loss": 1.5538,
+      "step": 275800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.556624585814106e-05,
+      "loss": 1.5497,
+      "step": 275900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.556081747158783e-05,
+      "loss": 1.5548,
+      "step": 276000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.555538908503459e-05,
+      "loss": 1.566,
+      "step": 276100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.554996069848135e-05,
+      "loss": 1.5541,
+      "step": 276200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.554453231192812e-05,
+      "loss": 1.5611,
+      "step": 276300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.5539103925374884e-05,
+      "loss": 1.5562,
+      "step": 276400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.553367553882165e-05,
+      "loss": 1.562,
+      "step": 276500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.552824715226842e-05,
+      "loss": 1.5387,
+      "step": 276600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.552281876571518e-05,
+      "loss": 1.5448,
+      "step": 276700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.5517390379161945e-05,
+      "loss": 1.5676,
+      "step": 276800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.5511961992608713e-05,
+      "loss": 1.5381,
+      "step": 276900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.5506533606055475e-05,
+      "loss": 1.5715,
+      "step": 277000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.550110521950224e-05,
+      "loss": 1.5469,
+      "step": 277100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.5495676832949006e-05,
+      "loss": 1.5773,
+      "step": 277200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.549024844639577e-05,
+      "loss": 1.5712,
+      "step": 277300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.548482005984253e-05,
+      "loss": 1.5421,
+      "step": 277400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.54793916732893e-05,
+      "loss": 1.5442,
+      "step": 277500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.547396328673607e-05,
+      "loss": 1.5309,
+      "step": 277600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.546853490018283e-05,
+      "loss": 1.5588,
+      "step": 277700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.54631065136296e-05,
+      "loss": 1.5598,
+      "step": 277800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.545767812707636e-05,
+      "loss": 1.5618,
+      "step": 277900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.545224974052313e-05,
+      "loss": 1.5579,
+      "step": 278000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.544682135396989e-05,
+      "loss": 1.5515,
+      "step": 278100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.544139296741665e-05,
+      "loss": 1.5779,
+      "step": 278200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.543596458086342e-05,
+      "loss": 1.5852,
+      "step": 278300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.543053619431018e-05,
+      "loss": 1.5487,
+      "step": 278400
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.5425107807756944e-05,
+      "loss": 1.5745,
+      "step": 278500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.541967942120372e-05,
+      "loss": 1.5421,
+      "step": 278600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.541425103465048e-05,
+      "loss": 1.5352,
+      "step": 278700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.540882264809724e-05,
+      "loss": 1.556,
+      "step": 278800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.540339426154401e-05,
+      "loss": 1.5488,
+      "step": 278900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.5397965874990774e-05,
+      "loss": 1.5424,
+      "step": 279000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.5392537488437536e-05,
+      "loss": 1.5643,
+      "step": 279100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.5387109101884304e-05,
+      "loss": 1.5561,
+      "step": 279200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.5381680715331066e-05,
+      "loss": 1.5653,
+      "step": 279300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.537625232877783e-05,
+      "loss": 1.5528,
+      "step": 279400
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.53708239422246e-05,
+      "loss": 1.566,
+      "step": 279500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.536539555567136e-05,
+      "loss": 1.5282,
+      "step": 279600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.535996716911813e-05,
+      "loss": 1.5516,
+      "step": 279700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.5354538782564896e-05,
+      "loss": 1.531,
+      "step": 279800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.534911039601166e-05,
+      "loss": 1.5397,
+      "step": 279900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.5343682009458426e-05,
+      "loss": 1.5497,
+      "step": 280000
+    },
+    {
+      "epoch": 1.2,
+      "eval_loss": 1.456358551979065,
+      "eval_runtime": 17.8134,
+      "eval_samples_per_second": 561.375,
+      "eval_steps_per_second": 17.571,
+      "step": 280000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.533825362290519e-05,
+      "loss": 1.5704,
+      "step": 280100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.533282523635195e-05,
+      "loss": 1.5544,
+      "step": 280200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.532739684979872e-05,
+      "loss": 1.5539,
+      "step": 280300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.532196846324548e-05,
+      "loss": 1.5754,
+      "step": 280400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.531654007669224e-05,
+      "loss": 1.5399,
+      "step": 280500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.531111169013901e-05,
+      "loss": 1.5785,
+      "step": 280600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.530568330358578e-05,
+      "loss": 1.5415,
+      "step": 280700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.530025491703254e-05,
+      "loss": 1.5825,
+      "step": 280800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.529482653047931e-05,
+      "loss": 1.5565,
+      "step": 280900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.528939814392607e-05,
+      "loss": 1.567,
+      "step": 281000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.5283969757372834e-05,
+      "loss": 1.5726,
+      "step": 281100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.52785413708196e-05,
+      "loss": 1.5311,
+      "step": 281200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.5273112984266365e-05,
+      "loss": 1.5515,
+      "step": 281300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.5267684597713127e-05,
+      "loss": 1.5515,
+      "step": 281400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.5262256211159895e-05,
+      "loss": 1.5597,
+      "step": 281500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.525682782460666e-05,
+      "loss": 1.5452,
+      "step": 281600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.5251399438053426e-05,
+      "loss": 1.5714,
+      "step": 281700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.5245971051500194e-05,
+      "loss": 1.5578,
+      "step": 281800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.5240542664946956e-05,
+      "loss": 1.5484,
+      "step": 281900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.5235114278393725e-05,
+      "loss": 1.5792,
+      "step": 282000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.522968589184049e-05,
+      "loss": 1.5652,
+      "step": 282100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.522425750528725e-05,
+      "loss": 1.5616,
+      "step": 282200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.521882911873402e-05,
+      "loss": 1.561,
+      "step": 282300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.521340073218078e-05,
+      "loss": 1.5664,
+      "step": 282400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.520797234562754e-05,
+      "loss": 1.5775,
+      "step": 282500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.520254395907431e-05,
+      "loss": 1.5332,
+      "step": 282600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.519711557252107e-05,
+      "loss": 1.5193,
+      "step": 282700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.519168718596784e-05,
+      "loss": 1.5546,
+      "step": 282800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.518625879941461e-05,
+      "loss": 1.5636,
+      "step": 282900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.518083041286137e-05,
+      "loss": 1.5514,
+      "step": 283000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.517540202630813e-05,
+      "loss": 1.5575,
+      "step": 283100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.51699736397549e-05,
+      "loss": 1.5341,
+      "step": 283200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.516454525320166e-05,
+      "loss": 1.5468,
+      "step": 283300
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.5159116866648425e-05,
+      "loss": 1.5506,
+      "step": 283400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.5153688480095194e-05,
+      "loss": 1.5696,
+      "step": 283500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.5148260093541956e-05,
+      "loss": 1.5704,
+      "step": 283600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.5142831706988724e-05,
+      "loss": 1.5557,
+      "step": 283700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.513740332043549e-05,
+      "loss": 1.5561,
+      "step": 283800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.5131974933882255e-05,
+      "loss": 1.5315,
+      "step": 283900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.5126546547329023e-05,
+      "loss": 1.5445,
+      "step": 284000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.5121118160775785e-05,
+      "loss": 1.5352,
+      "step": 284100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.511568977422255e-05,
+      "loss": 1.5458,
+      "step": 284200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.5110261387669316e-05,
+      "loss": 1.5889,
+      "step": 284300
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.510483300111608e-05,
+      "loss": 1.5682,
+      "step": 284400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.509940461456284e-05,
+      "loss": 1.5332,
+      "step": 284500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.509397622800961e-05,
+      "loss": 1.5871,
+      "step": 284600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.508854784145637e-05,
+      "loss": 1.5586,
+      "step": 284700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.508311945490313e-05,
+      "loss": 1.5607,
+      "step": 284800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.507769106834991e-05,
+      "loss": 1.576,
+      "step": 284900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.507226268179667e-05,
+      "loss": 1.5491,
+      "step": 285000
+    },
+    {
+      "epoch": 1.22,
+      "eval_loss": 1.4562804698944092,
+      "eval_runtime": 17.8003,
+      "eval_samples_per_second": 561.788,
+      "eval_steps_per_second": 17.584,
+      "step": 285000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.506683429524343e-05,
+      "loss": 1.5509,
+      "step": 285100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.50614059086902e-05,
+      "loss": 1.5571,
+      "step": 285200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.505597752213696e-05,
+      "loss": 1.541,
+      "step": 285300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.5050549135583724e-05,
+      "loss": 1.5445,
+      "step": 285400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.504512074903049e-05,
+      "loss": 1.5346,
+      "step": 285500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.5039692362477254e-05,
+      "loss": 1.5462,
+      "step": 285600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.503426397592402e-05,
+      "loss": 1.5539,
+      "step": 285700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.5028835589370785e-05,
+      "loss": 1.5642,
+      "step": 285800
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.502340720281755e-05,
+      "loss": 1.56,
+      "step": 285900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.5017978816264315e-05,
+      "loss": 1.5673,
+      "step": 286000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.5012550429711084e-05,
+      "loss": 1.5509,
+      "step": 286100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.5007122043157846e-05,
+      "loss": 1.5552,
+      "step": 286200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.5001693656604614e-05,
+      "loss": 1.5524,
+      "step": 286300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.4996265270051376e-05,
+      "loss": 1.5624,
+      "step": 286400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.499083688349814e-05,
+      "loss": 1.5801,
+      "step": 286500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.498540849694491e-05,
+      "loss": 1.5274,
+      "step": 286600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.497998011039167e-05,
+      "loss": 1.5369,
+      "step": 286700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.497455172383843e-05,
+      "loss": 1.5509,
+      "step": 286800
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.49691233372852e-05,
+      "loss": 1.5426,
+      "step": 286900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.496369495073197e-05,
+      "loss": 1.5654,
+      "step": 287000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.495826656417873e-05,
+      "loss": 1.55,
+      "step": 287100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.49528381776255e-05,
+      "loss": 1.5626,
+      "step": 287200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.494740979107226e-05,
+      "loss": 1.5696,
+      "step": 287300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.494198140451902e-05,
+      "loss": 1.5594,
+      "step": 287400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.493655301796579e-05,
+      "loss": 1.5418,
+      "step": 287500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.493112463141255e-05,
+      "loss": 1.5708,
+      "step": 287600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.492569624485932e-05,
+      "loss": 1.57,
+      "step": 287700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.492026785830608e-05,
+      "loss": 1.5632,
+      "step": 287800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.4914839471752845e-05,
+      "loss": 1.5588,
+      "step": 287900
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.4909411085199614e-05,
+      "loss": 1.5361,
+      "step": 288000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.490398269864638e-05,
+      "loss": 1.5445,
+      "step": 288100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.4898554312093144e-05,
+      "loss": 1.559,
+      "step": 288200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.489312592553991e-05,
+      "loss": 1.5863,
+      "step": 288300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.4887697538986675e-05,
+      "loss": 1.5221,
+      "step": 288400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.4882269152433437e-05,
+      "loss": 1.5596,
+      "step": 288500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.4876840765880205e-05,
+      "loss": 1.5405,
+      "step": 288600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.487141237932697e-05,
+      "loss": 1.5409,
+      "step": 288700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.486598399277373e-05,
+      "loss": 1.5524,
+      "step": 288800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.48605556062205e-05,
+      "loss": 1.5533,
+      "step": 288900
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.485512721966726e-05,
+      "loss": 1.5379,
+      "step": 289000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.484969883311403e-05,
+      "loss": 1.5547,
+      "step": 289100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.48442704465608e-05,
+      "loss": 1.5598,
+      "step": 289200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.483884206000756e-05,
+      "loss": 1.5472,
+      "step": 289300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.483341367345432e-05,
+      "loss": 1.5289,
+      "step": 289400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.482798528690109e-05,
+      "loss": 1.543,
+      "step": 289500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.482255690034785e-05,
+      "loss": 1.5766,
+      "step": 289600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.481712851379462e-05,
+      "loss": 1.5635,
+      "step": 289700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.481170012724138e-05,
+      "loss": 1.5564,
+      "step": 289800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.4806271740688143e-05,
+      "loss": 1.5679,
+      "step": 289900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.480084335413491e-05,
+      "loss": 1.5554,
+      "step": 290000
+    },
+    {
+      "epoch": 1.25,
+      "eval_loss": 1.4566130638122559,
+      "eval_runtime": 17.7457,
+      "eval_samples_per_second": 563.517,
+      "eval_steps_per_second": 17.638,
+      "step": 290000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.479541496758168e-05,
+      "loss": 1.5559,
+      "step": 290100
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.478998658102844e-05,
+      "loss": 1.5364,
+      "step": 290200
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.478455819447521e-05,
+      "loss": 1.5495,
+      "step": 290300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.477912980792197e-05,
+      "loss": 1.5425,
+      "step": 290400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.4773701421368735e-05,
+      "loss": 1.5416,
+      "step": 290500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.4768273034815504e-05,
+      "loss": 1.5573,
+      "step": 290600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.4762844648262266e-05,
+      "loss": 1.5505,
+      "step": 290700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.475741626170903e-05,
+      "loss": 1.5827,
+      "step": 290800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.4751987875155796e-05,
+      "loss": 1.5488,
+      "step": 290900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.474655948860256e-05,
+      "loss": 1.5206,
+      "step": 291000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.474113110204933e-05,
+      "loss": 1.562,
+      "step": 291100
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.4735702715496095e-05,
+      "loss": 1.552,
+      "step": 291200
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.473027432894286e-05,
+      "loss": 1.5442,
+      "step": 291300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.472484594238962e-05,
+      "loss": 1.5581,
+      "step": 291400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.471941755583639e-05,
+      "loss": 1.5588,
+      "step": 291500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.471398916928315e-05,
+      "loss": 1.5738,
+      "step": 291600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.470856078272992e-05,
+      "loss": 1.5371,
+      "step": 291700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.470313239617668e-05,
+      "loss": 1.5712,
+      "step": 291800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.469770400962344e-05,
+      "loss": 1.5889,
+      "step": 291900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.469227562307021e-05,
+      "loss": 1.5397,
+      "step": 292000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.468684723651697e-05,
+      "loss": 1.5654,
+      "step": 292100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.468141884996374e-05,
+      "loss": 1.5338,
+      "step": 292200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.467599046341051e-05,
+      "loss": 1.5349,
+      "step": 292300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.467056207685727e-05,
+      "loss": 1.553,
+      "step": 292400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.4665133690304034e-05,
+      "loss": 1.5506,
+      "step": 292500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.46597053037508e-05,
+      "loss": 1.5682,
+      "step": 292600
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.4654276917197564e-05,
+      "loss": 1.5506,
+      "step": 292700
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.4648848530644326e-05,
+      "loss": 1.549,
+      "step": 292800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.4643420144091095e-05,
+      "loss": 1.5393,
+      "step": 292900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.4637991757537856e-05,
+      "loss": 1.5481,
+      "step": 293000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.4632563370984625e-05,
+      "loss": 1.5265,
+      "step": 293100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.462713498443139e-05,
+      "loss": 1.5614,
+      "step": 293200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.4621706597878156e-05,
+      "loss": 1.5649,
+      "step": 293300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.461627821132492e-05,
+      "loss": 1.5426,
+      "step": 293400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.4610849824771686e-05,
+      "loss": 1.5627,
+      "step": 293500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.460542143821845e-05,
+      "loss": 1.5791,
+      "step": 293600
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.459999305166522e-05,
+      "loss": 1.5435,
+      "step": 293700
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.459456466511198e-05,
+      "loss": 1.5421,
+      "step": 293800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.458913627855874e-05,
+      "loss": 1.5323,
+      "step": 293900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.458370789200551e-05,
+      "loss": 1.5384,
+      "step": 294000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.457827950545227e-05,
+      "loss": 1.5803,
+      "step": 294100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.457285111889903e-05,
+      "loss": 1.5698,
+      "step": 294200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.456742273234581e-05,
+      "loss": 1.5649,
+      "step": 294300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.456199434579257e-05,
+      "loss": 1.587,
+      "step": 294400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.455656595923933e-05,
+      "loss": 1.5739,
+      "step": 294500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.45511375726861e-05,
+      "loss": 1.5239,
+      "step": 294600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.454570918613286e-05,
+      "loss": 1.5361,
+      "step": 294700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.4540280799579624e-05,
+      "loss": 1.5392,
+      "step": 294800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.453485241302639e-05,
+      "loss": 1.541,
+      "step": 294900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.4529424026473155e-05,
+      "loss": 1.5379,
+      "step": 295000
+    },
+    {
+      "epoch": 1.27,
+      "eval_loss": 1.4548817873001099,
+      "eval_runtime": 17.8195,
+      "eval_samples_per_second": 561.182,
+      "eval_steps_per_second": 17.565,
+      "step": 295000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.4523995639919924e-05,
+      "loss": 1.5534,
+      "step": 295100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.4518567253366685e-05,
+      "loss": 1.5415,
+      "step": 295200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.451313886681345e-05,
+      "loss": 1.5556,
+      "step": 295300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.4507710480260216e-05,
+      "loss": 1.5313,
+      "step": 295400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.4502282093706985e-05,
+      "loss": 1.5754,
+      "step": 295500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.4496853707153747e-05,
+      "loss": 1.5621,
+      "step": 295600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.4491425320600515e-05,
+      "loss": 1.5538,
+      "step": 295700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.448599693404728e-05,
+      "loss": 1.5616,
+      "step": 295800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.448056854749404e-05,
+      "loss": 1.554,
+      "step": 295900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.447514016094081e-05,
+      "loss": 1.5214,
+      "step": 296000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.446971177438757e-05,
+      "loss": 1.5608,
+      "step": 296100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.446428338783433e-05,
+      "loss": 1.5388,
+      "step": 296200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.44588550012811e-05,
+      "loss": 1.5748,
+      "step": 296300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.445342661472787e-05,
+      "loss": 1.5302,
+      "step": 296400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.444799822817463e-05,
+      "loss": 1.57,
+      "step": 296500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.44425698416214e-05,
+      "loss": 1.5821,
+      "step": 296600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.443714145506816e-05,
+      "loss": 1.5377,
+      "step": 296700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.443171306851492e-05,
+      "loss": 1.5542,
+      "step": 296800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.442628468196169e-05,
+      "loss": 1.574,
+      "step": 296900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.4420856295408453e-05,
+      "loss": 1.5629,
+      "step": 297000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.441542790885522e-05,
+      "loss": 1.4973,
+      "step": 297100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.4409999522301984e-05,
+      "loss": 1.5659,
+      "step": 297200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.4404571135748746e-05,
+      "loss": 1.5541,
+      "step": 297300
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.4399142749195515e-05,
+      "loss": 1.553,
+      "step": 297400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.439371436264228e-05,
+      "loss": 1.5844,
+      "step": 297500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.4388285976089045e-05,
+      "loss": 1.5459,
+      "step": 297600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.4382857589535814e-05,
+      "loss": 1.5314,
+      "step": 297700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.4377429202982576e-05,
+      "loss": 1.5155,
+      "step": 297800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.437200081642934e-05,
+      "loss": 1.5474,
+      "step": 297900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.4366572429876106e-05,
+      "loss": 1.539,
+      "step": 298000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.436114404332287e-05,
+      "loss": 1.5794,
+      "step": 298100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.435571565676963e-05,
+      "loss": 1.5276,
+      "step": 298200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.43502872702164e-05,
+      "loss": 1.5574,
+      "step": 298300
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.434485888366316e-05,
+      "loss": 1.5615,
+      "step": 298400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.433943049710993e-05,
+      "loss": 1.5511,
+      "step": 298500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.43340021105567e-05,
+      "loss": 1.5437,
+      "step": 298600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.432857372400346e-05,
+      "loss": 1.5773,
+      "step": 298700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.432314533745022e-05,
+      "loss": 1.5682,
+      "step": 298800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.431771695089699e-05,
+      "loss": 1.5283,
+      "step": 298900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.431228856434375e-05,
+      "loss": 1.5506,
+      "step": 299000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.430686017779052e-05,
+      "loss": 1.5324,
+      "step": 299100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.430143179123728e-05,
+      "loss": 1.5774,
+      "step": 299200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.4296003404684044e-05,
+      "loss": 1.5498,
+      "step": 299300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.429057501813081e-05,
+      "loss": 1.5267,
+      "step": 299400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.4285146631577575e-05,
+      "loss": 1.5507,
+      "step": 299500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.4279718245024344e-05,
+      "loss": 1.5379,
+      "step": 299600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.427428985847111e-05,
+      "loss": 1.5369,
+      "step": 299700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.4268861471917874e-05,
+      "loss": 1.552,
+      "step": 299800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.4263433085364636e-05,
+      "loss": 1.5454,
+      "step": 299900
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.4258004698811405e-05,
+      "loss": 1.5883,
+      "step": 300000
+    },
+    {
+      "epoch": 1.29,
+      "eval_loss": 1.4549944400787354,
+      "eval_runtime": 17.7682,
+      "eval_samples_per_second": 562.803,
+      "eval_steps_per_second": 17.616,
+      "step": 300000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.4252576312258166e-05,
+      "loss": 1.5578,
+      "step": 300100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.424714792570493e-05,
+      "loss": 1.5529,
+      "step": 300200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.42417195391517e-05,
+      "loss": 1.555,
+      "step": 300300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.423629115259846e-05,
+      "loss": 1.5347,
+      "step": 300400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.423086276604523e-05,
+      "loss": 1.5489,
+      "step": 300500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.4225434379491996e-05,
+      "loss": 1.5477,
+      "step": 300600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.422000599293876e-05,
+      "loss": 1.576,
+      "step": 300700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.421457760638552e-05,
+      "loss": 1.5734,
+      "step": 300800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.420914921983229e-05,
+      "loss": 1.5729,
+      "step": 300900
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.420372083327905e-05,
+      "loss": 1.5648,
+      "step": 301000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.419829244672582e-05,
+      "loss": 1.5804,
+      "step": 301100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.419286406017258e-05,
+      "loss": 1.554,
+      "step": 301200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.418743567361934e-05,
+      "loss": 1.543,
+      "step": 301300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.418200728706611e-05,
+      "loss": 1.5486,
+      "step": 301400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.417657890051287e-05,
+      "loss": 1.5243,
+      "step": 301500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.4171150513959635e-05,
+      "loss": 1.5519,
+      "step": 301600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.416572212740641e-05,
+      "loss": 1.5429,
+      "step": 301700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.416029374085317e-05,
+      "loss": 1.5501,
+      "step": 301800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.4154865354299934e-05,
+      "loss": 1.5872,
+      "step": 301900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.41494369677467e-05,
+      "loss": 1.5625,
+      "step": 302000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.4144008581193465e-05,
+      "loss": 1.5918,
+      "step": 302100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.413858019464023e-05,
+      "loss": 1.5764,
+      "step": 302200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.4133151808086995e-05,
+      "loss": 1.5277,
+      "step": 302300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.412772342153376e-05,
+      "loss": 1.5534,
+      "step": 302400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.4122295034980526e-05,
+      "loss": 1.562,
+      "step": 302500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.411686664842729e-05,
+      "loss": 1.5329,
+      "step": 302600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.4111438261874057e-05,
+      "loss": 1.575,
+      "step": 302700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.410600987532082e-05,
+      "loss": 1.5411,
+      "step": 302800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.410058148876759e-05,
+      "loss": 1.5783,
+      "step": 302900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.409515310221435e-05,
+      "loss": 1.5569,
+      "step": 303000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.408972471566112e-05,
+      "loss": 1.5529,
+      "step": 303100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.408429632910788e-05,
+      "loss": 1.5409,
+      "step": 303200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.407886794255464e-05,
+      "loss": 1.53,
+      "step": 303300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.407343955600141e-05,
+      "loss": 1.5868,
+      "step": 303400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.406801116944817e-05,
+      "loss": 1.5303,
+      "step": 303500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.4062582782894934e-05,
+      "loss": 1.5679,
+      "step": 303600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.405715439634171e-05,
+      "loss": 1.5392,
+      "step": 303700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.405172600978847e-05,
+      "loss": 1.5518,
+      "step": 303800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.404629762323523e-05,
+      "loss": 1.5485,
+      "step": 303900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.4040869236682e-05,
+      "loss": 1.5709,
+      "step": 304000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.4035440850128763e-05,
+      "loss": 1.558,
+      "step": 304100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.4030012463575525e-05,
+      "loss": 1.5485,
+      "step": 304200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.4024584077022294e-05,
+      "loss": 1.5486,
+      "step": 304300
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.4019155690469056e-05,
+      "loss": 1.5748,
+      "step": 304400
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.401372730391582e-05,
+      "loss": 1.5405,
+      "step": 304500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.4008298917362586e-05,
+      "loss": 1.5368,
+      "step": 304600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.400287053080935e-05,
+      "loss": 1.5462,
+      "step": 304700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.399744214425612e-05,
+      "loss": 1.5653,
+      "step": 304800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.3992013757702886e-05,
+      "loss": 1.5397,
+      "step": 304900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.398658537114965e-05,
+      "loss": 1.5411,
+      "step": 305000
+    },
+    {
+      "epoch": 1.31,
+      "eval_loss": 1.4525820016860962,
+      "eval_runtime": 17.7928,
+      "eval_samples_per_second": 562.025,
+      "eval_steps_per_second": 17.591,
+      "step": 305000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.3981156984596416e-05,
+      "loss": 1.5321,
+      "step": 305100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.397572859804318e-05,
+      "loss": 1.5782,
+      "step": 305200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.397030021148994e-05,
+      "loss": 1.5761,
+      "step": 305300
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.396487182493671e-05,
+      "loss": 1.5259,
+      "step": 305400
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.395944343838347e-05,
+      "loss": 1.5641,
+      "step": 305500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.395401505183023e-05,
+      "loss": 1.5323,
+      "step": 305600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.3948586665277e-05,
+      "loss": 1.555,
+      "step": 305700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.394315827872377e-05,
+      "loss": 1.5812,
+      "step": 305800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.393772989217053e-05,
+      "loss": 1.5379,
+      "step": 305900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.39323015056173e-05,
+      "loss": 1.5659,
+      "step": 306000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.392687311906406e-05,
+      "loss": 1.5332,
+      "step": 306100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.3921444732510824e-05,
+      "loss": 1.5472,
+      "step": 306200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.391601634595759e-05,
+      "loss": 1.5734,
+      "step": 306300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.3910587959404354e-05,
+      "loss": 1.5283,
+      "step": 306400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.3905159572851116e-05,
+      "loss": 1.5499,
+      "step": 306500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.3899731186297885e-05,
+      "loss": 1.5867,
+      "step": 306600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.389430279974465e-05,
+      "loss": 1.5538,
+      "step": 306700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.3888874413191415e-05,
+      "loss": 1.5579,
+      "step": 306800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.3883446026638184e-05,
+      "loss": 1.5155,
+      "step": 306900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.3878017640084946e-05,
+      "loss": 1.5217,
+      "step": 307000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.3872589253531715e-05,
+      "loss": 1.555,
+      "step": 307100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.3867160866978476e-05,
+      "loss": 1.5743,
+      "step": 307200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.386173248042524e-05,
+      "loss": 1.5423,
+      "step": 307300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.385630409387201e-05,
+      "loss": 1.54,
+      "step": 307400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.385087570731877e-05,
+      "loss": 1.5613,
+      "step": 307500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.384544732076553e-05,
+      "loss": 1.572,
+      "step": 307600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.38400189342123e-05,
+      "loss": 1.5829,
+      "step": 307700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.383459054765906e-05,
+      "loss": 1.5477,
+      "step": 307800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.382916216110583e-05,
+      "loss": 1.5112,
+      "step": 307900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.38237337745526e-05,
+      "loss": 1.5316,
+      "step": 308000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.381830538799936e-05,
+      "loss": 1.5668,
+      "step": 308100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.381287700144612e-05,
+      "loss": 1.5425,
+      "step": 308200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.380744861489289e-05,
+      "loss": 1.5763,
+      "step": 308300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.380202022833965e-05,
+      "loss": 1.5412,
+      "step": 308400
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3796591841786415e-05,
+      "loss": 1.5534,
+      "step": 308500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.379116345523318e-05,
+      "loss": 1.548,
+      "step": 308600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3785735068679945e-05,
+      "loss": 1.5331,
+      "step": 308700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3780306682126714e-05,
+      "loss": 1.555,
+      "step": 308800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3774878295573476e-05,
+      "loss": 1.5385,
+      "step": 308900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3769449909020244e-05,
+      "loss": 1.5703,
+      "step": 309000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.376402152246701e-05,
+      "loss": 1.5898,
+      "step": 309100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3758593135913775e-05,
+      "loss": 1.5576,
+      "step": 309200
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.375316474936054e-05,
+      "loss": 1.5749,
+      "step": 309300
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3747736362807305e-05,
+      "loss": 1.5374,
+      "step": 309400
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.374230797625407e-05,
+      "loss": 1.5718,
+      "step": 309500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.373687958970083e-05,
+      "loss": 1.5414,
+      "step": 309600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.37314512031476e-05,
+      "loss": 1.5578,
+      "step": 309700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.372602281659436e-05,
+      "loss": 1.558,
+      "step": 309800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.372059443004112e-05,
+      "loss": 1.5667,
+      "step": 309900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.37151660434879e-05,
+      "loss": 1.5619,
+      "step": 310000
+    },
+    {
+      "epoch": 1.33,
+      "eval_loss": 1.4491908550262451,
+      "eval_runtime": 17.778,
+      "eval_samples_per_second": 562.492,
+      "eval_steps_per_second": 17.606,
+      "step": 310000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.370973765693466e-05,
+      "loss": 1.5363,
+      "step": 310100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.370430927038142e-05,
+      "loss": 1.5495,
+      "step": 310200
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.369888088382819e-05,
+      "loss": 1.5621,
+      "step": 310300
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.369345249727495e-05,
+      "loss": 1.5361,
+      "step": 310400
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.368802411072171e-05,
+      "loss": 1.534,
+      "step": 310500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.368259572416848e-05,
+      "loss": 1.5424,
+      "step": 310600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3677167337615244e-05,
+      "loss": 1.5508,
+      "step": 310700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.367173895106201e-05,
+      "loss": 1.556,
+      "step": 310800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.3666310564508774e-05,
+      "loss": 1.5717,
+      "step": 310900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.3660882177955536e-05,
+      "loss": 1.538,
+      "step": 311000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.365545379140231e-05,
+      "loss": 1.5567,
+      "step": 311100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.3650025404849073e-05,
+      "loss": 1.5566,
+      "step": 311200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.3644597018295835e-05,
+      "loss": 1.5158,
+      "step": 311300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.3639168631742604e-05,
+      "loss": 1.5473,
+      "step": 311400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.3633740245189366e-05,
+      "loss": 1.5722,
+      "step": 311500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.362831185863613e-05,
+      "loss": 1.5427,
+      "step": 311600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.3622883472082896e-05,
+      "loss": 1.5306,
+      "step": 311700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.361745508552966e-05,
+      "loss": 1.5336,
+      "step": 311800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.361202669897642e-05,
+      "loss": 1.5604,
+      "step": 311900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.360659831242319e-05,
+      "loss": 1.5639,
+      "step": 312000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.360116992586996e-05,
+      "loss": 1.5306,
+      "step": 312100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.359574153931672e-05,
+      "loss": 1.5653,
+      "step": 312200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.359031315276349e-05,
+      "loss": 1.5584,
+      "step": 312300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.358488476621025e-05,
+      "loss": 1.5973,
+      "step": 312400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.357945637965701e-05,
+      "loss": 1.5377,
+      "step": 312500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.357402799310378e-05,
+      "loss": 1.5437,
+      "step": 312600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.356859960655054e-05,
+      "loss": 1.5638,
+      "step": 312700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.356317121999731e-05,
+      "loss": 1.5189,
+      "step": 312800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.355774283344407e-05,
+      "loss": 1.5334,
+      "step": 312900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.3552314446890835e-05,
+      "loss": 1.5373,
+      "step": 313000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.35468860603376e-05,
+      "loss": 1.5312,
+      "step": 313100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.354145767378437e-05,
+      "loss": 1.5472,
+      "step": 313200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.3536029287231134e-05,
+      "loss": 1.5665,
+      "step": 313300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.35306009006779e-05,
+      "loss": 1.5235,
+      "step": 313400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.3525172514124664e-05,
+      "loss": 1.5534,
+      "step": 313500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.3519744127571426e-05,
+      "loss": 1.5593,
+      "step": 313600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.3514315741018195e-05,
+      "loss": 1.5607,
+      "step": 313700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.350888735446496e-05,
+      "loss": 1.5367,
+      "step": 313800
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.350345896791172e-05,
+      "loss": 1.5468,
+      "step": 313900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.349803058135849e-05,
+      "loss": 1.5385,
+      "step": 314000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.349260219480525e-05,
+      "loss": 1.59,
+      "step": 314100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.348717380825202e-05,
+      "loss": 1.5429,
+      "step": 314200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.3481745421698786e-05,
+      "loss": 1.5347,
+      "step": 314300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.347631703514555e-05,
+      "loss": 1.5435,
+      "step": 314400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.347088864859231e-05,
+      "loss": 1.5158,
+      "step": 314500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.346546026203908e-05,
+      "loss": 1.5348,
+      "step": 314600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.346003187548584e-05,
+      "loss": 1.5336,
+      "step": 314700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.345460348893261e-05,
+      "loss": 1.5486,
+      "step": 314800
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.344917510237937e-05,
+      "loss": 1.5407,
+      "step": 314900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.344374671582613e-05,
+      "loss": 1.5393,
+      "step": 315000
+    },
+    {
+      "epoch": 1.35,
+      "eval_loss": 1.4514410495758057,
+      "eval_runtime": 17.8233,
+      "eval_samples_per_second": 561.064,
+      "eval_steps_per_second": 17.561,
+      "step": 315000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.34383183292729e-05,
+      "loss": 1.5579,
+      "step": 315100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.3432889942719664e-05,
+      "loss": 1.5455,
+      "step": 315200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.342746155616643e-05,
+      "loss": 1.514,
+      "step": 315300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.34220331696132e-05,
+      "loss": 1.5846,
+      "step": 315400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.341660478305996e-05,
+      "loss": 1.5471,
+      "step": 315500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.3411176396506725e-05,
+      "loss": 1.5392,
+      "step": 315600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.340574800995349e-05,
+      "loss": 1.5526,
+      "step": 315700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.3400319623400255e-05,
+      "loss": 1.5548,
+      "step": 315800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.339489123684702e-05,
+      "loss": 1.545,
+      "step": 315900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.3389462850293786e-05,
+      "loss": 1.5535,
+      "step": 316000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.338403446374055e-05,
+      "loss": 1.5481,
+      "step": 316100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.3378606077187316e-05,
+      "loss": 1.5492,
+      "step": 316200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.3373177690634085e-05,
+      "loss": 1.549,
+      "step": 316300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.336774930408085e-05,
+      "loss": 1.5482,
+      "step": 316400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.336232091752761e-05,
+      "loss": 1.5448,
+      "step": 316500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.335689253097438e-05,
+      "loss": 1.5604,
+      "step": 316600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.335146414442114e-05,
+      "loss": 1.5651,
+      "step": 316700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.334603575786791e-05,
+      "loss": 1.5387,
+      "step": 316800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.334060737131467e-05,
+      "loss": 1.5601,
+      "step": 316900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.333517898476143e-05,
+      "loss": 1.5667,
+      "step": 317000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.33297505982082e-05,
+      "loss": 1.5659,
+      "step": 317100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.332432221165496e-05,
+      "loss": 1.5477,
+      "step": 317200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.3318893825101724e-05,
+      "loss": 1.5303,
+      "step": 317300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.33134654385485e-05,
+      "loss": 1.5476,
+      "step": 317400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.330803705199526e-05,
+      "loss": 1.56,
+      "step": 317500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.330260866544202e-05,
+      "loss": 1.5144,
+      "step": 317600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.329718027888879e-05,
+      "loss": 1.5498,
+      "step": 317700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.3291751892335554e-05,
+      "loss": 1.5201,
+      "step": 317800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.3286323505782316e-05,
+      "loss": 1.5448,
+      "step": 317900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.3280895119229084e-05,
+      "loss": 1.5226,
+      "step": 318000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.3275466732675846e-05,
+      "loss": 1.5646,
+      "step": 318100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.3270038346122615e-05,
+      "loss": 1.5561,
+      "step": 318200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.3264609959569377e-05,
+      "loss": 1.5407,
+      "step": 318300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.3259181573016145e-05,
+      "loss": 1.5461,
+      "step": 318400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.325375318646291e-05,
+      "loss": 1.5291,
+      "step": 318500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.3248324799909676e-05,
+      "loss": 1.5518,
+      "step": 318600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.324289641335644e-05,
+      "loss": 1.5508,
+      "step": 318700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.3237468026803206e-05,
+      "loss": 1.5272,
+      "step": 318800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.323203964024997e-05,
+      "loss": 1.573,
+      "step": 318900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.322661125369673e-05,
+      "loss": 1.5179,
+      "step": 319000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.32211828671435e-05,
+      "loss": 1.5202,
+      "step": 319100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.321575448059026e-05,
+      "loss": 1.5343,
+      "step": 319200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.321032609403702e-05,
+      "loss": 1.5627,
+      "step": 319300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.320489770748379e-05,
+      "loss": 1.5392,
+      "step": 319400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.319946932093056e-05,
+      "loss": 1.5463,
+      "step": 319500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.319404093437732e-05,
+      "loss": 1.5394,
+      "step": 319600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.318861254782409e-05,
+      "loss": 1.5631,
+      "step": 319700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.318318416127085e-05,
+      "loss": 1.553,
+      "step": 319800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.3177755774717614e-05,
+      "loss": 1.5476,
+      "step": 319900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.317232738816438e-05,
+      "loss": 1.5625,
+      "step": 320000
+    },
+    {
+      "epoch": 1.37,
+      "eval_loss": 1.449994683265686,
+      "eval_runtime": 17.8575,
+      "eval_samples_per_second": 559.99,
+      "eval_steps_per_second": 17.528,
+      "step": 320000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.3166899001611145e-05,
+      "loss": 1.5589,
+      "step": 320100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.316147061505791e-05,
+      "loss": 1.5389,
+      "step": 320200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.3156042228504675e-05,
+      "loss": 1.5409,
+      "step": 320300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.315061384195144e-05,
+      "loss": 1.578,
+      "step": 320400
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.3145185455398206e-05,
+      "loss": 1.5632,
+      "step": 320500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.3139757068844974e-05,
+      "loss": 1.5289,
+      "step": 320600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.3134328682291736e-05,
+      "loss": 1.5462,
+      "step": 320700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.3128900295738505e-05,
+      "loss": 1.516,
+      "step": 320800
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.312347190918527e-05,
+      "loss": 1.5305,
+      "step": 320900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.311804352263203e-05,
+      "loss": 1.5529,
+      "step": 321000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.31126151360788e-05,
+      "loss": 1.5553,
+      "step": 321100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.310718674952556e-05,
+      "loss": 1.5441,
+      "step": 321200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.310175836297232e-05,
+      "loss": 1.5222,
+      "step": 321300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.309632997641909e-05,
+      "loss": 1.5558,
+      "step": 321400
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.309090158986585e-05,
+      "loss": 1.5436,
+      "step": 321500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.308547320331262e-05,
+      "loss": 1.5389,
+      "step": 321600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.308004481675939e-05,
+      "loss": 1.5522,
+      "step": 321700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.307461643020615e-05,
+      "loss": 1.5622,
+      "step": 321800
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.306918804365291e-05,
+      "loss": 1.565,
+      "step": 321900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.306375965709968e-05,
+      "loss": 1.5737,
+      "step": 322000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.305833127054644e-05,
+      "loss": 1.5529,
+      "step": 322100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.305290288399321e-05,
+      "loss": 1.5312,
+      "step": 322200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.3047474497439974e-05,
+      "loss": 1.547,
+      "step": 322300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.3042046110886735e-05,
+      "loss": 1.5254,
+      "step": 322400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.3036617724333504e-05,
+      "loss": 1.542,
+      "step": 322500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.303118933778027e-05,
+      "loss": 1.5424,
+      "step": 322600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.3025760951227035e-05,
+      "loss": 1.5567,
+      "step": 322700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.30203325646738e-05,
+      "loss": 1.5519,
+      "step": 322800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.3014904178120565e-05,
+      "loss": 1.5261,
+      "step": 322900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.300947579156733e-05,
+      "loss": 1.5377,
+      "step": 323000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.3004047405014096e-05,
+      "loss": 1.5222,
+      "step": 323100
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.299861901846086e-05,
+      "loss": 1.5657,
+      "step": 323200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.299319063190762e-05,
+      "loss": 1.593,
+      "step": 323300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.298776224535439e-05,
+      "loss": 1.579,
+      "step": 323400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.298233385880115e-05,
+      "loss": 1.5289,
+      "step": 323500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.297690547224792e-05,
+      "loss": 1.5249,
+      "step": 323600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.297147708569469e-05,
+      "loss": 1.5064,
+      "step": 323700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.296604869914145e-05,
+      "loss": 1.563,
+      "step": 323800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.296062031258821e-05,
+      "loss": 1.5391,
+      "step": 323900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.295519192603498e-05,
+      "loss": 1.5499,
+      "step": 324000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.294976353948174e-05,
+      "loss": 1.533,
+      "step": 324100
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.294433515292851e-05,
+      "loss": 1.5583,
+      "step": 324200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.293890676637527e-05,
+      "loss": 1.5807,
+      "step": 324300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.2933478379822034e-05,
+      "loss": 1.5936,
+      "step": 324400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.29280499932688e-05,
+      "loss": 1.5425,
+      "step": 324500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.2922621606715564e-05,
+      "loss": 1.5584,
+      "step": 324600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.291719322016233e-05,
+      "loss": 1.5303,
+      "step": 324700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.29117648336091e-05,
+      "loss": 1.513,
+      "step": 324800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.2906336447055864e-05,
+      "loss": 1.5649,
+      "step": 324900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.2900908060502626e-05,
+      "loss": 1.5681,
+      "step": 325000
+    },
+    {
+      "epoch": 1.4,
+      "eval_loss": 1.451581597328186,
+      "eval_runtime": 17.7955,
+      "eval_samples_per_second": 561.941,
+      "eval_steps_per_second": 17.589,
+      "step": 325000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.2895479673949394e-05,
+      "loss": 1.5595,
+      "step": 325100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.2890051287396156e-05,
+      "loss": 1.571,
+      "step": 325200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.288462290084292e-05,
+      "loss": 1.5265,
+      "step": 325300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.2879194514289687e-05,
+      "loss": 1.5329,
+      "step": 325400
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.287376612773645e-05,
+      "loss": 1.5374,
+      "step": 325500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.286833774118322e-05,
+      "loss": 1.5731,
+      "step": 325600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.2862909354629986e-05,
+      "loss": 1.5147,
+      "step": 325700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.285748096807675e-05,
+      "loss": 1.5584,
+      "step": 325800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.285205258152351e-05,
+      "loss": 1.5324,
+      "step": 325900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.284662419497028e-05,
+      "loss": 1.5545,
+      "step": 326000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.284119580841704e-05,
+      "loss": 1.5175,
+      "step": 326100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.283576742186381e-05,
+      "loss": 1.5394,
+      "step": 326200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.283033903531057e-05,
+      "loss": 1.5742,
+      "step": 326300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.282491064875733e-05,
+      "loss": 1.5592,
+      "step": 326400
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.28194822622041e-05,
+      "loss": 1.5361,
+      "step": 326500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.281405387565086e-05,
+      "loss": 1.5426,
+      "step": 326600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.2808625489097625e-05,
+      "loss": 1.5495,
+      "step": 326700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.28031971025444e-05,
+      "loss": 1.5557,
+      "step": 326800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.279776871599116e-05,
+      "loss": 1.5362,
+      "step": 326900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.2792340329437924e-05,
+      "loss": 1.5475,
+      "step": 327000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.278691194288469e-05,
+      "loss": 1.573,
+      "step": 327100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.2781483556331455e-05,
+      "loss": 1.5561,
+      "step": 327200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.2776055169778216e-05,
+      "loss": 1.563,
+      "step": 327300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.2770626783224985e-05,
+      "loss": 1.5532,
+      "step": 327400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.276519839667175e-05,
+      "loss": 1.5384,
+      "step": 327500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.2759770010118516e-05,
+      "loss": 1.5402,
+      "step": 327600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.275434162356528e-05,
+      "loss": 1.5416,
+      "step": 327700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.2748913237012046e-05,
+      "loss": 1.5601,
+      "step": 327800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.274348485045881e-05,
+      "loss": 1.5608,
+      "step": 327900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.273805646390558e-05,
+      "loss": 1.5469,
+      "step": 328000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.273262807735234e-05,
+      "loss": 1.5253,
+      "step": 328100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.272719969079911e-05,
+      "loss": 1.5558,
+      "step": 328200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.272177130424587e-05,
+      "loss": 1.5568,
+      "step": 328300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.271634291769263e-05,
+      "loss": 1.4968,
+      "step": 328400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.27109145311394e-05,
+      "loss": 1.5454,
+      "step": 328500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.270548614458616e-05,
+      "loss": 1.5199,
+      "step": 328600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.270005775803292e-05,
+      "loss": 1.5302,
+      "step": 328700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.269462937147969e-05,
+      "loss": 1.5254,
+      "step": 328800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.268920098492646e-05,
+      "loss": 1.5374,
+      "step": 328900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.268377259837322e-05,
+      "loss": 1.5471,
+      "step": 329000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.267834421181999e-05,
+      "loss": 1.5328,
+      "step": 329100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.267291582526675e-05,
+      "loss": 1.5556,
+      "step": 329200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.2667487438713515e-05,
+      "loss": 1.5498,
+      "step": 329300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.2662059052160284e-05,
+      "loss": 1.5521,
+      "step": 329400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.2656630665607045e-05,
+      "loss": 1.551,
+      "step": 329500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.2651202279053814e-05,
+      "loss": 1.5019,
+      "step": 329600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.2645773892500576e-05,
+      "loss": 1.5463,
+      "step": 329700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.264034550594734e-05,
+      "loss": 1.5388,
+      "step": 329800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.2634917119394106e-05,
+      "loss": 1.5319,
+      "step": 329900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.2629488732840875e-05,
+      "loss": 1.5459,
+      "step": 330000
+    },
+    {
+      "epoch": 1.42,
+      "eval_loss": 1.4472662210464478,
+      "eval_runtime": 17.7977,
+      "eval_samples_per_second": 561.871,
+      "eval_steps_per_second": 17.587,
+      "step": 330000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.262406034628764e-05,
+      "loss": 1.5692,
+      "step": 330100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.2618631959734406e-05,
+      "loss": 1.5344,
+      "step": 330200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.261320357318117e-05,
+      "loss": 1.5314,
+      "step": 330300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.260777518662793e-05,
+      "loss": 1.5389,
+      "step": 330400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.26023468000747e-05,
+      "loss": 1.5467,
+      "step": 330500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.259691841352146e-05,
+      "loss": 1.5338,
+      "step": 330600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.259149002696822e-05,
+      "loss": 1.5669,
+      "step": 330700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.258606164041499e-05,
+      "loss": 1.5344,
+      "step": 330800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.258063325386175e-05,
+      "loss": 1.5236,
+      "step": 330900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.257520486730852e-05,
+      "loss": 1.5438,
+      "step": 331000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.256977648075529e-05,
+      "loss": 1.5269,
+      "step": 331100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.256434809420205e-05,
+      "loss": 1.5401,
+      "step": 331200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.255891970764881e-05,
+      "loss": 1.5533,
+      "step": 331300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.255349132109558e-05,
+      "loss": 1.5514,
+      "step": 331400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.2548062934542344e-05,
+      "loss": 1.5562,
+      "step": 331500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.254263454798911e-05,
+      "loss": 1.5358,
+      "step": 331600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.2537206161435874e-05,
+      "loss": 1.5287,
+      "step": 331700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.2531777774882636e-05,
+      "loss": 1.5347,
+      "step": 331800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.2526349388329405e-05,
+      "loss": 1.5572,
+      "step": 331900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.2520921001776174e-05,
+      "loss": 1.5723,
+      "step": 332000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.2515492615222935e-05,
+      "loss": 1.5383,
+      "step": 332100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.2510064228669704e-05,
+      "loss": 1.5197,
+      "step": 332200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.2504635842116466e-05,
+      "loss": 1.5499,
+      "step": 332300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.249920745556323e-05,
+      "loss": 1.5379,
+      "step": 332400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.2493779069009997e-05,
+      "loss": 1.5418,
+      "step": 332500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.248835068245676e-05,
+      "loss": 1.5495,
+      "step": 332600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.248292229590352e-05,
+      "loss": 1.5348,
+      "step": 332700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.247749390935029e-05,
+      "loss": 1.55,
+      "step": 332800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.247206552279705e-05,
+      "loss": 1.5507,
+      "step": 332900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.246663713624381e-05,
+      "loss": 1.5419,
+      "step": 333000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.246120874969059e-05,
+      "loss": 1.5527,
+      "step": 333100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.245578036313735e-05,
+      "loss": 1.5638,
+      "step": 333200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.245035197658411e-05,
+      "loss": 1.5521,
+      "step": 333300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.244492359003088e-05,
+      "loss": 1.5566,
+      "step": 333400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.243949520347764e-05,
+      "loss": 1.5747,
+      "step": 333500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.243406681692441e-05,
+      "loss": 1.5436,
+      "step": 333600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.242863843037117e-05,
+      "loss": 1.5771,
+      "step": 333700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.2423210043817935e-05,
+      "loss": 1.5296,
+      "step": 333800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.2417781657264703e-05,
+      "loss": 1.5294,
+      "step": 333900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.2412353270711465e-05,
+      "loss": 1.5422,
+      "step": 334000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.2406924884158234e-05,
+      "loss": 1.5428,
+      "step": 334100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.2401496497605e-05,
+      "loss": 1.5283,
+      "step": 334200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.2396068111051764e-05,
+      "loss": 1.559,
+      "step": 334300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.2390639724498526e-05,
+      "loss": 1.5509,
+      "step": 334400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.2385211337945295e-05,
+      "loss": 1.5446,
+      "step": 334500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.237978295139206e-05,
+      "loss": 1.5647,
+      "step": 334600
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.237435456483882e-05,
+      "loss": 1.5359,
+      "step": 334700
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.236892617828559e-05,
+      "loss": 1.5501,
+      "step": 334800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.236349779173235e-05,
+      "loss": 1.5473,
+      "step": 334900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.235806940517911e-05,
+      "loss": 1.5413,
+      "step": 335000
+    },
+    {
+      "epoch": 1.44,
+      "eval_loss": 1.4457744359970093,
+      "eval_runtime": 17.8184,
+      "eval_samples_per_second": 561.217,
+      "eval_steps_per_second": 17.566,
+      "step": 335000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.235264101862588e-05,
+      "loss": 1.5361,
+      "step": 335100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.234721263207265e-05,
+      "loss": 1.5402,
+      "step": 335200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.234178424551941e-05,
+      "loss": 1.5459,
+      "step": 335300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.233635585896618e-05,
+      "loss": 1.5213,
+      "step": 335400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.233092747241294e-05,
+      "loss": 1.5281,
+      "step": 335500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.232549908585971e-05,
+      "loss": 1.547,
+      "step": 335600
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.232007069930647e-05,
+      "loss": 1.5668,
+      "step": 335700
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.231464231275323e-05,
+      "loss": 1.533,
+      "step": 335800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.23092139262e-05,
+      "loss": 1.5421,
+      "step": 335900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.2303785539646764e-05,
+      "loss": 1.5544,
+      "step": 336000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.2298357153093526e-05,
+      "loss": 1.5418,
+      "step": 336100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.22929287665403e-05,
+      "loss": 1.5508,
+      "step": 336200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.228750037998706e-05,
+      "loss": 1.5832,
+      "step": 336300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.2282071993433825e-05,
+      "loss": 1.5215,
+      "step": 336400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.2276643606880594e-05,
+      "loss": 1.5246,
+      "step": 336500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.2271215220327355e-05,
+      "loss": 1.5504,
+      "step": 336600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.226578683377412e-05,
+      "loss": 1.5488,
+      "step": 336700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.2260358447220886e-05,
+      "loss": 1.5597,
+      "step": 336800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.225493006066765e-05,
+      "loss": 1.5257,
+      "step": 336900
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.224950167411441e-05,
+      "loss": 1.5254,
+      "step": 337000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.224407328756118e-05,
+      "loss": 1.5362,
+      "step": 337100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.223864490100794e-05,
+      "loss": 1.5569,
+      "step": 337200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.223321651445471e-05,
+      "loss": 1.5443,
+      "step": 337300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.222778812790148e-05,
+      "loss": 1.5064,
+      "step": 337400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.222235974134824e-05,
+      "loss": 1.5525,
+      "step": 337500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.221693135479501e-05,
+      "loss": 1.5267,
+      "step": 337600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.221150296824177e-05,
+      "loss": 1.498,
+      "step": 337700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.220607458168853e-05,
+      "loss": 1.5337,
+      "step": 337800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.22006461951353e-05,
+      "loss": 1.5393,
+      "step": 337900
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.219521780858206e-05,
+      "loss": 1.5382,
+      "step": 338000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.2189789422028824e-05,
+      "loss": 1.5627,
+      "step": 338100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.218436103547559e-05,
+      "loss": 1.531,
+      "step": 338200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.217893264892236e-05,
+      "loss": 1.5392,
+      "step": 338300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.217350426236912e-05,
+      "loss": 1.5617,
+      "step": 338400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.216807587581589e-05,
+      "loss": 1.5522,
+      "step": 338500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.2162647489262654e-05,
+      "loss": 1.5298,
+      "step": 338600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.2157219102709416e-05,
+      "loss": 1.5476,
+      "step": 338700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.2151790716156184e-05,
+      "loss": 1.542,
+      "step": 338800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.2146362329602946e-05,
+      "loss": 1.5461,
+      "step": 338900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.214093394304971e-05,
+      "loss": 1.5699,
+      "step": 339000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.213550555649648e-05,
+      "loss": 1.5064,
+      "step": 339100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.213007716994324e-05,
+      "loss": 1.5375,
+      "step": 339200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.212464878339001e-05,
+      "loss": 1.5728,
+      "step": 339300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.2119220396836776e-05,
+      "loss": 1.5027,
+      "step": 339400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.211379201028354e-05,
+      "loss": 1.5562,
+      "step": 339500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.2108363623730307e-05,
+      "loss": 1.5628,
+      "step": 339600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.210293523717707e-05,
+      "loss": 1.5528,
+      "step": 339700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.209750685062383e-05,
+      "loss": 1.5226,
+      "step": 339800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.20920784640706e-05,
+      "loss": 1.5621,
+      "step": 339900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.208665007751736e-05,
+      "loss": 1.5492,
+      "step": 340000
+    },
+    {
+      "epoch": 1.46,
+      "eval_loss": 1.4450249671936035,
+      "eval_runtime": 17.7966,
+      "eval_samples_per_second": 561.904,
+      "eval_steps_per_second": 17.588,
+      "step": 340000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.208122169096412e-05,
+      "loss": 1.5556,
+      "step": 340100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.207579330441089e-05,
+      "loss": 1.5418,
+      "step": 340200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.207036491785765e-05,
+      "loss": 1.5288,
+      "step": 340300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.206493653130442e-05,
+      "loss": 1.5588,
+      "step": 340400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.205950814475119e-05,
+      "loss": 1.54,
+      "step": 340500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.205407975819795e-05,
+      "loss": 1.5419,
+      "step": 340600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.2048651371644714e-05,
+      "loss": 1.5594,
+      "step": 340700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.204322298509148e-05,
+      "loss": 1.5502,
+      "step": 340800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.2037794598538245e-05,
+      "loss": 1.5399,
+      "step": 340900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.203236621198501e-05,
+      "loss": 1.5661,
+      "step": 341000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.2026937825431775e-05,
+      "loss": 1.5435,
+      "step": 341100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.202150943887854e-05,
+      "loss": 1.5386,
+      "step": 341200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.2016081052325306e-05,
+      "loss": 1.5634,
+      "step": 341300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.201065266577207e-05,
+      "loss": 1.5215,
+      "step": 341400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.2005224279218836e-05,
+      "loss": 1.5667,
+      "step": 341500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.1999795892665605e-05,
+      "loss": 1.5362,
+      "step": 341600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.199436750611237e-05,
+      "loss": 1.5467,
+      "step": 341700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.198893911955913e-05,
+      "loss": 1.5496,
+      "step": 341800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.19835107330059e-05,
+      "loss": 1.5406,
+      "step": 341900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.197808234645266e-05,
+      "loss": 1.5568,
+      "step": 342000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.197265395989942e-05,
+      "loss": 1.5319,
+      "step": 342100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.196722557334619e-05,
+      "loss": 1.5338,
+      "step": 342200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.196179718679295e-05,
+      "loss": 1.5516,
+      "step": 342300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.1956368800239714e-05,
+      "loss": 1.5399,
+      "step": 342400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.195094041368649e-05,
+      "loss": 1.545,
+      "step": 342500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.194551202713325e-05,
+      "loss": 1.5531,
+      "step": 342600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.194008364058001e-05,
+      "loss": 1.5147,
+      "step": 342700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.193465525402678e-05,
+      "loss": 1.5641,
+      "step": 342800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.192922686747354e-05,
+      "loss": 1.5532,
+      "step": 342900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.1923798480920305e-05,
+      "loss": 1.5282,
+      "step": 343000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.1918370094367074e-05,
+      "loss": 1.5503,
+      "step": 343100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.1912941707813836e-05,
+      "loss": 1.5604,
+      "step": 343200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.1907513321260604e-05,
+      "loss": 1.527,
+      "step": 343300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1902084934707366e-05,
+      "loss": 1.547,
+      "step": 343400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1896656548154135e-05,
+      "loss": 1.5202,
+      "step": 343500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1891228161600903e-05,
+      "loss": 1.5646,
+      "step": 343600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1885799775047665e-05,
+      "loss": 1.5329,
+      "step": 343700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.188037138849443e-05,
+      "loss": 1.5512,
+      "step": 343800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1874943001941196e-05,
+      "loss": 1.5299,
+      "step": 343900
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.186951461538796e-05,
+      "loss": 1.5369,
+      "step": 344000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.186408622883472e-05,
+      "loss": 1.5414,
+      "step": 344100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.185865784228149e-05,
+      "loss": 1.5322,
+      "step": 344200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.185322945572825e-05,
+      "loss": 1.5215,
+      "step": 344300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.184780106917501e-05,
+      "loss": 1.5493,
+      "step": 344400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.184237268262178e-05,
+      "loss": 1.5299,
+      "step": 344500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.183694429606855e-05,
+      "loss": 1.5392,
+      "step": 344600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.183151590951531e-05,
+      "loss": 1.5435,
+      "step": 344700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.182608752296208e-05,
+      "loss": 1.5557,
+      "step": 344800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.182065913640884e-05,
+      "loss": 1.5388,
+      "step": 344900
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1815230749855604e-05,
+      "loss": 1.5414,
+      "step": 345000
+    },
+    {
+      "epoch": 1.48,
+      "eval_loss": 1.4478392601013184,
+      "eval_runtime": 17.7608,
+      "eval_samples_per_second": 563.038,
+      "eval_steps_per_second": 17.623,
+      "step": 345000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.180980236330237e-05,
+      "loss": 1.5371,
+      "step": 345100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1804373976749134e-05,
+      "loss": 1.5326,
+      "step": 345200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.17989455901959e-05,
+      "loss": 1.5514,
+      "step": 345300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1793517203642665e-05,
+      "loss": 1.5307,
+      "step": 345400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1788088817089427e-05,
+      "loss": 1.5185,
+      "step": 345500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1782660430536195e-05,
+      "loss": 1.575,
+      "step": 345600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1777232043982964e-05,
+      "loss": 1.5719,
+      "step": 345700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1771803657429726e-05,
+      "loss": 1.5208,
+      "step": 345800
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1766375270876494e-05,
+      "loss": 1.5388,
+      "step": 345900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1760946884323256e-05,
+      "loss": 1.5615,
+      "step": 346000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.175551849777002e-05,
+      "loss": 1.5374,
+      "step": 346100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.175009011121679e-05,
+      "loss": 1.5419,
+      "step": 346200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.174466172466355e-05,
+      "loss": 1.5355,
+      "step": 346300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.173923333811031e-05,
+      "loss": 1.5732,
+      "step": 346400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.173380495155708e-05,
+      "loss": 1.514,
+      "step": 346500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.172837656500384e-05,
+      "loss": 1.5063,
+      "step": 346600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.172294817845061e-05,
+      "loss": 1.5282,
+      "step": 346700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.171751979189738e-05,
+      "loss": 1.5447,
+      "step": 346800
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.171209140534414e-05,
+      "loss": 1.553,
+      "step": 346900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.17066630187909e-05,
+      "loss": 1.5592,
+      "step": 347000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.170123463223767e-05,
+      "loss": 1.562,
+      "step": 347100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.169580624568443e-05,
+      "loss": 1.5373,
+      "step": 347200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.16903778591312e-05,
+      "loss": 1.5465,
+      "step": 347300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.168494947257796e-05,
+      "loss": 1.5292,
+      "step": 347400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1679521086024725e-05,
+      "loss": 1.5545,
+      "step": 347500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1674092699471494e-05,
+      "loss": 1.5225,
+      "step": 347600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.166866431291826e-05,
+      "loss": 1.5182,
+      "step": 347700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1663235926365024e-05,
+      "loss": 1.5689,
+      "step": 347800
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.165780753981179e-05,
+      "loss": 1.535,
+      "step": 347900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1652379153258555e-05,
+      "loss": 1.5145,
+      "step": 348000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1646950766705317e-05,
+      "loss": 1.5291,
+      "step": 348100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1641522380152085e-05,
+      "loss": 1.5232,
+      "step": 348200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.163609399359885e-05,
+      "loss": 1.5489,
+      "step": 348300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.163066560704561e-05,
+      "loss": 1.5601,
+      "step": 348400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.162523722049238e-05,
+      "loss": 1.5143,
+      "step": 348500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.161980883393914e-05,
+      "loss": 1.5376,
+      "step": 348600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.161438044738591e-05,
+      "loss": 1.5192,
+      "step": 348700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.160895206083268e-05,
+      "loss": 1.5509,
+      "step": 348800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.160352367427944e-05,
+      "loss": 1.5483,
+      "step": 348900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.15980952877262e-05,
+      "loss": 1.5048,
+      "step": 349000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.159266690117297e-05,
+      "loss": 1.522,
+      "step": 349100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.158723851461973e-05,
+      "loss": 1.5402,
+      "step": 349200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.15818101280665e-05,
+      "loss": 1.5543,
+      "step": 349300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.157638174151326e-05,
+      "loss": 1.5409,
+      "step": 349400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1570953354960024e-05,
+      "loss": 1.5243,
+      "step": 349500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.156552496840679e-05,
+      "loss": 1.5425,
+      "step": 349600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1560096581853554e-05,
+      "loss": 1.5237,
+      "step": 349700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.155466819530032e-05,
+      "loss": 1.541,
+      "step": 349800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.154923980874709e-05,
+      "loss": 1.5575,
+      "step": 349900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.154381142219385e-05,
+      "loss": 1.5517,
+      "step": 350000
+    },
+    {
+      "epoch": 1.5,
+      "eval_loss": 1.4470714330673218,
+      "eval_runtime": 17.6899,
+      "eval_samples_per_second": 565.294,
+      "eval_steps_per_second": 17.694,
+      "step": 350000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1538383035640615e-05,
+      "loss": 1.532,
+      "step": 350100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1532954649087384e-05,
+      "loss": 1.5497,
+      "step": 350200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1527526262534146e-05,
+      "loss": 1.545,
+      "step": 350300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.152209787598091e-05,
+      "loss": 1.5249,
+      "step": 350400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1516669489427676e-05,
+      "loss": 1.5464,
+      "step": 350500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.151124110287444e-05,
+      "loss": 1.557,
+      "step": 350600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.150581271632121e-05,
+      "loss": 1.5432,
+      "step": 350700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.150038432976797e-05,
+      "loss": 1.5386,
+      "step": 350800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.149495594321474e-05,
+      "loss": 1.5456,
+      "step": 350900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.14895275566615e-05,
+      "loss": 1.538,
+      "step": 351000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.148409917010827e-05,
+      "loss": 1.5359,
+      "step": 351100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.147867078355503e-05,
+      "loss": 1.5138,
+      "step": 351200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.14732423970018e-05,
+      "loss": 1.5448,
+      "step": 351300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.146781401044856e-05,
+      "loss": 1.536,
+      "step": 351400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.146238562389532e-05,
+      "loss": 1.5433,
+      "step": 351500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.145695723734209e-05,
+      "loss": 1.5384,
+      "step": 351600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.145152885078885e-05,
+      "loss": 1.5072,
+      "step": 351700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1446100464235614e-05,
+      "loss": 1.5633,
+      "step": 351800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.144067207768239e-05,
+      "loss": 1.5238,
+      "step": 351900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.143524369112915e-05,
+      "loss": 1.5614,
+      "step": 352000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1429815304575914e-05,
+      "loss": 1.5744,
+      "step": 352100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.142438691802268e-05,
+      "loss": 1.5098,
+      "step": 352200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1418958531469444e-05,
+      "loss": 1.5568,
+      "step": 352300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1413530144916206e-05,
+      "loss": 1.5406,
+      "step": 352400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1408101758362975e-05,
+      "loss": 1.5192,
+      "step": 352500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1402673371809737e-05,
+      "loss": 1.5363,
+      "step": 352600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1397244985256505e-05,
+      "loss": 1.5626,
+      "step": 352700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.139181659870327e-05,
+      "loss": 1.5644,
+      "step": 352800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.138638821215003e-05,
+      "loss": 1.5126,
+      "step": 352900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.13809598255968e-05,
+      "loss": 1.5045,
+      "step": 353000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1375531439043566e-05,
+      "loss": 1.5488,
+      "step": 353100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.137010305249033e-05,
+      "loss": 1.5111,
+      "step": 353200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.13646746659371e-05,
+      "loss": 1.5539,
+      "step": 353300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.135924627938386e-05,
+      "loss": 1.5193,
+      "step": 353400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.135381789283062e-05,
+      "loss": 1.5483,
+      "step": 353500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.134838950627739e-05,
+      "loss": 1.5216,
+      "step": 353600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.134296111972415e-05,
+      "loss": 1.5613,
+      "step": 353700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.133753273317091e-05,
+      "loss": 1.5414,
+      "step": 353800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.133210434661768e-05,
+      "loss": 1.5307,
+      "step": 353900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.132667596006445e-05,
+      "loss": 1.5392,
+      "step": 354000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.132124757351121e-05,
+      "loss": 1.5457,
+      "step": 354100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.131581918695798e-05,
+      "loss": 1.539,
+      "step": 354200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.131039080040474e-05,
+      "loss": 1.5216,
+      "step": 354300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1304962413851504e-05,
+      "loss": 1.5752,
+      "step": 354400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.129953402729827e-05,
+      "loss": 1.5301,
+      "step": 354500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1294105640745035e-05,
+      "loss": 1.5457,
+      "step": 354600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1288677254191804e-05,
+      "loss": 1.5437,
+      "step": 354700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1283248867638566e-05,
+      "loss": 1.5305,
+      "step": 354800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.127782048108533e-05,
+      "loss": 1.5045,
+      "step": 354900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1272392094532096e-05,
+      "loss": 1.5433,
+      "step": 355000
+    },
+    {
+      "epoch": 1.53,
+      "eval_loss": 1.4460771083831787,
+      "eval_runtime": 17.7968,
+      "eval_samples_per_second": 561.9,
+      "eval_steps_per_second": 17.587,
+      "step": 355000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1266963707978865e-05,
+      "loss": 1.5781,
+      "step": 355100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1261535321425627e-05,
+      "loss": 1.5645,
+      "step": 355200
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1256106934872395e-05,
+      "loss": 1.5349,
+      "step": 355300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.125067854831916e-05,
+      "loss": 1.5459,
+      "step": 355400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.124525016176592e-05,
+      "loss": 1.5356,
+      "step": 355500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.123982177521269e-05,
+      "loss": 1.556,
+      "step": 355600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.123439338865945e-05,
+      "loss": 1.5376,
+      "step": 355700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.122896500210621e-05,
+      "loss": 1.5308,
+      "step": 355800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.122353661555298e-05,
+      "loss": 1.5548,
+      "step": 355900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.121810822899974e-05,
+      "loss": 1.5357,
+      "step": 356000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.121267984244651e-05,
+      "loss": 1.5202,
+      "step": 356100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.120725145589328e-05,
+      "loss": 1.5478,
+      "step": 356200
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.120182306934004e-05,
+      "loss": 1.5502,
+      "step": 356300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.11963946827868e-05,
+      "loss": 1.5432,
+      "step": 356400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.119096629623357e-05,
+      "loss": 1.5386,
+      "step": 356500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1185537909680333e-05,
+      "loss": 1.5578,
+      "step": 356600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.11801095231271e-05,
+      "loss": 1.5365,
+      "step": 356700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1174681136573864e-05,
+      "loss": 1.5718,
+      "step": 356800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1169252750020626e-05,
+      "loss": 1.5305,
+      "step": 356900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1163824363467395e-05,
+      "loss": 1.5333,
+      "step": 357000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1158395976914156e-05,
+      "loss": 1.5484,
+      "step": 357100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1152967590360925e-05,
+      "loss": 1.5521,
+      "step": 357200
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1147539203807694e-05,
+      "loss": 1.5207,
+      "step": 357300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.1142110817254456e-05,
+      "loss": 1.5525,
+      "step": 357400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.113668243070122e-05,
+      "loss": 1.5497,
+      "step": 357500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.1131254044147986e-05,
+      "loss": 1.5305,
+      "step": 357600
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.112582565759475e-05,
+      "loss": 1.5587,
+      "step": 357700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.112039727104151e-05,
+      "loss": 1.5261,
+      "step": 357800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.111496888448828e-05,
+      "loss": 1.5192,
+      "step": 357900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.110954049793504e-05,
+      "loss": 1.5383,
+      "step": 358000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.110411211138181e-05,
+      "loss": 1.5502,
+      "step": 358100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.109868372482858e-05,
+      "loss": 1.5433,
+      "step": 358200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.109325533827534e-05,
+      "loss": 1.5171,
+      "step": 358300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.10878269517221e-05,
+      "loss": 1.5241,
+      "step": 358400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.108239856516887e-05,
+      "loss": 1.5281,
+      "step": 358500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.107697017861563e-05,
+      "loss": 1.5371,
+      "step": 358600
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.10715417920624e-05,
+      "loss": 1.5413,
+      "step": 358700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.106611340550916e-05,
+      "loss": 1.509,
+      "step": 358800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.1060685018955924e-05,
+      "loss": 1.5464,
+      "step": 358900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.105525663240269e-05,
+      "loss": 1.5556,
+      "step": 359000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.1049828245849455e-05,
+      "loss": 1.5318,
+      "step": 359100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.104439985929622e-05,
+      "loss": 1.5198,
+      "step": 359200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.103897147274299e-05,
+      "loss": 1.529,
+      "step": 359300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.1033543086189754e-05,
+      "loss": 1.5501,
+      "step": 359400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.1028114699636516e-05,
+      "loss": 1.5562,
+      "step": 359500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.1022686313083285e-05,
+      "loss": 1.5365,
+      "step": 359600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.1017257926530046e-05,
+      "loss": 1.5487,
+      "step": 359700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.101182953997681e-05,
+      "loss": 1.5359,
+      "step": 359800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.100640115342358e-05,
+      "loss": 1.5248,
+      "step": 359900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.100097276687034e-05,
+      "loss": 1.5636,
+      "step": 360000
+    },
+    {
+      "epoch": 1.55,
+      "eval_loss": 1.4448529481887817,
+      "eval_runtime": 17.7635,
+      "eval_samples_per_second": 562.952,
+      "eval_steps_per_second": 17.62,
+      "step": 360000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.099554438031711e-05,
+      "loss": 1.5315,
+      "step": 360100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.099011599376387e-05,
+      "loss": 1.5473,
+      "step": 360200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.098468760721064e-05,
+      "loss": 1.5161,
+      "step": 360300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.09792592206574e-05,
+      "loss": 1.5373,
+      "step": 360400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.097383083410417e-05,
+      "loss": 1.5096,
+      "step": 360500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.096840244755093e-05,
+      "loss": 1.5239,
+      "step": 360600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.09629740609977e-05,
+      "loss": 1.563,
+      "step": 360700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.095754567444446e-05,
+      "loss": 1.5418,
+      "step": 360800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.095211728789122e-05,
+      "loss": 1.5406,
+      "step": 360900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.094668890133799e-05,
+      "loss": 1.5123,
+      "step": 361000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0941260514784753e-05,
+      "loss": 1.5302,
+      "step": 361100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0935832128231515e-05,
+      "loss": 1.5363,
+      "step": 361200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0930403741678284e-05,
+      "loss": 1.5542,
+      "step": 361300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.092497535512505e-05,
+      "loss": 1.5293,
+      "step": 361400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0919546968571814e-05,
+      "loss": 1.5282,
+      "step": 361500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.091411858201858e-05,
+      "loss": 1.5087,
+      "step": 361600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0908690195465345e-05,
+      "loss": 1.5276,
+      "step": 361700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.090326180891211e-05,
+      "loss": 1.5427,
+      "step": 361800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0897833422358876e-05,
+      "loss": 1.5427,
+      "step": 361900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.089240503580564e-05,
+      "loss": 1.5539,
+      "step": 362000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0886976649252406e-05,
+      "loss": 1.539,
+      "step": 362100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.088154826269917e-05,
+      "loss": 1.5205,
+      "step": 362200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.087611987614593e-05,
+      "loss": 1.549,
+      "step": 362300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.08706914895927e-05,
+      "loss": 1.5274,
+      "step": 362400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.086526310303947e-05,
+      "loss": 1.5495,
+      "step": 362500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.085983471648623e-05,
+      "loss": 1.5387,
+      "step": 362600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0854406329933e-05,
+      "loss": 1.5476,
+      "step": 362700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.084897794337976e-05,
+      "loss": 1.5327,
+      "step": 362800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.084354955682652e-05,
+      "loss": 1.5387,
+      "step": 362900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.083812117027329e-05,
+      "loss": 1.5682,
+      "step": 363000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.083269278372005e-05,
+      "loss": 1.5104,
+      "step": 363100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0827264397166814e-05,
+      "loss": 1.5487,
+      "step": 363200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.082183601061358e-05,
+      "loss": 1.534,
+      "step": 363300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.081640762406035e-05,
+      "loss": 1.5264,
+      "step": 363400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.081097923750711e-05,
+      "loss": 1.5203,
+      "step": 363500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.080555085095388e-05,
+      "loss": 1.5561,
+      "step": 363600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0800122464400643e-05,
+      "loss": 1.5083,
+      "step": 363700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0794694077847405e-05,
+      "loss": 1.5377,
+      "step": 363800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0789265691294174e-05,
+      "loss": 1.5362,
+      "step": 363900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0783837304740936e-05,
+      "loss": 1.548,
+      "step": 364000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.07784089181877e-05,
+      "loss": 1.5398,
+      "step": 364100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0772980531634466e-05,
+      "loss": 1.5074,
+      "step": 364200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.076755214508123e-05,
+      "loss": 1.5492,
+      "step": 364300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.0762123758528e-05,
+      "loss": 1.5581,
+      "step": 364400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.0756695371974766e-05,
+      "loss": 1.5623,
+      "step": 364500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.075126698542153e-05,
+      "loss": 1.5235,
+      "step": 364600
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.0745838598868296e-05,
+      "loss": 1.5464,
+      "step": 364700
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.074041021231506e-05,
+      "loss": 1.5277,
+      "step": 364800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.073498182576182e-05,
+      "loss": 1.5038,
+      "step": 364900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.072955343920859e-05,
+      "loss": 1.5474,
+      "step": 365000
+    },
+    {
+      "epoch": 1.57,
+      "eval_loss": 1.447064995765686,
+      "eval_runtime": 17.8278,
+      "eval_samples_per_second": 560.922,
+      "eval_steps_per_second": 17.557,
+      "step": 365000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.072412505265535e-05,
+      "loss": 1.5234,
+      "step": 365100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.071869666610211e-05,
+      "loss": 1.5393,
+      "step": 365200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.071326827954888e-05,
+      "loss": 1.5358,
+      "step": 365300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.070783989299564e-05,
+      "loss": 1.5163,
+      "step": 365400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.070241150644241e-05,
+      "loss": 1.5246,
+      "step": 365500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.069698311988918e-05,
+      "loss": 1.5233,
+      "step": 365600
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.069155473333594e-05,
+      "loss": 1.5309,
+      "step": 365700
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.0686126346782704e-05,
+      "loss": 1.5457,
+      "step": 365800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.068069796022947e-05,
+      "loss": 1.5595,
+      "step": 365900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.0675269573676234e-05,
+      "loss": 1.5376,
+      "step": 366000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.0669841187122996e-05,
+      "loss": 1.548,
+      "step": 366100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.0664412800569765e-05,
+      "loss": 1.5257,
+      "step": 366200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.065898441401653e-05,
+      "loss": 1.5412,
+      "step": 366300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.0653556027463295e-05,
+      "loss": 1.5206,
+      "step": 366400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.064812764091006e-05,
+      "loss": 1.5396,
+      "step": 366500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.0642699254356826e-05,
+      "loss": 1.5389,
+      "step": 366600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.0637270867803595e-05,
+      "loss": 1.5348,
+      "step": 366700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.0631842481250356e-05,
+      "loss": 1.5322,
+      "step": 366800
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.062641409469712e-05,
+      "loss": 1.5203,
+      "step": 366900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.062098570814389e-05,
+      "loss": 1.573,
+      "step": 367000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.061555732159065e-05,
+      "loss": 1.53,
+      "step": 367100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.061012893503741e-05,
+      "loss": 1.5678,
+      "step": 367200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.060470054848418e-05,
+      "loss": 1.5372,
+      "step": 367300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.059927216193094e-05,
+      "loss": 1.5358,
+      "step": 367400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.05938437753777e-05,
+      "loss": 1.5054,
+      "step": 367500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.058841538882448e-05,
+      "loss": 1.5148,
+      "step": 367600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.058298700227124e-05,
+      "loss": 1.5508,
+      "step": 367700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.0577558615718e-05,
+      "loss": 1.5353,
+      "step": 367800
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.057213022916477e-05,
+      "loss": 1.5434,
+      "step": 367900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.056670184261153e-05,
+      "loss": 1.5367,
+      "step": 368000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.0561273456058295e-05,
+      "loss": 1.5396,
+      "step": 368100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.055584506950506e-05,
+      "loss": 1.51,
+      "step": 368200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.0550416682951825e-05,
+      "loss": 1.5025,
+      "step": 368300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.0544988296398594e-05,
+      "loss": 1.5592,
+      "step": 368400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.0539559909845356e-05,
+      "loss": 1.5472,
+      "step": 368500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.053413152329212e-05,
+      "loss": 1.5681,
+      "step": 368600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.052870313673889e-05,
+      "loss": 1.5353,
+      "step": 368700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.0523274750185655e-05,
+      "loss": 1.5521,
+      "step": 368800
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.051784636363242e-05,
+      "loss": 1.5408,
+      "step": 368900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0512417977079182e-05,
+      "loss": 1.5242,
+      "step": 369000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0506989590525947e-05,
+      "loss": 1.5627,
+      "step": 369100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0501561203972713e-05,
+      "loss": 1.5177,
+      "step": 369200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0496132817419474e-05,
+      "loss": 1.5447,
+      "step": 369300
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.049070443086624e-05,
+      "loss": 1.5235,
+      "step": 369400
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0485276044313005e-05,
+      "loss": 1.5045,
+      "step": 369500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.047984765775977e-05,
+      "loss": 1.5419,
+      "step": 369600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.047441927120654e-05,
+      "loss": 1.514,
+      "step": 369700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0468990884653304e-05,
+      "loss": 1.5397,
+      "step": 369800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.046356249810007e-05,
+      "loss": 1.5312,
+      "step": 369900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.045813411154683e-05,
+      "loss": 1.547,
+      "step": 370000
+    },
+    {
+      "epoch": 1.59,
+      "eval_loss": 1.443129301071167,
+      "eval_runtime": 17.8114,
+      "eval_samples_per_second": 561.438,
+      "eval_steps_per_second": 17.573,
+      "step": 370000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0452705724993597e-05,
+      "loss": 1.5362,
+      "step": 370100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0447277338440362e-05,
+      "loss": 1.5216,
+      "step": 370200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0441848951887124e-05,
+      "loss": 1.542,
+      "step": 370300
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.043642056533389e-05,
+      "loss": 1.5317,
+      "step": 370400
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0430992178780654e-05,
+      "loss": 1.5284,
+      "step": 370500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.042556379222742e-05,
+      "loss": 1.527,
+      "step": 370600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.042013540567418e-05,
+      "loss": 1.5274,
+      "step": 370700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0414707019120953e-05,
+      "loss": 1.5469,
+      "step": 370800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.040927863256772e-05,
+      "loss": 1.5412,
+      "step": 370900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.040385024601448e-05,
+      "loss": 1.5134,
+      "step": 371000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0398421859461246e-05,
+      "loss": 1.4993,
+      "step": 371100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.039299347290801e-05,
+      "loss": 1.5266,
+      "step": 371200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0387565086354773e-05,
+      "loss": 1.5317,
+      "step": 371300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0382136699801538e-05,
+      "loss": 1.5536,
+      "step": 371400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0376708313248304e-05,
+      "loss": 1.5476,
+      "step": 371500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.037127992669507e-05,
+      "loss": 1.566,
+      "step": 371600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.036585154014183e-05,
+      "loss": 1.5351,
+      "step": 371700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0360423153588603e-05,
+      "loss": 1.5477,
+      "step": 371800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0354994767035368e-05,
+      "loss": 1.5356,
+      "step": 371900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.034956638048213e-05,
+      "loss": 1.5352,
+      "step": 372000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0344137993928895e-05,
+      "loss": 1.5359,
+      "step": 372100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.033870960737566e-05,
+      "loss": 1.536,
+      "step": 372200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0333281220822422e-05,
+      "loss": 1.5461,
+      "step": 372300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0327852834269187e-05,
+      "loss": 1.5335,
+      "step": 372400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0322424447715953e-05,
+      "loss": 1.5377,
+      "step": 372500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0316996061162718e-05,
+      "loss": 1.5455,
+      "step": 372600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.031156767460948e-05,
+      "loss": 1.5413,
+      "step": 372700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0306139288056245e-05,
+      "loss": 1.5349,
+      "step": 372800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0300710901503017e-05,
+      "loss": 1.5368,
+      "step": 372900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.029528251494978e-05,
+      "loss": 1.517,
+      "step": 373000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0289854128396544e-05,
+      "loss": 1.5187,
+      "step": 373100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.028442574184331e-05,
+      "loss": 1.5592,
+      "step": 373200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.027899735529007e-05,
+      "loss": 1.55,
+      "step": 373300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0273568968736837e-05,
+      "loss": 1.5557,
+      "step": 373400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0268140582183602e-05,
+      "loss": 1.5562,
+      "step": 373500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0262712195630367e-05,
+      "loss": 1.5256,
+      "step": 373600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.025728380907713e-05,
+      "loss": 1.5164,
+      "step": 373700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0251855422523894e-05,
+      "loss": 1.5493,
+      "step": 373800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0246427035970663e-05,
+      "loss": 1.4844,
+      "step": 373900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.024099864941743e-05,
+      "loss": 1.527,
+      "step": 374000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0235570262864194e-05,
+      "loss": 1.5316,
+      "step": 374100
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.023014187631096e-05,
+      "loss": 1.5353,
+      "step": 374200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.022471348975772e-05,
+      "loss": 1.5089,
+      "step": 374300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0219285103204486e-05,
+      "loss": 1.5295,
+      "step": 374400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.021385671665125e-05,
+      "loss": 1.5455,
+      "step": 374500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0208428330098016e-05,
+      "loss": 1.5158,
+      "step": 374600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.020299994354478e-05,
+      "loss": 1.5624,
+      "step": 374700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0197571556991544e-05,
+      "loss": 1.5486,
+      "step": 374800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.019214317043831e-05,
+      "loss": 1.5279,
+      "step": 374900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0186714783885078e-05,
+      "loss": 1.545,
+      "step": 375000
+    },
+    {
+      "epoch": 1.61,
+      "eval_loss": 1.44380521774292,
+      "eval_runtime": 17.8282,
+      "eval_samples_per_second": 560.909,
+      "eval_steps_per_second": 17.556,
+      "step": 375000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0181286397331843e-05,
+      "loss": 1.5086,
+      "step": 375100
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0175858010778608e-05,
+      "loss": 1.5125,
+      "step": 375200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.017042962422537e-05,
+      "loss": 1.5614,
+      "step": 375300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0165001237672135e-05,
+      "loss": 1.5253,
+      "step": 375400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.01595728511189e-05,
+      "loss": 1.5391,
+      "step": 375500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0154144464565666e-05,
+      "loss": 1.5503,
+      "step": 375600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0148716078012428e-05,
+      "loss": 1.5573,
+      "step": 375700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0143287691459193e-05,
+      "loss": 1.5312,
+      "step": 375800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.0137859304905958e-05,
+      "loss": 1.5472,
+      "step": 375900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0132430918352727e-05,
+      "loss": 1.5398,
+      "step": 376000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0127002531799492e-05,
+      "loss": 1.5737,
+      "step": 376100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0121574145246257e-05,
+      "loss": 1.5449,
+      "step": 376200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.011614575869302e-05,
+      "loss": 1.5374,
+      "step": 376300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0110717372139784e-05,
+      "loss": 1.5009,
+      "step": 376400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.010528898558655e-05,
+      "loss": 1.5409,
+      "step": 376500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0099860599033315e-05,
+      "loss": 1.5352,
+      "step": 376600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0094432212480077e-05,
+      "loss": 1.5282,
+      "step": 376700
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0089003825926842e-05,
+      "loss": 1.5344,
+      "step": 376800
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0083575439373607e-05,
+      "loss": 1.5274,
+      "step": 376900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0078147052820373e-05,
+      "loss": 1.5241,
+      "step": 377000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.007271866626714e-05,
+      "loss": 1.5125,
+      "step": 377100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0067290279713907e-05,
+      "loss": 1.5378,
+      "step": 377200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.006186189316067e-05,
+      "loss": 1.5027,
+      "step": 377300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0056433506607434e-05,
+      "loss": 1.5362,
+      "step": 377400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.00510051200542e-05,
+      "loss": 1.5468,
+      "step": 377500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0045576733500964e-05,
+      "loss": 1.5328,
+      "step": 377600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0040148346947726e-05,
+      "loss": 1.553,
+      "step": 377700
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.003471996039449e-05,
+      "loss": 1.5313,
+      "step": 377800
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0029291573841257e-05,
+      "loss": 1.5494,
+      "step": 377900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0023863187288022e-05,
+      "loss": 1.5392,
+      "step": 378000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.001843480073479e-05,
+      "loss": 1.5492,
+      "step": 378100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.0013006414181556e-05,
+      "loss": 1.5513,
+      "step": 378200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.0007578027628318e-05,
+      "loss": 1.5523,
+      "step": 378300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.0002149641075083e-05,
+      "loss": 1.5254,
+      "step": 378400
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9996721254521848e-05,
+      "loss": 1.5396,
+      "step": 378500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9991292867968613e-05,
+      "loss": 1.5326,
+      "step": 378600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9985864481415375e-05,
+      "loss": 1.5232,
+      "step": 378700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.998043609486214e-05,
+      "loss": 1.5402,
+      "step": 378800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9975007708308906e-05,
+      "loss": 1.5314,
+      "step": 378900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.996957932175567e-05,
+      "loss": 1.5268,
+      "step": 379000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9964150935202433e-05,
+      "loss": 1.5378,
+      "step": 379100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9958722548649205e-05,
+      "loss": 1.5408,
+      "step": 379200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9953294162095967e-05,
+      "loss": 1.504,
+      "step": 379300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9947865775542732e-05,
+      "loss": 1.5343,
+      "step": 379400
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9942437388989497e-05,
+      "loss": 1.5036,
+      "step": 379500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9937009002436263e-05,
+      "loss": 1.5604,
+      "step": 379600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9931580615883025e-05,
+      "loss": 1.5606,
+      "step": 379700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.992615222932979e-05,
+      "loss": 1.5496,
+      "step": 379800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9920723842776555e-05,
+      "loss": 1.541,
+      "step": 379900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.991529545622332e-05,
+      "loss": 1.519,
+      "step": 380000
+    },
+    {
+      "epoch": 1.63,
+      "eval_loss": 1.441465973854065,
+      "eval_runtime": 17.8531,
+      "eval_samples_per_second": 560.125,
+      "eval_steps_per_second": 17.532,
+      "step": 380000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9909867069670082e-05,
+      "loss": 1.5429,
+      "step": 380100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9904438683116854e-05,
+      "loss": 1.5333,
+      "step": 380200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9899010296563616e-05,
+      "loss": 1.5395,
+      "step": 380300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.989358191001038e-05,
+      "loss": 1.5221,
+      "step": 380400
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9888153523457147e-05,
+      "loss": 1.5362,
+      "step": 380500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9882725136903912e-05,
+      "loss": 1.5406,
+      "step": 380600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9877296750350674e-05,
+      "loss": 1.5323,
+      "step": 380700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.987186836379744e-05,
+      "loss": 1.5644,
+      "step": 380800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9866439977244204e-05,
+      "loss": 1.5275,
+      "step": 380900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.986101159069097e-05,
+      "loss": 1.5485,
+      "step": 381000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.985558320413773e-05,
+      "loss": 1.5074,
+      "step": 381100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9850154817584497e-05,
+      "loss": 1.5068,
+      "step": 381200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9844726431031265e-05,
+      "loss": 1.5295,
+      "step": 381300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.983929804447803e-05,
+      "loss": 1.5248,
+      "step": 381400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9833869657924796e-05,
+      "loss": 1.5372,
+      "step": 381500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.982844127137156e-05,
+      "loss": 1.5467,
+      "step": 381600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9823012884818323e-05,
+      "loss": 1.508,
+      "step": 381700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.981758449826509e-05,
+      "loss": 1.545,
+      "step": 381800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9812156111711854e-05,
+      "loss": 1.5113,
+      "step": 381900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.980672772515862e-05,
+      "loss": 1.5503,
+      "step": 382000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.980129933860538e-05,
+      "loss": 1.5146,
+      "step": 382100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9795870952052146e-05,
+      "loss": 1.5072,
+      "step": 382200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9790442565498915e-05,
+      "loss": 1.5394,
+      "step": 382300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.978501417894568e-05,
+      "loss": 1.5449,
+      "step": 382400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9779585792392445e-05,
+      "loss": 1.5293,
+      "step": 382500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.977415740583921e-05,
+      "loss": 1.5363,
+      "step": 382600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9768729019285972e-05,
+      "loss": 1.5216,
+      "step": 382700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9763300632732738e-05,
+      "loss": 1.5477,
+      "step": 382800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9757872246179503e-05,
+      "loss": 1.5265,
+      "step": 382900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.9752443859626268e-05,
+      "loss": 1.4951,
+      "step": 383000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.974701547307303e-05,
+      "loss": 1.5332,
+      "step": 383100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.9741587086519795e-05,
+      "loss": 1.5358,
+      "step": 383200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.9736158699966564e-05,
+      "loss": 1.5612,
+      "step": 383300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.973073031341333e-05,
+      "loss": 1.5337,
+      "step": 383400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.9725301926860094e-05,
+      "loss": 1.5483,
+      "step": 383500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.971987354030686e-05,
+      "loss": 1.503,
+      "step": 383600
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.971444515375362e-05,
+      "loss": 1.555,
+      "step": 383700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.9709016767200387e-05,
+      "loss": 1.5329,
+      "step": 383800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.9703588380647152e-05,
+      "loss": 1.5388,
+      "step": 383900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.9698159994093917e-05,
+      "loss": 1.5073,
+      "step": 384000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.969273160754068e-05,
+      "loss": 1.5039,
+      "step": 384100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.9687303220987444e-05,
+      "loss": 1.5433,
+      "step": 384200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.968187483443421e-05,
+      "loss": 1.5517,
+      "step": 384300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.967644644788098e-05,
+      "loss": 1.5244,
+      "step": 384400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.9671018061327744e-05,
+      "loss": 1.5426,
+      "step": 384500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.966558967477451e-05,
+      "loss": 1.5364,
+      "step": 384600
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.966016128822127e-05,
+      "loss": 1.561,
+      "step": 384700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.9654732901668036e-05,
+      "loss": 1.5231,
+      "step": 384800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.96493045151148e-05,
+      "loss": 1.5599,
+      "step": 384900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.9643876128561567e-05,
+      "loss": 1.5445,
+      "step": 385000
+    },
+    {
+      "epoch": 1.65,
+      "eval_loss": 1.4401415586471558,
+      "eval_runtime": 17.7618,
+      "eval_samples_per_second": 563.005,
+      "eval_steps_per_second": 17.622,
+      "step": 385000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.963844774200833e-05,
+      "loss": 1.5467,
+      "step": 385100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.9633019355455094e-05,
+      "loss": 1.5363,
+      "step": 385200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.962759096890186e-05,
+      "loss": 1.5448,
+      "step": 385300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9622162582348628e-05,
+      "loss": 1.5497,
+      "step": 385400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9616734195795393e-05,
+      "loss": 1.5068,
+      "step": 385500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9611305809242158e-05,
+      "loss": 1.5253,
+      "step": 385600
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.960587742268892e-05,
+      "loss": 1.5221,
+      "step": 385700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9600449036135685e-05,
+      "loss": 1.5336,
+      "step": 385800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.959502064958245e-05,
+      "loss": 1.5336,
+      "step": 385900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9589592263029216e-05,
+      "loss": 1.5239,
+      "step": 386000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9584163876475978e-05,
+      "loss": 1.5261,
+      "step": 386100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9578735489922743e-05,
+      "loss": 1.5146,
+      "step": 386200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9573307103369508e-05,
+      "loss": 1.5527,
+      "step": 386300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9567878716816274e-05,
+      "loss": 1.5197,
+      "step": 386400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9562450330263042e-05,
+      "loss": 1.5463,
+      "step": 386500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9557021943709807e-05,
+      "loss": 1.5406,
+      "step": 386600
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.955159355715657e-05,
+      "loss": 1.536,
+      "step": 386700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9546165170603335e-05,
+      "loss": 1.5257,
+      "step": 386800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.95407367840501e-05,
+      "loss": 1.5449,
+      "step": 386900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9535308397496865e-05,
+      "loss": 1.5352,
+      "step": 387000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9529880010943627e-05,
+      "loss": 1.5447,
+      "step": 387100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9524451624390392e-05,
+      "loss": 1.5273,
+      "step": 387200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9519023237837157e-05,
+      "loss": 1.515,
+      "step": 387300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9513594851283923e-05,
+      "loss": 1.5312,
+      "step": 387400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.950816646473069e-05,
+      "loss": 1.5419,
+      "step": 387500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9502738078177457e-05,
+      "loss": 1.5344,
+      "step": 387600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.949730969162422e-05,
+      "loss": 1.5083,
+      "step": 387700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9491881305070984e-05,
+      "loss": 1.5281,
+      "step": 387800
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.948645291851775e-05,
+      "loss": 1.5292,
+      "step": 387900
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9481024531964514e-05,
+      "loss": 1.5412,
+      "step": 388000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9475596145411276e-05,
+      "loss": 1.5267,
+      "step": 388100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.947016775885804e-05,
+      "loss": 1.5315,
+      "step": 388200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9464739372304807e-05,
+      "loss": 1.5206,
+      "step": 388300
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9459310985751572e-05,
+      "loss": 1.5419,
+      "step": 388400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9453882599198334e-05,
+      "loss": 1.5717,
+      "step": 388500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9448454212645106e-05,
+      "loss": 1.5388,
+      "step": 388600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9443025826091868e-05,
+      "loss": 1.5255,
+      "step": 388700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9437597439538633e-05,
+      "loss": 1.5424,
+      "step": 388800
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.94321690529854e-05,
+      "loss": 1.5145,
+      "step": 388900
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9426740666432164e-05,
+      "loss": 1.5352,
+      "step": 389000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9421312279878925e-05,
+      "loss": 1.5473,
+      "step": 389100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.941588389332569e-05,
+      "loss": 1.5378,
+      "step": 389200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9410455506772456e-05,
+      "loss": 1.5358,
+      "step": 389300
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.940502712021922e-05,
+      "loss": 1.5273,
+      "step": 389400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9399598733665983e-05,
+      "loss": 1.5294,
+      "step": 389500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9394170347112755e-05,
+      "loss": 1.5519,
+      "step": 389600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9388741960559517e-05,
+      "loss": 1.5548,
+      "step": 389700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9383313574006282e-05,
+      "loss": 1.5096,
+      "step": 389800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9377885187453048e-05,
+      "loss": 1.5136,
+      "step": 389900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9372456800899813e-05,
+      "loss": 1.5348,
+      "step": 390000
+    },
+    {
+      "epoch": 1.68,
+      "eval_loss": 1.4426006078720093,
+      "eval_runtime": 36.5743,
+      "eval_samples_per_second": 273.416,
+      "eval_steps_per_second": 8.558,
+      "step": 390000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9367028414346575e-05,
+      "loss": 1.5546,
+      "step": 390100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.936160002779334e-05,
+      "loss": 1.5413,
+      "step": 390200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9356171641240105e-05,
+      "loss": 1.5366,
+      "step": 390300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.935074325468687e-05,
+      "loss": 1.5408,
+      "step": 390400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9345314868133632e-05,
+      "loss": 1.53,
+      "step": 390500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9339886481580398e-05,
+      "loss": 1.4864,
+      "step": 390600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9334458095027166e-05,
+      "loss": 1.5162,
+      "step": 390700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.932902970847393e-05,
+      "loss": 1.5158,
+      "step": 390800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9323601321920697e-05,
+      "loss": 1.5527,
+      "step": 390900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9318172935367462e-05,
+      "loss": 1.5186,
+      "step": 391000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9312744548814224e-05,
+      "loss": 1.5328,
+      "step": 391100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.930731616226099e-05,
+      "loss": 1.5606,
+      "step": 391200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9301887775707754e-05,
+      "loss": 1.5256,
+      "step": 391300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.929645938915452e-05,
+      "loss": 1.5284,
+      "step": 391400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.929103100260128e-05,
+      "loss": 1.5403,
+      "step": 391500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9285602616048047e-05,
+      "loss": 1.5277,
+      "step": 391600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9280174229494816e-05,
+      "loss": 1.5158,
+      "step": 391700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.927474584294158e-05,
+      "loss": 1.5156,
+      "step": 391800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9269317456388346e-05,
+      "loss": 1.519,
+      "step": 391900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.926388906983511e-05,
+      "loss": 1.5105,
+      "step": 392000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9258460683281873e-05,
+      "loss": 1.4996,
+      "step": 392100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.925303229672864e-05,
+      "loss": 1.538,
+      "step": 392200
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9247603910175404e-05,
+      "loss": 1.5354,
+      "step": 392300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9242175523622166e-05,
+      "loss": 1.5445,
+      "step": 392400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.923674713706893e-05,
+      "loss": 1.5375,
+      "step": 392500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9231318750515696e-05,
+      "loss": 1.5331,
+      "step": 392600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.922589036396246e-05,
+      "loss": 1.5468,
+      "step": 392700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.922046197740923e-05,
+      "loss": 1.5413,
+      "step": 392800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9215033590855995e-05,
+      "loss": 1.5525,
+      "step": 392900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.920960520430276e-05,
+      "loss": 1.5149,
+      "step": 393000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9204176817749522e-05,
+      "loss": 1.5498,
+      "step": 393100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9198748431196288e-05,
+      "loss": 1.5412,
+      "step": 393200
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9193320044643053e-05,
+      "loss": 1.5368,
+      "step": 393300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9187891658089815e-05,
+      "loss": 1.542,
+      "step": 393400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.918246327153658e-05,
+      "loss": 1.5243,
+      "step": 393500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9177034884983345e-05,
+      "loss": 1.5567,
+      "step": 393600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.917160649843011e-05,
+      "loss": 1.5155,
+      "step": 393700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.916617811187688e-05,
+      "loss": 1.5208,
+      "step": 393800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9160749725323645e-05,
+      "loss": 1.5428,
+      "step": 393900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.915532133877041e-05,
+      "loss": 1.5396,
+      "step": 394000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.914989295221717e-05,
+      "loss": 1.5331,
+      "step": 394100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9144464565663937e-05,
+      "loss": 1.5322,
+      "step": 394200
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9139036179110702e-05,
+      "loss": 1.57,
+      "step": 394300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.9133607792557464e-05,
+      "loss": 1.5523,
+      "step": 394400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.912817940600423e-05,
+      "loss": 1.515,
+      "step": 394500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9122751019450995e-05,
+      "loss": 1.5288,
+      "step": 394600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.911732263289776e-05,
+      "loss": 1.5688,
+      "step": 394700
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9111894246344522e-05,
+      "loss": 1.5503,
+      "step": 394800
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9106465859791294e-05,
+      "loss": 1.5064,
+      "step": 394900
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.910103747323806e-05,
+      "loss": 1.525,
+      "step": 395000
+    },
+    {
+      "epoch": 1.7,
+      "eval_loss": 1.441019892692566,
+      "eval_runtime": 17.8041,
+      "eval_samples_per_second": 561.668,
+      "eval_steps_per_second": 17.58,
+      "step": 395000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.909560908668482e-05,
+      "loss": 1.5334,
+      "step": 395100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9090180700131586e-05,
+      "loss": 1.5313,
+      "step": 395200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.908475231357835e-05,
+      "loss": 1.5513,
+      "step": 395300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9079323927025113e-05,
+      "loss": 1.5522,
+      "step": 395400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.907389554047188e-05,
+      "loss": 1.5336,
+      "step": 395500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9068467153918644e-05,
+      "loss": 1.5244,
+      "step": 395600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.906303876736541e-05,
+      "loss": 1.5485,
+      "step": 395700
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.905761038081217e-05,
+      "loss": 1.562,
+      "step": 395800
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9052181994258943e-05,
+      "loss": 1.5201,
+      "step": 395900
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9046753607705708e-05,
+      "loss": 1.5629,
+      "step": 396000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.904132522115247e-05,
+      "loss": 1.5387,
+      "step": 396100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9035896834599235e-05,
+      "loss": 1.5281,
+      "step": 396200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9030468448046e-05,
+      "loss": 1.5412,
+      "step": 396300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9025040061492763e-05,
+      "loss": 1.5418,
+      "step": 396400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9019611674939528e-05,
+      "loss": 1.5426,
+      "step": 396500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9014183288386293e-05,
+      "loss": 1.5182,
+      "step": 396600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.900875490183306e-05,
+      "loss": 1.5376,
+      "step": 396700
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.900332651527982e-05,
+      "loss": 1.5392,
+      "step": 396800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8997898128726585e-05,
+      "loss": 1.5274,
+      "step": 396900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8992469742173358e-05,
+      "loss": 1.5162,
+      "step": 397000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.898704135562012e-05,
+      "loss": 1.5394,
+      "step": 397100
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8981612969066885e-05,
+      "loss": 1.5618,
+      "step": 397200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.897618458251365e-05,
+      "loss": 1.5106,
+      "step": 397300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8970756195960412e-05,
+      "loss": 1.5487,
+      "step": 397400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8965327809407177e-05,
+      "loss": 1.5158,
+      "step": 397500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8959899422853942e-05,
+      "loss": 1.5026,
+      "step": 397600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8954471036300708e-05,
+      "loss": 1.5133,
+      "step": 397700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.894904264974747e-05,
+      "loss": 1.5608,
+      "step": 397800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8943614263194235e-05,
+      "loss": 1.5655,
+      "step": 397900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8938185876641007e-05,
+      "loss": 1.5126,
+      "step": 398000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.893275749008777e-05,
+      "loss": 1.556,
+      "step": 398100
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8927329103534534e-05,
+      "loss": 1.5335,
+      "step": 398200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.89219007169813e-05,
+      "loss": 1.5421,
+      "step": 398300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.891647233042806e-05,
+      "loss": 1.5166,
+      "step": 398400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8911043943874826e-05,
+      "loss": 1.5319,
+      "step": 398500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.890561555732159e-05,
+      "loss": 1.5482,
+      "step": 398600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8900187170768357e-05,
+      "loss": 1.5344,
+      "step": 398700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.889475878421512e-05,
+      "loss": 1.5221,
+      "step": 398800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8889330397661884e-05,
+      "loss": 1.5241,
+      "step": 398900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.888390201110865e-05,
+      "loss": 1.5421,
+      "step": 399000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8878473624555418e-05,
+      "loss": 1.5389,
+      "step": 399100
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8873045238002183e-05,
+      "loss": 1.5364,
+      "step": 399200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.886761685144895e-05,
+      "loss": 1.5273,
+      "step": 399300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.886218846489571e-05,
+      "loss": 1.5398,
+      "step": 399400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8856760078342476e-05,
+      "loss": 1.5343,
+      "step": 399500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.885133169178924e-05,
+      "loss": 1.5356,
+      "step": 399600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8845903305236006e-05,
+      "loss": 1.5525,
+      "step": 399700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8840474918682768e-05,
+      "loss": 1.539,
+      "step": 399800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8835046532129533e-05,
+      "loss": 1.5196,
+      "step": 399900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.88296181455763e-05,
+      "loss": 1.5199,
+      "step": 400000
+    },
+    {
+      "epoch": 1.72,
+      "eval_loss": 1.4402216672897339,
+      "eval_runtime": 17.7998,
+      "eval_samples_per_second": 561.804,
+      "eval_steps_per_second": 17.584,
+      "step": 400000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8824189759023067e-05,
+      "loss": 1.5239,
+      "step": 400100
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8818761372469832e-05,
+      "loss": 1.553,
+      "step": 400200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8813332985916598e-05,
+      "loss": 1.4987,
+      "step": 400300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.880790459936336e-05,
+      "loss": 1.5189,
+      "step": 400400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8802476212810125e-05,
+      "loss": 1.5222,
+      "step": 400500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.879704782625689e-05,
+      "loss": 1.5357,
+      "step": 400600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8791619439703655e-05,
+      "loss": 1.5154,
+      "step": 400700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8786191053150417e-05,
+      "loss": 1.5355,
+      "step": 400800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8780762666597182e-05,
+      "loss": 1.5319,
+      "step": 400900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8775334280043948e-05,
+      "loss": 1.5456,
+      "step": 401000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8769905893490713e-05,
+      "loss": 1.5551,
+      "step": 401100
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.876447750693748e-05,
+      "loss": 1.5271,
+      "step": 401200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8759049120384247e-05,
+      "loss": 1.5595,
+      "step": 401300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.875362073383101e-05,
+      "loss": 1.5349,
+      "step": 401400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8748192347277774e-05,
+      "loss": 1.5408,
+      "step": 401500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.874276396072454e-05,
+      "loss": 1.5176,
+      "step": 401600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8737335574171305e-05,
+      "loss": 1.4892,
+      "step": 401700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8731907187618066e-05,
+      "loss": 1.5339,
+      "step": 401800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8726478801064832e-05,
+      "loss": 1.5057,
+      "step": 401900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8721050414511597e-05,
+      "loss": 1.5326,
+      "step": 402000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8715622027958362e-05,
+      "loss": 1.5439,
+      "step": 402100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.871019364140513e-05,
+      "loss": 1.5396,
+      "step": 402200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8704765254851896e-05,
+      "loss": 1.5363,
+      "step": 402300
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8699336868298658e-05,
+      "loss": 1.5016,
+      "step": 402400
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8693908481745423e-05,
+      "loss": 1.522,
+      "step": 402500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.868848009519219e-05,
+      "loss": 1.5246,
+      "step": 402600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8683051708638954e-05,
+      "loss": 1.5332,
+      "step": 402700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8677623322085716e-05,
+      "loss": 1.5089,
+      "step": 402800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.867219493553248e-05,
+      "loss": 1.5207,
+      "step": 402900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8666766548979246e-05,
+      "loss": 1.5402,
+      "step": 403000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.866133816242601e-05,
+      "loss": 1.5388,
+      "step": 403100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8655909775872773e-05,
+      "loss": 1.5333,
+      "step": 403200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8650481389319545e-05,
+      "loss": 1.5238,
+      "step": 403300
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8645053002766307e-05,
+      "loss": 1.4922,
+      "step": 403400
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8639624616213073e-05,
+      "loss": 1.5461,
+      "step": 403500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8634196229659838e-05,
+      "loss": 1.5317,
+      "step": 403600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8628767843106603e-05,
+      "loss": 1.543,
+      "step": 403700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8623339456553365e-05,
+      "loss": 1.5036,
+      "step": 403800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.861791107000013e-05,
+      "loss": 1.5589,
+      "step": 403900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8612482683446895e-05,
+      "loss": 1.5317,
+      "step": 404000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.860705429689366e-05,
+      "loss": 1.5291,
+      "step": 404100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8601625910340423e-05,
+      "loss": 1.5626,
+      "step": 404200
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8596197523787195e-05,
+      "loss": 1.5296,
+      "step": 404300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8590769137233957e-05,
+      "loss": 1.5301,
+      "step": 404400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8585340750680722e-05,
+      "loss": 1.528,
+      "step": 404500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8579912364127487e-05,
+      "loss": 1.5344,
+      "step": 404600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8574483977574252e-05,
+      "loss": 1.5603,
+      "step": 404700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8569055591021014e-05,
+      "loss": 1.5,
+      "step": 404800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.856362720446778e-05,
+      "loss": 1.519,
+      "step": 404900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8558198817914545e-05,
+      "loss": 1.5268,
+      "step": 405000
+    },
+    {
+      "epoch": 1.74,
+      "eval_loss": 1.4383865594863892,
+      "eval_runtime": 17.8138,
+      "eval_samples_per_second": 561.361,
+      "eval_steps_per_second": 17.571,
+      "step": 405000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.855277043136131e-05,
+      "loss": 1.5201,
+      "step": 405100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8547342044808072e-05,
+      "loss": 1.531,
+      "step": 405200
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8541913658254844e-05,
+      "loss": 1.5284,
+      "step": 405300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8536485271701606e-05,
+      "loss": 1.5239,
+      "step": 405400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.853105688514837e-05,
+      "loss": 1.5457,
+      "step": 405500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8525628498595136e-05,
+      "loss": 1.517,
+      "step": 405600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.85202001120419e-05,
+      "loss": 1.5408,
+      "step": 405700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8514771725488663e-05,
+      "loss": 1.5529,
+      "step": 405800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.850934333893543e-05,
+      "loss": 1.5393,
+      "step": 405900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8503914952382194e-05,
+      "loss": 1.5388,
+      "step": 406000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.849848656582896e-05,
+      "loss": 1.5186,
+      "step": 406100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.849305817927572e-05,
+      "loss": 1.5518,
+      "step": 406200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8487629792722486e-05,
+      "loss": 1.543,
+      "step": 406300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8482201406169255e-05,
+      "loss": 1.5168,
+      "step": 406400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.847677301961602e-05,
+      "loss": 1.5481,
+      "step": 406500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8471344633062786e-05,
+      "loss": 1.5403,
+      "step": 406600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.846591624650955e-05,
+      "loss": 1.517,
+      "step": 406700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8460487859956313e-05,
+      "loss": 1.5456,
+      "step": 406800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8455059473403078e-05,
+      "loss": 1.5158,
+      "step": 406900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8449631086849843e-05,
+      "loss": 1.5657,
+      "step": 407000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.844420270029661e-05,
+      "loss": 1.537,
+      "step": 407100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.843877431374337e-05,
+      "loss": 1.5425,
+      "step": 407200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8433345927190136e-05,
+      "loss": 1.4908,
+      "step": 407300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8427917540636904e-05,
+      "loss": 1.5385,
+      "step": 407400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.842248915408367e-05,
+      "loss": 1.5428,
+      "step": 407500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8417060767530435e-05,
+      "loss": 1.5489,
+      "step": 407600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.84116323809772e-05,
+      "loss": 1.5266,
+      "step": 407700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8406203994423962e-05,
+      "loss": 1.524,
+      "step": 407800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8400775607870727e-05,
+      "loss": 1.5025,
+      "step": 407900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8395347221317492e-05,
+      "loss": 1.5191,
+      "step": 408000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8389918834764258e-05,
+      "loss": 1.5252,
+      "step": 408100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.838449044821102e-05,
+      "loss": 1.5154,
+      "step": 408200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8379062061657785e-05,
+      "loss": 1.5204,
+      "step": 408300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.837363367510455e-05,
+      "loss": 1.5414,
+      "step": 408400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.836820528855132e-05,
+      "loss": 1.5467,
+      "step": 408500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.8362776901998084e-05,
+      "loss": 1.5474,
+      "step": 408600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.835734851544485e-05,
+      "loss": 1.5368,
+      "step": 408700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.835192012889161e-05,
+      "loss": 1.5173,
+      "step": 408800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.8346491742338376e-05,
+      "loss": 1.5255,
+      "step": 408900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.834106335578514e-05,
+      "loss": 1.5475,
+      "step": 409000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.8335634969231907e-05,
+      "loss": 1.5244,
+      "step": 409100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.833020658267867e-05,
+      "loss": 1.5163,
+      "step": 409200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.8324778196125434e-05,
+      "loss": 1.532,
+      "step": 409300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.83193498095722e-05,
+      "loss": 1.5529,
+      "step": 409400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.8313921423018968e-05,
+      "loss": 1.5279,
+      "step": 409500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.8308493036465733e-05,
+      "loss": 1.5326,
+      "step": 409600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.83030646499125e-05,
+      "loss": 1.4819,
+      "step": 409700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.829763626335926e-05,
+      "loss": 1.5172,
+      "step": 409800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.8292207876806026e-05,
+      "loss": 1.5296,
+      "step": 409900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.828677949025279e-05,
+      "loss": 1.5574,
+      "step": 410000
+    },
+    {
+      "epoch": 1.76,
+      "eval_loss": 1.4406307935714722,
+      "eval_runtime": 17.8024,
+      "eval_samples_per_second": 561.722,
+      "eval_steps_per_second": 17.582,
+      "step": 410000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.8281351103699556e-05,
+      "loss": 1.5008,
+      "step": 410100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.8275922717146318e-05,
+      "loss": 1.5263,
+      "step": 410200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.8270494330593083e-05,
+      "loss": 1.5134,
+      "step": 410300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.826506594403985e-05,
+      "loss": 1.532,
+      "step": 410400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.8259637557486614e-05,
+      "loss": 1.5147,
+      "step": 410500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.8254209170933383e-05,
+      "loss": 1.4813,
+      "step": 410600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.8248780784380148e-05,
+      "loss": 1.5178,
+      "step": 410700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.824335239782691e-05,
+      "loss": 1.5388,
+      "step": 410800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8237924011273675e-05,
+      "loss": 1.5144,
+      "step": 410900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.823249562472044e-05,
+      "loss": 1.4999,
+      "step": 411000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8227067238167205e-05,
+      "loss": 1.5258,
+      "step": 411100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8221638851613967e-05,
+      "loss": 1.5123,
+      "step": 411200
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8216210465060733e-05,
+      "loss": 1.5162,
+      "step": 411300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8210782078507498e-05,
+      "loss": 1.5132,
+      "step": 411400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8205353691954263e-05,
+      "loss": 1.5276,
+      "step": 411500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8199925305401032e-05,
+      "loss": 1.4997,
+      "step": 411600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8194496918847797e-05,
+      "loss": 1.5132,
+      "step": 411700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.818906853229456e-05,
+      "loss": 1.5267,
+      "step": 411800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8183640145741324e-05,
+      "loss": 1.5039,
+      "step": 411900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.817821175918809e-05,
+      "loss": 1.5142,
+      "step": 412000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8172783372634855e-05,
+      "loss": 1.526,
+      "step": 412100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8167354986081617e-05,
+      "loss": 1.5519,
+      "step": 412200
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8161926599528382e-05,
+      "loss": 1.5155,
+      "step": 412300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8156498212975147e-05,
+      "loss": 1.5072,
+      "step": 412400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8151069826421912e-05,
+      "loss": 1.5312,
+      "step": 412500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8145641439868674e-05,
+      "loss": 1.5432,
+      "step": 412600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8140213053315446e-05,
+      "loss": 1.5491,
+      "step": 412700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8134784666762208e-05,
+      "loss": 1.5052,
+      "step": 412800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8129356280208973e-05,
+      "loss": 1.5212,
+      "step": 412900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.812392789365574e-05,
+      "loss": 1.5209,
+      "step": 413000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.8118499507102504e-05,
+      "loss": 1.5077,
+      "step": 413100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8113071120549266e-05,
+      "loss": 1.4798,
+      "step": 413200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.810764273399603e-05,
+      "loss": 1.5385,
+      "step": 413300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8102214347442796e-05,
+      "loss": 1.5457,
+      "step": 413400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.809678596088956e-05,
+      "loss": 1.523,
+      "step": 413500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8091357574336323e-05,
+      "loss": 1.5384,
+      "step": 413600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8085929187783096e-05,
+      "loss": 1.5193,
+      "step": 413700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8080500801229857e-05,
+      "loss": 1.5332,
+      "step": 413800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8075072414676623e-05,
+      "loss": 1.5462,
+      "step": 413900
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8069644028123388e-05,
+      "loss": 1.5654,
+      "step": 414000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8064215641570153e-05,
+      "loss": 1.5277,
+      "step": 414100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8058787255016915e-05,
+      "loss": 1.485,
+      "step": 414200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.805335886846368e-05,
+      "loss": 1.4895,
+      "step": 414300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8047930481910446e-05,
+      "loss": 1.5262,
+      "step": 414400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.804250209535721e-05,
+      "loss": 1.5392,
+      "step": 414500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8037073708803973e-05,
+      "loss": 1.4971,
+      "step": 414600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8031645322250738e-05,
+      "loss": 1.5281,
+      "step": 414700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8026216935697507e-05,
+      "loss": 1.5242,
+      "step": 414800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8020788549144272e-05,
+      "loss": 1.5357,
+      "step": 414900
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8015360162591037e-05,
+      "loss": 1.5335,
+      "step": 415000
+    },
+    {
+      "epoch": 1.78,
+      "eval_loss": 1.4382429122924805,
+      "eval_runtime": 17.7575,
+      "eval_samples_per_second": 563.143,
+      "eval_steps_per_second": 17.626,
+      "step": 415000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8009931776037802e-05,
+      "loss": 1.5204,
+      "step": 415100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.8004503389484564e-05,
+      "loss": 1.5121,
+      "step": 415200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.799907500293133e-05,
+      "loss": 1.5203,
+      "step": 415300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.7993646616378095e-05,
+      "loss": 1.5089,
+      "step": 415400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.798821822982486e-05,
+      "loss": 1.5304,
+      "step": 415500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7982789843271622e-05,
+      "loss": 1.524,
+      "step": 415600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7977361456718387e-05,
+      "loss": 1.5185,
+      "step": 415700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7971933070165156e-05,
+      "loss": 1.5323,
+      "step": 415800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.796650468361192e-05,
+      "loss": 1.5207,
+      "step": 415900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7961076297058686e-05,
+      "loss": 1.5296,
+      "step": 416000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.795564791050545e-05,
+      "loss": 1.51,
+      "step": 416100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7950219523952214e-05,
+      "loss": 1.5265,
+      "step": 416200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.794479113739898e-05,
+      "loss": 1.5355,
+      "step": 416300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7939362750845744e-05,
+      "loss": 1.5426,
+      "step": 416400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.793393436429251e-05,
+      "loss": 1.5439,
+      "step": 416500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.792850597773927e-05,
+      "loss": 1.5535,
+      "step": 416600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7923077591186036e-05,
+      "loss": 1.516,
+      "step": 416700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7917649204632802e-05,
+      "loss": 1.5376,
+      "step": 416800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.791222081807957e-05,
+      "loss": 1.528,
+      "step": 416900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7906792431526336e-05,
+      "loss": 1.5282,
+      "step": 417000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.79013640449731e-05,
+      "loss": 1.528,
+      "step": 417100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7895935658419863e-05,
+      "loss": 1.5245,
+      "step": 417200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7890507271866628e-05,
+      "loss": 1.5342,
+      "step": 417300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7885078885313393e-05,
+      "loss": 1.5092,
+      "step": 417400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.787965049876016e-05,
+      "loss": 1.5356,
+      "step": 417500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.787422211220692e-05,
+      "loss": 1.5228,
+      "step": 417600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7868793725653686e-05,
+      "loss": 1.5355,
+      "step": 417700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.786336533910045e-05,
+      "loss": 1.5256,
+      "step": 417800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.785793695254722e-05,
+      "loss": 1.4689,
+      "step": 417900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7852508565993985e-05,
+      "loss": 1.5218,
+      "step": 418000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.784708017944075e-05,
+      "loss": 1.5373,
+      "step": 418100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7841651792887512e-05,
+      "loss": 1.5249,
+      "step": 418200
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7836223406334277e-05,
+      "loss": 1.526,
+      "step": 418300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7830795019781043e-05,
+      "loss": 1.5213,
+      "step": 418400
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7825366633227808e-05,
+      "loss": 1.5095,
+      "step": 418500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.781993824667457e-05,
+      "loss": 1.5196,
+      "step": 418600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7814509860121335e-05,
+      "loss": 1.5147,
+      "step": 418700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.78090814735681e-05,
+      "loss": 1.5175,
+      "step": 418800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7803653087014862e-05,
+      "loss": 1.535,
+      "step": 418900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7798224700461634e-05,
+      "loss": 1.4684,
+      "step": 419000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.77927963139084e-05,
+      "loss": 1.5448,
+      "step": 419100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.778736792735516e-05,
+      "loss": 1.5211,
+      "step": 419200
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7781939540801927e-05,
+      "loss": 1.5303,
+      "step": 419300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7776511154248692e-05,
+      "loss": 1.5182,
+      "step": 419400
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7771082767695457e-05,
+      "loss": 1.5291,
+      "step": 419500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.776565438114222e-05,
+      "loss": 1.5256,
+      "step": 419600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7760225994588984e-05,
+      "loss": 1.5461,
+      "step": 419700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.775479760803575e-05,
+      "loss": 1.5385,
+      "step": 419800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.774936922148251e-05,
+      "loss": 1.547,
+      "step": 419900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7743940834929283e-05,
+      "loss": 1.5163,
+      "step": 420000
+    },
+    {
+      "epoch": 1.8,
+      "eval_loss": 1.4372782707214355,
+      "eval_runtime": 17.8039,
+      "eval_samples_per_second": 561.676,
+      "eval_steps_per_second": 17.58,
+      "step": 420000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.773851244837605e-05,
+      "loss": 1.4974,
+      "step": 420100
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.773308406182281e-05,
+      "loss": 1.5112,
+      "step": 420200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7727655675269576e-05,
+      "loss": 1.5071,
+      "step": 420300
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.772222728871634e-05,
+      "loss": 1.5225,
+      "step": 420400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7716798902163106e-05,
+      "loss": 1.5471,
+      "step": 420500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7711370515609868e-05,
+      "loss": 1.5217,
+      "step": 420600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7705942129056633e-05,
+      "loss": 1.4961,
+      "step": 420700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.77005137425034e-05,
+      "loss": 1.5081,
+      "step": 420800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.769508535595016e-05,
+      "loss": 1.5024,
+      "step": 420900
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7689656969396926e-05,
+      "loss": 1.5408,
+      "step": 421000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7684228582843698e-05,
+      "loss": 1.537,
+      "step": 421100
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.767880019629046e-05,
+      "loss": 1.5233,
+      "step": 421200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7673371809737225e-05,
+      "loss": 1.5183,
+      "step": 421300
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.766794342318399e-05,
+      "loss": 1.5317,
+      "step": 421400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7662515036630756e-05,
+      "loss": 1.5374,
+      "step": 421500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7657086650077517e-05,
+      "loss": 1.525,
+      "step": 421600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7651658263524283e-05,
+      "loss": 1.5106,
+      "step": 421700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7646229876971048e-05,
+      "loss": 1.4967,
+      "step": 421800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.764080149041781e-05,
+      "loss": 1.5214,
+      "step": 421900
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7635373103864575e-05,
+      "loss": 1.522,
+      "step": 422000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7629944717311347e-05,
+      "loss": 1.5259,
+      "step": 422100
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.762451633075811e-05,
+      "loss": 1.4915,
+      "step": 422200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7619087944204874e-05,
+      "loss": 1.531,
+      "step": 422300
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.761365955765164e-05,
+      "loss": 1.5303,
+      "step": 422400
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7608231171098405e-05,
+      "loss": 1.505,
+      "step": 422500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7602802784545167e-05,
+      "loss": 1.5445,
+      "step": 422600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7597374397991932e-05,
+      "loss": 1.5211,
+      "step": 422700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7591946011438697e-05,
+      "loss": 1.5107,
+      "step": 422800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.758651762488546e-05,
+      "loss": 1.4987,
+      "step": 422900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7581089238332224e-05,
+      "loss": 1.5144,
+      "step": 423000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.757566085177899e-05,
+      "loss": 1.5307,
+      "step": 423100
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7570232465225758e-05,
+      "loss": 1.5336,
+      "step": 423200
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7564804078672524e-05,
+      "loss": 1.5058,
+      "step": 423300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.755937569211929e-05,
+      "loss": 1.5249,
+      "step": 423400
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7553947305566054e-05,
+      "loss": 1.5157,
+      "step": 423500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7548518919012816e-05,
+      "loss": 1.5169,
+      "step": 423600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.754309053245958e-05,
+      "loss": 1.5482,
+      "step": 423700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7537662145906346e-05,
+      "loss": 1.5145,
+      "step": 423800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.753223375935311e-05,
+      "loss": 1.5434,
+      "step": 423900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7526805372799874e-05,
+      "loss": 1.505,
+      "step": 424000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.752137698624664e-05,
+      "loss": 1.5246,
+      "step": 424100
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7515948599693407e-05,
+      "loss": 1.532,
+      "step": 424200
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7510520213140173e-05,
+      "loss": 1.4976,
+      "step": 424300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7505091826586938e-05,
+      "loss": 1.5232,
+      "step": 424400
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7499663440033703e-05,
+      "loss": 1.5196,
+      "step": 424500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7494235053480465e-05,
+      "loss": 1.5297,
+      "step": 424600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.748880666692723e-05,
+      "loss": 1.5444,
+      "step": 424700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7483378280373996e-05,
+      "loss": 1.5246,
+      "step": 424800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7477949893820758e-05,
+      "loss": 1.5539,
+      "step": 424900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7472521507267523e-05,
+      "loss": 1.5018,
+      "step": 425000
+    },
+    {
+      "epoch": 1.83,
+      "eval_loss": 1.439378261566162,
+      "eval_runtime": 17.7944,
+      "eval_samples_per_second": 561.975,
+      "eval_steps_per_second": 17.59,
+      "step": 425000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7467093120714288e-05,
+      "loss": 1.5323,
+      "step": 425100
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7461664734161057e-05,
+      "loss": 1.557,
+      "step": 425200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7456236347607822e-05,
+      "loss": 1.5208,
+      "step": 425300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7450807961054587e-05,
+      "loss": 1.5268,
+      "step": 425400
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7445379574501353e-05,
+      "loss": 1.5093,
+      "step": 425500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7439951187948114e-05,
+      "loss": 1.5231,
+      "step": 425600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.743452280139488e-05,
+      "loss": 1.5062,
+      "step": 425700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7429094414841645e-05,
+      "loss": 1.5069,
+      "step": 425800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7423666028288407e-05,
+      "loss": 1.5228,
+      "step": 425900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7418237641735172e-05,
+      "loss": 1.5146,
+      "step": 426000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7412809255181937e-05,
+      "loss": 1.5084,
+      "step": 426100
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7407380868628703e-05,
+      "loss": 1.5113,
+      "step": 426200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.740195248207547e-05,
+      "loss": 1.523,
+      "step": 426300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7396524095522237e-05,
+      "loss": 1.5522,
+      "step": 426400
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7391095708969002e-05,
+      "loss": 1.5223,
+      "step": 426500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7385667322415764e-05,
+      "loss": 1.5028,
+      "step": 426600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.738023893586253e-05,
+      "loss": 1.5345,
+      "step": 426700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7374810549309294e-05,
+      "loss": 1.5229,
+      "step": 426800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7369382162756056e-05,
+      "loss": 1.5139,
+      "step": 426900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.736395377620282e-05,
+      "loss": 1.5395,
+      "step": 427000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7358525389649587e-05,
+      "loss": 1.542,
+      "step": 427100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7353097003096352e-05,
+      "loss": 1.5018,
+      "step": 427200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.734766861654312e-05,
+      "loss": 1.5242,
+      "step": 427300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7342240229989886e-05,
+      "loss": 1.5297,
+      "step": 427400
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.733681184343665e-05,
+      "loss": 1.5024,
+      "step": 427500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7331383456883413e-05,
+      "loss": 1.5456,
+      "step": 427600
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7325955070330178e-05,
+      "loss": 1.5168,
+      "step": 427700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7320526683776943e-05,
+      "loss": 1.5301,
+      "step": 427800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7315098297223705e-05,
+      "loss": 1.5365,
+      "step": 427900
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.730966991067047e-05,
+      "loss": 1.5342,
+      "step": 428000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7304241524117236e-05,
+      "loss": 1.543,
+      "step": 428100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7298813137564e-05,
+      "loss": 1.5274,
+      "step": 428200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7293384751010763e-05,
+      "loss": 1.5155,
+      "step": 428300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7287956364457535e-05,
+      "loss": 1.5238,
+      "step": 428400
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.72825279779043e-05,
+      "loss": 1.5082,
+      "step": 428500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7277099591351062e-05,
+      "loss": 1.5241,
+      "step": 428600
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7271671204797827e-05,
+      "loss": 1.538,
+      "step": 428700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7266242818244593e-05,
+      "loss": 1.5274,
+      "step": 428800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7260814431691355e-05,
+      "loss": 1.5441,
+      "step": 428900
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.725538604513812e-05,
+      "loss": 1.5111,
+      "step": 429000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7249957658584885e-05,
+      "loss": 1.5125,
+      "step": 429100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.724452927203165e-05,
+      "loss": 1.5252,
+      "step": 429200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7239100885478412e-05,
+      "loss": 1.5299,
+      "step": 429300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7233672498925184e-05,
+      "loss": 1.512,
+      "step": 429400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.722824411237195e-05,
+      "loss": 1.5447,
+      "step": 429500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.722281572581871e-05,
+      "loss": 1.5315,
+      "step": 429600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7217387339265477e-05,
+      "loss": 1.5102,
+      "step": 429700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7211958952712242e-05,
+      "loss": 1.5258,
+      "step": 429800
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7206530566159004e-05,
+      "loss": 1.5251,
+      "step": 429900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.720110217960577e-05,
+      "loss": 1.5062,
+      "step": 430000
+    },
+    {
+      "epoch": 1.85,
+      "eval_loss": 1.4390578269958496,
+      "eval_runtime": 17.787,
+      "eval_samples_per_second": 562.209,
+      "eval_steps_per_second": 17.597,
+      "step": 430000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7195673793052534e-05,
+      "loss": 1.5278,
+      "step": 430100
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.71902454064993e-05,
+      "loss": 1.5421,
+      "step": 430200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.718481701994606e-05,
+      "loss": 1.5117,
+      "step": 430300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7179388633392827e-05,
+      "loss": 1.5179,
+      "step": 430400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.71739602468396e-05,
+      "loss": 1.5315,
+      "step": 430500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.716853186028636e-05,
+      "loss": 1.5208,
+      "step": 430600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7163103473733126e-05,
+      "loss": 1.5164,
+      "step": 430700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.715767508717989e-05,
+      "loss": 1.5265,
+      "step": 430800
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7152246700626653e-05,
+      "loss": 1.5322,
+      "step": 430900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7146818314073418e-05,
+      "loss": 1.5207,
+      "step": 431000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7141389927520184e-05,
+      "loss": 1.5297,
+      "step": 431100
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.713596154096695e-05,
+      "loss": 1.5421,
+      "step": 431200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.713053315441371e-05,
+      "loss": 1.5178,
+      "step": 431300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7125104767860476e-05,
+      "loss": 1.5537,
+      "step": 431400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7119676381307248e-05,
+      "loss": 1.5231,
+      "step": 431500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.711424799475401e-05,
+      "loss": 1.5325,
+      "step": 431600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7108819608200775e-05,
+      "loss": 1.5152,
+      "step": 431700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.710339122164754e-05,
+      "loss": 1.5053,
+      "step": 431800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7097962835094302e-05,
+      "loss": 1.5255,
+      "step": 431900
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7092534448541068e-05,
+      "loss": 1.5174,
+      "step": 432000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7087106061987833e-05,
+      "loss": 1.5404,
+      "step": 432100
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7081677675434598e-05,
+      "loss": 1.5016,
+      "step": 432200
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.707624928888136e-05,
+      "loss": 1.5244,
+      "step": 432300
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7070820902328125e-05,
+      "loss": 1.5198,
+      "step": 432400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.706539251577489e-05,
+      "loss": 1.5196,
+      "step": 432500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.705996412922166e-05,
+      "loss": 1.5185,
+      "step": 432600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7054535742668424e-05,
+      "loss": 1.5117,
+      "step": 432700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.704910735611519e-05,
+      "loss": 1.5151,
+      "step": 432800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.704367896956195e-05,
+      "loss": 1.5069,
+      "step": 432900
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7038250583008717e-05,
+      "loss": 1.4896,
+      "step": 433000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7032822196455482e-05,
+      "loss": 1.5276,
+      "step": 433100
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7027393809902247e-05,
+      "loss": 1.5253,
+      "step": 433200
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.702196542334901e-05,
+      "loss": 1.5171,
+      "step": 433300
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7016537036795774e-05,
+      "loss": 1.5322,
+      "step": 433400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.701110865024254e-05,
+      "loss": 1.53,
+      "step": 433500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.700568026368931e-05,
+      "loss": 1.5743,
+      "step": 433600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7000251877136074e-05,
+      "loss": 1.551,
+      "step": 433700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.699482349058284e-05,
+      "loss": 1.5115,
+      "step": 433800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.69893951040296e-05,
+      "loss": 1.5231,
+      "step": 433900
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6983966717476366e-05,
+      "loss": 1.5368,
+      "step": 434000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.697853833092313e-05,
+      "loss": 1.5166,
+      "step": 434100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6973109944369897e-05,
+      "loss": 1.4913,
+      "step": 434200
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.696768155781666e-05,
+      "loss": 1.5306,
+      "step": 434300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6962253171263424e-05,
+      "loss": 1.5197,
+      "step": 434400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.695682478471019e-05,
+      "loss": 1.5224,
+      "step": 434500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6951396398156954e-05,
+      "loss": 1.55,
+      "step": 434600
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6945968011603723e-05,
+      "loss": 1.5211,
+      "step": 434700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6940539625050488e-05,
+      "loss": 1.521,
+      "step": 434800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.693511123849725e-05,
+      "loss": 1.5298,
+      "step": 434900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6929682851944015e-05,
+      "loss": 1.5378,
+      "step": 435000
+    },
+    {
+      "epoch": 1.87,
+      "eval_loss": 1.4357421398162842,
+      "eval_runtime": 17.828,
+      "eval_samples_per_second": 560.916,
+      "eval_steps_per_second": 17.557,
+      "step": 435000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.692425446539078e-05,
+      "loss": 1.5364,
+      "step": 435100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6918826078837546e-05,
+      "loss": 1.5312,
+      "step": 435200
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6913397692284308e-05,
+      "loss": 1.5099,
+      "step": 435300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6907969305731073e-05,
+      "loss": 1.5087,
+      "step": 435400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6902540919177838e-05,
+      "loss": 1.5656,
+      "step": 435500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6897112532624603e-05,
+      "loss": 1.5303,
+      "step": 435600
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6891684146071372e-05,
+      "loss": 1.5082,
+      "step": 435700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6886255759518137e-05,
+      "loss": 1.5383,
+      "step": 435800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.68808273729649e-05,
+      "loss": 1.5201,
+      "step": 435900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6875398986411665e-05,
+      "loss": 1.5027,
+      "step": 436000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.686997059985843e-05,
+      "loss": 1.5225,
+      "step": 436100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6864542213305195e-05,
+      "loss": 1.4969,
+      "step": 436200
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6859113826751957e-05,
+      "loss": 1.5574,
+      "step": 436300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6853685440198722e-05,
+      "loss": 1.5404,
+      "step": 436400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6848257053645487e-05,
+      "loss": 1.5161,
+      "step": 436500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6842828667092253e-05,
+      "loss": 1.5114,
+      "step": 436600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6837400280539015e-05,
+      "loss": 1.514,
+      "step": 436700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6831971893985787e-05,
+      "loss": 1.5196,
+      "step": 436800
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.682654350743255e-05,
+      "loss": 1.5169,
+      "step": 436900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6821115120879314e-05,
+      "loss": 1.513,
+      "step": 437000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.681568673432608e-05,
+      "loss": 1.5279,
+      "step": 437100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6810258347772844e-05,
+      "loss": 1.5097,
+      "step": 437200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6804829961219606e-05,
+      "loss": 1.522,
+      "step": 437300
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.679940157466637e-05,
+      "loss": 1.5354,
+      "step": 437400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6793973188113137e-05,
+      "loss": 1.5158,
+      "step": 437500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6788544801559902e-05,
+      "loss": 1.5274,
+      "step": 437600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6783116415006664e-05,
+      "loss": 1.5224,
+      "step": 437700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6777688028453436e-05,
+      "loss": 1.5052,
+      "step": 437800
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6772259641900198e-05,
+      "loss": 1.5173,
+      "step": 437900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6766831255346963e-05,
+      "loss": 1.4881,
+      "step": 438000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6761402868793728e-05,
+      "loss": 1.514,
+      "step": 438100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6755974482240494e-05,
+      "loss": 1.5351,
+      "step": 438200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6750546095687255e-05,
+      "loss": 1.5091,
+      "step": 438300
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.674511770913402e-05,
+      "loss": 1.5445,
+      "step": 438400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6739689322580786e-05,
+      "loss": 1.5062,
+      "step": 438500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.673426093602755e-05,
+      "loss": 1.558,
+      "step": 438600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6728832549474313e-05,
+      "loss": 1.5366,
+      "step": 438700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.672340416292108e-05,
+      "loss": 1.5452,
+      "step": 438800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6717975776367847e-05,
+      "loss": 1.5135,
+      "step": 438900
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6712547389814612e-05,
+      "loss": 1.5453,
+      "step": 439000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6707119003261377e-05,
+      "loss": 1.5165,
+      "step": 439100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6701690616708143e-05,
+      "loss": 1.5297,
+      "step": 439200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6696262230154905e-05,
+      "loss": 1.5236,
+      "step": 439300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.669083384360167e-05,
+      "loss": 1.5378,
+      "step": 439400
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6685405457048435e-05,
+      "loss": 1.5295,
+      "step": 439500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.66799770704952e-05,
+      "loss": 1.5415,
+      "step": 439600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6674548683941962e-05,
+      "loss": 1.5155,
+      "step": 439700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6669120297388728e-05,
+      "loss": 1.5277,
+      "step": 439800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6663691910835496e-05,
+      "loss": 1.535,
+      "step": 439900
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.665826352428226e-05,
+      "loss": 1.5224,
+      "step": 440000
+    },
+    {
+      "epoch": 1.89,
+      "eval_loss": 1.4352110624313354,
+      "eval_runtime": 17.8498,
+      "eval_samples_per_second": 560.232,
+      "eval_steps_per_second": 17.535,
+      "step": 440000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6652835137729027e-05,
+      "loss": 1.5307,
+      "step": 440100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6647406751175792e-05,
+      "loss": 1.5424,
+      "step": 440200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6641978364622554e-05,
+      "loss": 1.5518,
+      "step": 440300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.663654997806932e-05,
+      "loss": 1.5074,
+      "step": 440400
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6631121591516084e-05,
+      "loss": 1.5107,
+      "step": 440500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.662569320496285e-05,
+      "loss": 1.5267,
+      "step": 440600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.662026481840961e-05,
+      "loss": 1.5299,
+      "step": 440700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6614836431856377e-05,
+      "loss": 1.5614,
+      "step": 440800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6609408045303142e-05,
+      "loss": 1.5349,
+      "step": 440900
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.660397965874991e-05,
+      "loss": 1.5416,
+      "step": 441000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6598551272196676e-05,
+      "loss": 1.5203,
+      "step": 441100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.659312288564344e-05,
+      "loss": 1.5251,
+      "step": 441200
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6587694499090203e-05,
+      "loss": 1.5086,
+      "step": 441300
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.658226611253697e-05,
+      "loss": 1.5369,
+      "step": 441400
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6576837725983734e-05,
+      "loss": 1.5259,
+      "step": 441500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.65714093394305e-05,
+      "loss": 1.5062,
+      "step": 441600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.656598095287726e-05,
+      "loss": 1.4988,
+      "step": 441700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6560552566324026e-05,
+      "loss": 1.5246,
+      "step": 441800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.655512417977079e-05,
+      "loss": 1.5177,
+      "step": 441900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.654969579321756e-05,
+      "loss": 1.5016,
+      "step": 442000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6544267406664325e-05,
+      "loss": 1.5054,
+      "step": 442100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.653883902011109e-05,
+      "loss": 1.5185,
+      "step": 442200
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6533410633557852e-05,
+      "loss": 1.5031,
+      "step": 442300
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6527982247004618e-05,
+      "loss": 1.5084,
+      "step": 442400
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6522553860451383e-05,
+      "loss": 1.5021,
+      "step": 442500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6517125473898148e-05,
+      "loss": 1.5188,
+      "step": 442600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.651169708734491e-05,
+      "loss": 1.5382,
+      "step": 442700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6506268700791675e-05,
+      "loss": 1.5199,
+      "step": 442800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.650084031423844e-05,
+      "loss": 1.5069,
+      "step": 442900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6495411927685206e-05,
+      "loss": 1.4994,
+      "step": 443000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6489983541131974e-05,
+      "loss": 1.5141,
+      "step": 443100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.648455515457874e-05,
+      "loss": 1.5473,
+      "step": 443200
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.64791267680255e-05,
+      "loss": 1.549,
+      "step": 443300
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6473698381472267e-05,
+      "loss": 1.5292,
+      "step": 443400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6468269994919032e-05,
+      "loss": 1.535,
+      "step": 443500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6462841608365797e-05,
+      "loss": 1.5304,
+      "step": 443600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.645741322181256e-05,
+      "loss": 1.5453,
+      "step": 443700
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6451984835259325e-05,
+      "loss": 1.5084,
+      "step": 443800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.644655644870609e-05,
+      "loss": 1.5249,
+      "step": 443900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6441128062152855e-05,
+      "loss": 1.5258,
+      "step": 444000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6435699675599624e-05,
+      "loss": 1.5305,
+      "step": 444100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.643027128904639e-05,
+      "loss": 1.5316,
+      "step": 444200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.642484290249315e-05,
+      "loss": 1.5522,
+      "step": 444300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6419414515939916e-05,
+      "loss": 1.5034,
+      "step": 444400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.641398612938668e-05,
+      "loss": 1.532,
+      "step": 444500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6408557742833447e-05,
+      "loss": 1.5135,
+      "step": 444600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.640312935628021e-05,
+      "loss": 1.501,
+      "step": 444700
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6397700969726974e-05,
+      "loss": 1.5275,
+      "step": 444800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.639227258317374e-05,
+      "loss": 1.5153,
+      "step": 444900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6386844196620504e-05,
+      "loss": 1.5199,
+      "step": 445000
+    },
+    {
+      "epoch": 1.91,
+      "eval_loss": 1.4345873594284058,
+      "eval_runtime": 17.7853,
+      "eval_samples_per_second": 562.264,
+      "eval_steps_per_second": 17.599,
+      "step": 445000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6381415810067273e-05,
+      "loss": 1.5318,
+      "step": 445100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6375987423514038e-05,
+      "loss": 1.5213,
+      "step": 445200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.63705590369608e-05,
+      "loss": 1.5152,
+      "step": 445300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6365130650407565e-05,
+      "loss": 1.5457,
+      "step": 445400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.635970226385433e-05,
+      "loss": 1.5134,
+      "step": 445500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6354273877301096e-05,
+      "loss": 1.5224,
+      "step": 445600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6348845490747858e-05,
+      "loss": 1.4994,
+      "step": 445700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6343417104194623e-05,
+      "loss": 1.537,
+      "step": 445800
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6337988717641388e-05,
+      "loss": 1.5304,
+      "step": 445900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6332560331088154e-05,
+      "loss": 1.5238,
+      "step": 446000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6327131944534915e-05,
+      "loss": 1.5099,
+      "step": 446100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6321703557981687e-05,
+      "loss": 1.5341,
+      "step": 446200
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.631627517142845e-05,
+      "loss": 1.5001,
+      "step": 446300
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6310846784875215e-05,
+      "loss": 1.5191,
+      "step": 446400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.630541839832198e-05,
+      "loss": 1.5184,
+      "step": 446500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6299990011768745e-05,
+      "loss": 1.5,
+      "step": 446600
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6294561625215507e-05,
+      "loss": 1.522,
+      "step": 446700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6289133238662272e-05,
+      "loss": 1.5686,
+      "step": 446800
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6283704852109038e-05,
+      "loss": 1.5029,
+      "step": 446900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6278276465555803e-05,
+      "loss": 1.5285,
+      "step": 447000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6272848079002565e-05,
+      "loss": 1.5039,
+      "step": 447100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6267419692449337e-05,
+      "loss": 1.5305,
+      "step": 447200
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.62619913058961e-05,
+      "loss": 1.5144,
+      "step": 447300
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6256562919342864e-05,
+      "loss": 1.5168,
+      "step": 447400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.625113453278963e-05,
+      "loss": 1.5162,
+      "step": 447500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6245706146236394e-05,
+      "loss": 1.5354,
+      "step": 447600
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6240277759683156e-05,
+      "loss": 1.5558,
+      "step": 447700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.623484937312992e-05,
+      "loss": 1.491,
+      "step": 447800
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6229420986576687e-05,
+      "loss": 1.5375,
+      "step": 447900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6223992600023452e-05,
+      "loss": 1.5197,
+      "step": 448000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6218564213470214e-05,
+      "loss": 1.5252,
+      "step": 448100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.621313582691698e-05,
+      "loss": 1.4979,
+      "step": 448200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6207707440363748e-05,
+      "loss": 1.5203,
+      "step": 448300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6202279053810513e-05,
+      "loss": 1.4894,
+      "step": 448400
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.619685066725728e-05,
+      "loss": 1.5117,
+      "step": 448500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6191422280704044e-05,
+      "loss": 1.5073,
+      "step": 448600
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6185993894150805e-05,
+      "loss": 1.5084,
+      "step": 448700
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.618056550759757e-05,
+      "loss": 1.5031,
+      "step": 448800
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6175137121044336e-05,
+      "loss": 1.5309,
+      "step": 448900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.61697087344911e-05,
+      "loss": 1.5146,
+      "step": 449000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6164280347937863e-05,
+      "loss": 1.5523,
+      "step": 449100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.615885196138463e-05,
+      "loss": 1.54,
+      "step": 449200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6153423574831397e-05,
+      "loss": 1.5049,
+      "step": 449300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6147995188278162e-05,
+      "loss": 1.531,
+      "step": 449400
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6142566801724928e-05,
+      "loss": 1.4931,
+      "step": 449500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6137138415171693e-05,
+      "loss": 1.4995,
+      "step": 449600
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6131710028618455e-05,
+      "loss": 1.5199,
+      "step": 449700
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.612628164206522e-05,
+      "loss": 1.4933,
+      "step": 449800
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6120853255511985e-05,
+      "loss": 1.5186,
+      "step": 449900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.611542486895875e-05,
+      "loss": 1.5041,
+      "step": 450000
+    },
+    {
+      "epoch": 1.93,
+      "eval_loss": 1.4356664419174194,
+      "eval_runtime": 17.7962,
+      "eval_samples_per_second": 561.918,
+      "eval_steps_per_second": 17.588,
+      "step": 450000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6109996482405512e-05,
+      "loss": 1.487,
+      "step": 450100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6104568095852278e-05,
+      "loss": 1.5048,
+      "step": 450200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.6099139709299043e-05,
+      "loss": 1.5171,
+      "step": 450300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.609371132274581e-05,
+      "loss": 1.5121,
+      "step": 450400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6088282936192577e-05,
+      "loss": 1.528,
+      "step": 450500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6082854549639342e-05,
+      "loss": 1.5088,
+      "step": 450600
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6077426163086104e-05,
+      "loss": 1.5042,
+      "step": 450700
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.607199777653287e-05,
+      "loss": 1.5142,
+      "step": 450800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6066569389979635e-05,
+      "loss": 1.532,
+      "step": 450900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6061141003426396e-05,
+      "loss": 1.5488,
+      "step": 451000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.605571261687316e-05,
+      "loss": 1.5321,
+      "step": 451100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6050284230319927e-05,
+      "loss": 1.521,
+      "step": 451200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6044855843766692e-05,
+      "loss": 1.5191,
+      "step": 451300
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.603942745721346e-05,
+      "loss": 1.501,
+      "step": 451400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6033999070660226e-05,
+      "loss": 1.5219,
+      "step": 451500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.602857068410699e-05,
+      "loss": 1.5292,
+      "step": 451600
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6023142297553753e-05,
+      "loss": 1.5018,
+      "step": 451700
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.601771391100052e-05,
+      "loss": 1.55,
+      "step": 451800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6012285524447284e-05,
+      "loss": 1.511,
+      "step": 451900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6006857137894046e-05,
+      "loss": 1.5349,
+      "step": 452000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.600142875134081e-05,
+      "loss": 1.5489,
+      "step": 452100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.5996000364787576e-05,
+      "loss": 1.5337,
+      "step": 452200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.599057197823434e-05,
+      "loss": 1.5176,
+      "step": 452300
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.5985143591681103e-05,
+      "loss": 1.4814,
+      "step": 452400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.5979715205127875e-05,
+      "loss": 1.5198,
+      "step": 452500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.597428681857464e-05,
+      "loss": 1.4852,
+      "step": 452600
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.5968858432021402e-05,
+      "loss": 1.5318,
+      "step": 452700
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5963430045468168e-05,
+      "loss": 1.5186,
+      "step": 452800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5958001658914933e-05,
+      "loss": 1.5258,
+      "step": 452900
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5952573272361695e-05,
+      "loss": 1.5404,
+      "step": 453000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.594714488580846e-05,
+      "loss": 1.4918,
+      "step": 453100
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5941716499255225e-05,
+      "loss": 1.5234,
+      "step": 453200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.593628811270199e-05,
+      "loss": 1.5085,
+      "step": 453300
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5930859726148753e-05,
+      "loss": 1.5206,
+      "step": 453400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5925431339595525e-05,
+      "loss": 1.5355,
+      "step": 453500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.592000295304229e-05,
+      "loss": 1.5291,
+      "step": 453600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5914574566489052e-05,
+      "loss": 1.5067,
+      "step": 453700
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5909146179935817e-05,
+      "loss": 1.5327,
+      "step": 453800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5903717793382582e-05,
+      "loss": 1.5209,
+      "step": 453900
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5898289406829344e-05,
+      "loss": 1.4975,
+      "step": 454000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.589286102027611e-05,
+      "loss": 1.526,
+      "step": 454100
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5887432633722875e-05,
+      "loss": 1.5319,
+      "step": 454200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.588200424716964e-05,
+      "loss": 1.5306,
+      "step": 454300
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5876575860616402e-05,
+      "loss": 1.5138,
+      "step": 454400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5871147474063167e-05,
+      "loss": 1.5177,
+      "step": 454500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.586571908750994e-05,
+      "loss": 1.487,
+      "step": 454600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.58602907009567e-05,
+      "loss": 1.5251,
+      "step": 454700
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5854862314403466e-05,
+      "loss": 1.5141,
+      "step": 454800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.584943392785023e-05,
+      "loss": 1.5246,
+      "step": 454900
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5844005541296993e-05,
+      "loss": 1.4794,
+      "step": 455000
+    },
+    {
+      "epoch": 1.95,
+      "eval_loss": 1.437150239944458,
+      "eval_runtime": 17.7517,
+      "eval_samples_per_second": 563.327,
+      "eval_steps_per_second": 17.632,
+      "step": 455000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.583857715474376e-05,
+      "loss": 1.5201,
+      "step": 455100
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5833148768190524e-05,
+      "loss": 1.4924,
+      "step": 455200
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.582772038163729e-05,
+      "loss": 1.5248,
+      "step": 455300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.582229199508405e-05,
+      "loss": 1.5092,
+      "step": 455400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5816863608530816e-05,
+      "loss": 1.521,
+      "step": 455500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.581143522197759e-05,
+      "loss": 1.5166,
+      "step": 455600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.580600683542435e-05,
+      "loss": 1.5099,
+      "step": 455700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5800578448871115e-05,
+      "loss": 1.5127,
+      "step": 455800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.579515006231788e-05,
+      "loss": 1.5218,
+      "step": 455900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5789721675764643e-05,
+      "loss": 1.527,
+      "step": 456000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5784293289211408e-05,
+      "loss": 1.532,
+      "step": 456100
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5778864902658173e-05,
+      "loss": 1.5171,
+      "step": 456200
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.577343651610494e-05,
+      "loss": 1.5231,
+      "step": 456300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.57680081295517e-05,
+      "loss": 1.5034,
+      "step": 456400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5762579742998466e-05,
+      "loss": 1.5194,
+      "step": 456500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.575715135644523e-05,
+      "loss": 1.5078,
+      "step": 456600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5751722969892e-05,
+      "loss": 1.5181,
+      "step": 456700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5746294583338765e-05,
+      "loss": 1.5411,
+      "step": 456800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.574086619678553e-05,
+      "loss": 1.5068,
+      "step": 456900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5735437810232292e-05,
+      "loss": 1.5259,
+      "step": 457000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5730009423679057e-05,
+      "loss": 1.5365,
+      "step": 457100
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5724581037125822e-05,
+      "loss": 1.5247,
+      "step": 457200
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5719152650572588e-05,
+      "loss": 1.5386,
+      "step": 457300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.571372426401935e-05,
+      "loss": 1.4939,
+      "step": 457400
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5708295877466115e-05,
+      "loss": 1.5226,
+      "step": 457500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.570286749091288e-05,
+      "loss": 1.5273,
+      "step": 457600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.569743910435965e-05,
+      "loss": 1.5175,
+      "step": 457700
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5692010717806414e-05,
+      "loss": 1.5396,
+      "step": 457800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.568658233125318e-05,
+      "loss": 1.5206,
+      "step": 457900
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.568115394469994e-05,
+      "loss": 1.4951,
+      "step": 458000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5675725558146706e-05,
+      "loss": 1.5071,
+      "step": 458100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.567029717159347e-05,
+      "loss": 1.5149,
+      "step": 458200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5664868785040237e-05,
+      "loss": 1.5149,
+      "step": 458300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5659440398487e-05,
+      "loss": 1.4868,
+      "step": 458400
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5654012011933764e-05,
+      "loss": 1.5077,
+      "step": 458500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.564858362538053e-05,
+      "loss": 1.5018,
+      "step": 458600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5643155238827295e-05,
+      "loss": 1.5123,
+      "step": 458700
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5637726852274063e-05,
+      "loss": 1.4973,
+      "step": 458800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.563229846572083e-05,
+      "loss": 1.4967,
+      "step": 458900
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.562687007916759e-05,
+      "loss": 1.516,
+      "step": 459000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5621441692614356e-05,
+      "loss": 1.5316,
+      "step": 459100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.561601330606112e-05,
+      "loss": 1.5445,
+      "step": 459200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5610584919507886e-05,
+      "loss": 1.5118,
+      "step": 459300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5605156532954648e-05,
+      "loss": 1.5051,
+      "step": 459400
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5599728146401413e-05,
+      "loss": 1.5285,
+      "step": 459500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.559429975984818e-05,
+      "loss": 1.5028,
+      "step": 459600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5588871373294944e-05,
+      "loss": 1.5188,
+      "step": 459700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5583442986741712e-05,
+      "loss": 1.5244,
+      "step": 459800
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5578014600188478e-05,
+      "loss": 1.5232,
+      "step": 459900
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.557258621363524e-05,
+      "loss": 1.5254,
+      "step": 460000
+    },
+    {
+      "epoch": 1.98,
+      "eval_loss": 1.4372631311416626,
+      "eval_runtime": 17.8072,
+      "eval_samples_per_second": 561.57,
+      "eval_steps_per_second": 17.577,
+      "step": 460000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5567157827082005e-05,
+      "loss": 1.5217,
+      "step": 460100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.556172944052877e-05,
+      "loss": 1.5216,
+      "step": 460200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5556301053975535e-05,
+      "loss": 1.524,
+      "step": 460300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5550872667422297e-05,
+      "loss": 1.5024,
+      "step": 460400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5545444280869063e-05,
+      "loss": 1.5246,
+      "step": 460500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5540015894315828e-05,
+      "loss": 1.5202,
+      "step": 460600
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5534587507762593e-05,
+      "loss": 1.5033,
+      "step": 460700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5529159121209355e-05,
+      "loss": 1.5379,
+      "step": 460800
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5523730734656127e-05,
+      "loss": 1.5156,
+      "step": 460900
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.551830234810289e-05,
+      "loss": 1.514,
+      "step": 461000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5512873961549654e-05,
+      "loss": 1.4981,
+      "step": 461100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.550744557499642e-05,
+      "loss": 1.47,
+      "step": 461200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5502017188443185e-05,
+      "loss": 1.5092,
+      "step": 461300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5496588801889946e-05,
+      "loss": 1.5191,
+      "step": 461400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5491160415336712e-05,
+      "loss": 1.504,
+      "step": 461500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5485732028783477e-05,
+      "loss": 1.5084,
+      "step": 461600
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5480303642230242e-05,
+      "loss": 1.4905,
+      "step": 461700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5474875255677004e-05,
+      "loss": 1.5079,
+      "step": 461800
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5469446869123776e-05,
+      "loss": 1.4988,
+      "step": 461900
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5464018482570538e-05,
+      "loss": 1.5224,
+      "step": 462000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5458590096017303e-05,
+      "loss": 1.5292,
+      "step": 462100
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.545316170946407e-05,
+      "loss": 1.5291,
+      "step": 462200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5447733322910834e-05,
+      "loss": 1.5155,
+      "step": 462300
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5442304936357596e-05,
+      "loss": 1.5209,
+      "step": 462400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.543687654980436e-05,
+      "loss": 1.5217,
+      "step": 462500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5431448163251126e-05,
+      "loss": 1.5291,
+      "step": 462600
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.542601977669789e-05,
+      "loss": 1.5398,
+      "step": 462700
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5420591390144653e-05,
+      "loss": 1.5249,
+      "step": 462800
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.541516300359142e-05,
+      "loss": 1.5094,
+      "step": 462900
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5409734617038187e-05,
+      "loss": 1.5231,
+      "step": 463000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5404306230484953e-05,
+      "loss": 1.4996,
+      "step": 463100
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5398877843931718e-05,
+      "loss": 1.5071,
+      "step": 463200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5393449457378483e-05,
+      "loss": 1.5225,
+      "step": 463300
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5388021070825245e-05,
+      "loss": 1.5091,
+      "step": 463400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.538259268427201e-05,
+      "loss": 1.5096,
+      "step": 463500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5377164297718776e-05,
+      "loss": 1.5081,
+      "step": 463600
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.537173591116554e-05,
+      "loss": 1.5027,
+      "step": 463700
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5366307524612303e-05,
+      "loss": 1.5129,
+      "step": 463800
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5360879138059068e-05,
+      "loss": 1.5128,
+      "step": 463900
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5355450751505837e-05,
+      "loss": 1.5264,
+      "step": 464000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5350022364952602e-05,
+      "loss": 1.5181,
+      "step": 464100
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5344593978399367e-05,
+      "loss": 1.492,
+      "step": 464200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5339165591846132e-05,
+      "loss": 1.5279,
+      "step": 464300
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5333737205292894e-05,
+      "loss": 1.5017,
+      "step": 464400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.532830881873966e-05,
+      "loss": 1.4949,
+      "step": 464500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5322880432186425e-05,
+      "loss": 1.5386,
+      "step": 464600
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.531745204563319e-05,
+      "loss": 1.501,
+      "step": 464700
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5312023659079952e-05,
+      "loss": 1.556,
+      "step": 464800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5306595272526717e-05,
+      "loss": 1.5504,
+      "step": 464900
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5301166885973486e-05,
+      "loss": 1.5136,
+      "step": 465000
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 1.4351954460144043,
+      "eval_runtime": 17.8126,
+      "eval_samples_per_second": 561.4,
+      "eval_steps_per_second": 17.572,
+      "step": 465000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.529573849942025e-05,
+      "loss": 1.537,
+      "step": 465100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5290310112867016e-05,
+      "loss": 1.5129,
+      "step": 465200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.528488172631378e-05,
+      "loss": 1.533,
+      "step": 465300
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5279453339760543e-05,
+      "loss": 1.5411,
+      "step": 465400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.527402495320731e-05,
+      "loss": 1.5306,
+      "step": 465500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5268596566654074e-05,
+      "loss": 1.4965,
+      "step": 465600
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.526316818010084e-05,
+      "loss": 1.5094,
+      "step": 465700
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.52577397935476e-05,
+      "loss": 1.481,
+      "step": 465800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5252311406994366e-05,
+      "loss": 1.4859,
+      "step": 465900
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.524688302044113e-05,
+      "loss": 1.4919,
+      "step": 466000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.52414546338879e-05,
+      "loss": 1.5023,
+      "step": 466100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5236026247334666e-05,
+      "loss": 1.4949,
+      "step": 466200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.523059786078143e-05,
+      "loss": 1.4836,
+      "step": 466300
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5225169474228193e-05,
+      "loss": 1.4577,
+      "step": 466400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5219741087674958e-05,
+      "loss": 1.5039,
+      "step": 466500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5214312701121723e-05,
+      "loss": 1.4643,
+      "step": 466600
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.520888431456849e-05,
+      "loss": 1.5069,
+      "step": 466700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.520345592801525e-05,
+      "loss": 1.4909,
+      "step": 466800
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.5198027541462016e-05,
+      "loss": 1.4872,
+      "step": 466900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.519259915490878e-05,
+      "loss": 1.4882,
+      "step": 467000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.518717076835555e-05,
+      "loss": 1.4986,
+      "step": 467100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.5181742381802315e-05,
+      "loss": 1.4678,
+      "step": 467200
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.517631399524908e-05,
+      "loss": 1.4588,
+      "step": 467300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.5170885608695842e-05,
+      "loss": 1.4589,
+      "step": 467400
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.5165457222142607e-05,
+      "loss": 1.4516,
+      "step": 467500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.5160028835589372e-05,
+      "loss": 1.4877,
+      "step": 467600
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.5154600449036138e-05,
+      "loss": 1.46,
+      "step": 467700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.51491720624829e-05,
+      "loss": 1.4712,
+      "step": 467800
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.5143743675929665e-05,
+      "loss": 1.4915,
+      "step": 467900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.513831528937643e-05,
+      "loss": 1.4865,
+      "step": 468000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.5132886902823195e-05,
+      "loss": 1.5087,
+      "step": 468100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.5127458516269964e-05,
+      "loss": 1.4844,
+      "step": 468200
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.512203012971673e-05,
+      "loss": 1.4879,
+      "step": 468300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.511660174316349e-05,
+      "loss": 1.4974,
+      "step": 468400
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.5111173356610256e-05,
+      "loss": 1.4683,
+      "step": 468500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.5105744970057022e-05,
+      "loss": 1.4931,
+      "step": 468600
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.5100316583503787e-05,
+      "loss": 1.4729,
+      "step": 468700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.509488819695055e-05,
+      "loss": 1.4603,
+      "step": 468800
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.5089459810397314e-05,
+      "loss": 1.4908,
+      "step": 468900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.508403142384408e-05,
+      "loss": 1.4753,
+      "step": 469000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.5078603037290845e-05,
+      "loss": 1.5068,
+      "step": 469100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.5073174650737613e-05,
+      "loss": 1.48,
+      "step": 469200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.506774626418438e-05,
+      "loss": 1.465,
+      "step": 469300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.506231787763114e-05,
+      "loss": 1.4885,
+      "step": 469400
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.5056889491077906e-05,
+      "loss": 1.5117,
+      "step": 469500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.505146110452467e-05,
+      "loss": 1.4701,
+      "step": 469600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.5046032717971436e-05,
+      "loss": 1.4813,
+      "step": 469700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.5040604331418198e-05,
+      "loss": 1.489,
+      "step": 469800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.5035175944864963e-05,
+      "loss": 1.4867,
+      "step": 469900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.502974755831173e-05,
+      "loss": 1.4885,
+      "step": 470000
+    },
+    {
+      "epoch": 2.02,
+      "eval_loss": 1.4379206895828247,
+      "eval_runtime": 17.8048,
+      "eval_samples_per_second": 561.647,
+      "eval_steps_per_second": 17.58,
+      "step": 470000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.5024319171758494e-05,
+      "loss": 1.4895,
+      "step": 470100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.5018890785205256e-05,
+      "loss": 1.4791,
+      "step": 470200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.5013462398652028e-05,
+      "loss": 1.4837,
+      "step": 470300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.500803401209879e-05,
+      "loss": 1.4775,
+      "step": 470400
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.5002605625545555e-05,
+      "loss": 1.519,
+      "step": 470500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.499717723899232e-05,
+      "loss": 1.4725,
+      "step": 470600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.4991748852439085e-05,
+      "loss": 1.5076,
+      "step": 470700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.4986320465885847e-05,
+      "loss": 1.4859,
+      "step": 470800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.4980892079332613e-05,
+      "loss": 1.4572,
+      "step": 470900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.4975463692779378e-05,
+      "loss": 1.4833,
+      "step": 471000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.4970035306226143e-05,
+      "loss": 1.4604,
+      "step": 471100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.496460691967291e-05,
+      "loss": 1.4769,
+      "step": 471200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.4959178533119674e-05,
+      "loss": 1.4817,
+      "step": 471300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.495375014656644e-05,
+      "loss": 1.4724,
+      "step": 471400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.49483217600132e-05,
+      "loss": 1.4648,
+      "step": 471500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.494289337345997e-05,
+      "loss": 1.4813,
+      "step": 471600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4937464986906735e-05,
+      "loss": 1.4614,
+      "step": 471700
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4932036600353497e-05,
+      "loss": 1.4949,
+      "step": 471800
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4926608213800262e-05,
+      "loss": 1.4819,
+      "step": 471900
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4921179827247027e-05,
+      "loss": 1.4853,
+      "step": 472000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4915751440693792e-05,
+      "loss": 1.4949,
+      "step": 472100
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4910323054140558e-05,
+      "loss": 1.4927,
+      "step": 472200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4904894667587323e-05,
+      "loss": 1.4945,
+      "step": 472300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4899466281034088e-05,
+      "loss": 1.4696,
+      "step": 472400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.489403789448085e-05,
+      "loss": 1.447,
+      "step": 472500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.488860950792762e-05,
+      "loss": 1.4982,
+      "step": 472600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4883181121374384e-05,
+      "loss": 1.4921,
+      "step": 472700
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4877752734821146e-05,
+      "loss": 1.4724,
+      "step": 472800
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.487232434826791e-05,
+      "loss": 1.5049,
+      "step": 472900
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4866895961714676e-05,
+      "loss": 1.4724,
+      "step": 473000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.486146757516144e-05,
+      "loss": 1.4891,
+      "step": 473100
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4856039188608207e-05,
+      "loss": 1.4846,
+      "step": 473200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4850610802054972e-05,
+      "loss": 1.4849,
+      "step": 473300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4845182415501737e-05,
+      "loss": 1.4698,
+      "step": 473400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.48397540289485e-05,
+      "loss": 1.4848,
+      "step": 473500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4834325642395265e-05,
+      "loss": 1.5008,
+      "step": 473600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4828897255842033e-05,
+      "loss": 1.4955,
+      "step": 473700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4823468869288795e-05,
+      "loss": 1.5106,
+      "step": 473800
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.481804048273556e-05,
+      "loss": 1.4705,
+      "step": 473900
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4812612096182326e-05,
+      "loss": 1.4965,
+      "step": 474000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.480718370962909e-05,
+      "loss": 1.4966,
+      "step": 474100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4801755323075856e-05,
+      "loss": 1.4636,
+      "step": 474200
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.479632693652262e-05,
+      "loss": 1.493,
+      "step": 474300
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4790898549969387e-05,
+      "loss": 1.4891,
+      "step": 474400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.478547016341615e-05,
+      "loss": 1.47,
+      "step": 474500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4780041776862914e-05,
+      "loss": 1.4939,
+      "step": 474600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4774613390309682e-05,
+      "loss": 1.5032,
+      "step": 474700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4769185003756444e-05,
+      "loss": 1.4862,
+      "step": 474800
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.476375661720321e-05,
+      "loss": 1.4728,
+      "step": 474900
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4758328230649975e-05,
+      "loss": 1.4978,
+      "step": 475000
+    },
+    {
+      "epoch": 2.04,
+      "eval_loss": 1.4373873472213745,
+      "eval_runtime": 17.7804,
+      "eval_samples_per_second": 562.418,
+      "eval_steps_per_second": 17.604,
+      "step": 475000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.475289984409674e-05,
+      "loss": 1.4943,
+      "step": 475100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4747471457543502e-05,
+      "loss": 1.4735,
+      "step": 475200
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.474204307099027e-05,
+      "loss": 1.4863,
+      "step": 475300
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4736614684437036e-05,
+      "loss": 1.4672,
+      "step": 475400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4731186297883798e-05,
+      "loss": 1.5015,
+      "step": 475500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4725757911330563e-05,
+      "loss": 1.4682,
+      "step": 475600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.472032952477733e-05,
+      "loss": 1.4963,
+      "step": 475700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4714901138224094e-05,
+      "loss": 1.4793,
+      "step": 475800
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.470947275167086e-05,
+      "loss": 1.4853,
+      "step": 475900
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4704044365117624e-05,
+      "loss": 1.4781,
+      "step": 476000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.469861597856439e-05,
+      "loss": 1.4825,
+      "step": 476100
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.469318759201115e-05,
+      "loss": 1.502,
+      "step": 476200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.468775920545792e-05,
+      "loss": 1.4894,
+      "step": 476300
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4682330818904685e-05,
+      "loss": 1.5123,
+      "step": 476400
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4676902432351447e-05,
+      "loss": 1.4597,
+      "step": 476500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4671474045798212e-05,
+      "loss": 1.4996,
+      "step": 476600
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4666045659244978e-05,
+      "loss": 1.4949,
+      "step": 476700
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4660617272691743e-05,
+      "loss": 1.4974,
+      "step": 476800
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4655188886138508e-05,
+      "loss": 1.4989,
+      "step": 476900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4649760499585273e-05,
+      "loss": 1.4714,
+      "step": 477000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.464433211303204e-05,
+      "loss": 1.4798,
+      "step": 477100
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.46389037264788e-05,
+      "loss": 1.4677,
+      "step": 477200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4633475339925566e-05,
+      "loss": 1.4871,
+      "step": 477300
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4628046953372334e-05,
+      "loss": 1.5236,
+      "step": 477400
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4622618566819096e-05,
+      "loss": 1.48,
+      "step": 477500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.461719018026586e-05,
+      "loss": 1.473,
+      "step": 477600
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4611761793712627e-05,
+      "loss": 1.4726,
+      "step": 477700
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4606333407159392e-05,
+      "loss": 1.4829,
+      "step": 477800
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4600905020606157e-05,
+      "loss": 1.4723,
+      "step": 477900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4595476634052923e-05,
+      "loss": 1.4876,
+      "step": 478000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4590048247499688e-05,
+      "loss": 1.4885,
+      "step": 478100
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.458461986094645e-05,
+      "loss": 1.486,
+      "step": 478200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4579191474393215e-05,
+      "loss": 1.4968,
+      "step": 478300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4573763087839984e-05,
+      "loss": 1.4917,
+      "step": 478400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4568334701286746e-05,
+      "loss": 1.4664,
+      "step": 478500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.456290631473351e-05,
+      "loss": 1.5124,
+      "step": 478600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4557477928180276e-05,
+      "loss": 1.4945,
+      "step": 478700
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.455204954162704e-05,
+      "loss": 1.48,
+      "step": 478800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4546621155073807e-05,
+      "loss": 1.4764,
+      "step": 478900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4541192768520572e-05,
+      "loss": 1.51,
+      "step": 479000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4535764381967337e-05,
+      "loss": 1.4715,
+      "step": 479100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.45303359954141e-05,
+      "loss": 1.4857,
+      "step": 479200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4524907608860864e-05,
+      "loss": 1.5054,
+      "step": 479300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4519479222307633e-05,
+      "loss": 1.4588,
+      "step": 479400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4514050835754395e-05,
+      "loss": 1.4906,
+      "step": 479500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.450862244920116e-05,
+      "loss": 1.4416,
+      "step": 479600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4503194062647925e-05,
+      "loss": 1.4945,
+      "step": 479700
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.449776567609469e-05,
+      "loss": 1.4888,
+      "step": 479800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4492337289541452e-05,
+      "loss": 1.4918,
+      "step": 479900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.448690890298822e-05,
+      "loss": 1.4872,
+      "step": 480000
+    },
+    {
+      "epoch": 2.06,
+      "eval_loss": 1.435674786567688,
+      "eval_runtime": 17.7907,
+      "eval_samples_per_second": 562.091,
+      "eval_steps_per_second": 17.593,
+      "step": 480000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4481480516434986e-05,
+      "loss": 1.4666,
+      "step": 480100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4476052129881748e-05,
+      "loss": 1.491,
+      "step": 480200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4470623743328513e-05,
+      "loss": 1.482,
+      "step": 480300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.446519535677528e-05,
+      "loss": 1.4848,
+      "step": 480400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4459766970222044e-05,
+      "loss": 1.4843,
+      "step": 480500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.445433858366881e-05,
+      "loss": 1.4776,
+      "step": 480600
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4448910197115575e-05,
+      "loss": 1.4761,
+      "step": 480700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4443481810562336e-05,
+      "loss": 1.5118,
+      "step": 480800
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.44380534240091e-05,
+      "loss": 1.4673,
+      "step": 480900
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.443262503745587e-05,
+      "loss": 1.5171,
+      "step": 481000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4427196650902636e-05,
+      "loss": 1.4556,
+      "step": 481100
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4421768264349397e-05,
+      "loss": 1.5094,
+      "step": 481200
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4416339877796163e-05,
+      "loss": 1.4814,
+      "step": 481300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4410911491242928e-05,
+      "loss": 1.4673,
+      "step": 481400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4405483104689693e-05,
+      "loss": 1.4845,
+      "step": 481500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.440005471813646e-05,
+      "loss": 1.507,
+      "step": 481600
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4394626331583224e-05,
+      "loss": 1.4716,
+      "step": 481700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4389197945029986e-05,
+      "loss": 1.4578,
+      "step": 481800
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.438376955847675e-05,
+      "loss": 1.4706,
+      "step": 481900
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4378341171923516e-05,
+      "loss": 1.4801,
+      "step": 482000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4372912785370285e-05,
+      "loss": 1.4767,
+      "step": 482100
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4367484398817047e-05,
+      "loss": 1.4646,
+      "step": 482200
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4362056012263812e-05,
+      "loss": 1.4711,
+      "step": 482300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4356627625710577e-05,
+      "loss": 1.5042,
+      "step": 482400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.435119923915734e-05,
+      "loss": 1.5114,
+      "step": 482500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4345770852604108e-05,
+      "loss": 1.4726,
+      "step": 482600
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4340342466050873e-05,
+      "loss": 1.4743,
+      "step": 482700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4334914079497635e-05,
+      "loss": 1.4796,
+      "step": 482800
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.43294856929444e-05,
+      "loss": 1.4428,
+      "step": 482900
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4324057306391165e-05,
+      "loss": 1.5025,
+      "step": 483000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4318628919837934e-05,
+      "loss": 1.5128,
+      "step": 483100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4313200533284696e-05,
+      "loss": 1.4972,
+      "step": 483200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.430777214673146e-05,
+      "loss": 1.4691,
+      "step": 483300
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4302343760178226e-05,
+      "loss": 1.4655,
+      "step": 483400
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.429691537362499e-05,
+      "loss": 1.4713,
+      "step": 483500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4291486987071757e-05,
+      "loss": 1.4963,
+      "step": 483600
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4286058600518522e-05,
+      "loss": 1.4704,
+      "step": 483700
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4280630213965284e-05,
+      "loss": 1.4644,
+      "step": 483800
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.427520182741205e-05,
+      "loss": 1.5307,
+      "step": 483900
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4269773440858815e-05,
+      "loss": 1.4934,
+      "step": 484000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.426434505430558e-05,
+      "loss": 1.4855,
+      "step": 484100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4258916667752345e-05,
+      "loss": 1.4747,
+      "step": 484200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.425348828119911e-05,
+      "loss": 1.4986,
+      "step": 484300
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4248059894645876e-05,
+      "loss": 1.496,
+      "step": 484400
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4242631508092638e-05,
+      "loss": 1.4976,
+      "step": 484500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4237203121539403e-05,
+      "loss": 1.4877,
+      "step": 484600
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.423177473498617e-05,
+      "loss": 1.5055,
+      "step": 484700
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4226346348432933e-05,
+      "loss": 1.4922,
+      "step": 484800
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.42209179618797e-05,
+      "loss": 1.4723,
+      "step": 484900
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4215489575326464e-05,
+      "loss": 1.5055,
+      "step": 485000
+    },
+    {
+      "epoch": 2.08,
+      "eval_loss": 1.433396816253662,
+      "eval_runtime": 17.803,
+      "eval_samples_per_second": 561.704,
+      "eval_steps_per_second": 17.581,
+      "step": 485000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.421006118877323e-05,
+      "loss": 1.4748,
+      "step": 485100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4204632802219994e-05,
+      "loss": 1.451,
+      "step": 485200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.419920441566676e-05,
+      "loss": 1.4868,
+      "step": 485300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4193776029113525e-05,
+      "loss": 1.4999,
+      "step": 485400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4188347642560287e-05,
+      "loss": 1.4661,
+      "step": 485500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4182919256007052e-05,
+      "loss": 1.5062,
+      "step": 485600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.417749086945382e-05,
+      "loss": 1.4775,
+      "step": 485700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4172062482900583e-05,
+      "loss": 1.4957,
+      "step": 485800
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4166634096347348e-05,
+      "loss": 1.5192,
+      "step": 485900
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4161205709794113e-05,
+      "loss": 1.4921,
+      "step": 486000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.415577732324088e-05,
+      "loss": 1.4891,
+      "step": 486100
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.415034893668764e-05,
+      "loss": 1.4821,
+      "step": 486200
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.414492055013441e-05,
+      "loss": 1.491,
+      "step": 486300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4139492163581174e-05,
+      "loss": 1.5049,
+      "step": 486400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4134063777027936e-05,
+      "loss": 1.5108,
+      "step": 486500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.41286353904747e-05,
+      "loss": 1.4862,
+      "step": 486600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4123207003921467e-05,
+      "loss": 1.4695,
+      "step": 486700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4117778617368232e-05,
+      "loss": 1.4921,
+      "step": 486800
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4112350230814997e-05,
+      "loss": 1.4882,
+      "step": 486900
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4106921844261762e-05,
+      "loss": 1.4948,
+      "step": 487000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4101493457708528e-05,
+      "loss": 1.4979,
+      "step": 487100
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.409606507115529e-05,
+      "loss": 1.4862,
+      "step": 487200
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4090636684602058e-05,
+      "loss": 1.4722,
+      "step": 487300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4085208298048823e-05,
+      "loss": 1.4701,
+      "step": 487400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.4079779911495585e-05,
+      "loss": 1.46,
+      "step": 487500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.407435152494235e-05,
+      "loss": 1.4932,
+      "step": 487600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4068923138389116e-05,
+      "loss": 1.482,
+      "step": 487700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.406349475183588e-05,
+      "loss": 1.4611,
+      "step": 487800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4058066365282646e-05,
+      "loss": 1.4995,
+      "step": 487900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.405263797872941e-05,
+      "loss": 1.451,
+      "step": 488000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4047209592176177e-05,
+      "loss": 1.4817,
+      "step": 488100
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.404178120562294e-05,
+      "loss": 1.4892,
+      "step": 488200
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4036352819069704e-05,
+      "loss": 1.4875,
+      "step": 488300
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4030924432516473e-05,
+      "loss": 1.4717,
+      "step": 488400
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4025496045963235e-05,
+      "loss": 1.4998,
+      "step": 488500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.402006765941e-05,
+      "loss": 1.4945,
+      "step": 488600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4014639272856765e-05,
+      "loss": 1.4908,
+      "step": 488700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.400921088630353e-05,
+      "loss": 1.4813,
+      "step": 488800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4003782499750296e-05,
+      "loss": 1.4892,
+      "step": 488900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.399835411319706e-05,
+      "loss": 1.4752,
+      "step": 489000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3992925726643826e-05,
+      "loss": 1.5083,
+      "step": 489100
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3987497340090588e-05,
+      "loss": 1.4873,
+      "step": 489200
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3982068953537353e-05,
+      "loss": 1.4875,
+      "step": 489300
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3976640566984122e-05,
+      "loss": 1.4657,
+      "step": 489400
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3971212180430884e-05,
+      "loss": 1.4937,
+      "step": 489500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.396578379387765e-05,
+      "loss": 1.4957,
+      "step": 489600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3960355407324414e-05,
+      "loss": 1.497,
+      "step": 489700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.395492702077118e-05,
+      "loss": 1.4825,
+      "step": 489800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3949498634217945e-05,
+      "loss": 1.481,
+      "step": 489900
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.394407024766471e-05,
+      "loss": 1.4926,
+      "step": 490000
+    },
+    {
+      "epoch": 2.11,
+      "eval_loss": 1.4341309070587158,
+      "eval_runtime": 17.8295,
+      "eval_samples_per_second": 560.867,
+      "eval_steps_per_second": 17.555,
+      "step": 490000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3938641861111475e-05,
+      "loss": 1.4879,
+      "step": 490100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3933213474558237e-05,
+      "loss": 1.4901,
+      "step": 490200
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3927785088005003e-05,
+      "loss": 1.4596,
+      "step": 490300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.392235670145177e-05,
+      "loss": 1.5034,
+      "step": 490400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3916928314898533e-05,
+      "loss": 1.4965,
+      "step": 490500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.39114999283453e-05,
+      "loss": 1.4744,
+      "step": 490600
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3906071541792064e-05,
+      "loss": 1.4832,
+      "step": 490700
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.390064315523883e-05,
+      "loss": 1.4838,
+      "step": 490800
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.389521476868559e-05,
+      "loss": 1.474,
+      "step": 490900
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.388978638213236e-05,
+      "loss": 1.502,
+      "step": 491000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3884357995579125e-05,
+      "loss": 1.5034,
+      "step": 491100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3878929609025887e-05,
+      "loss": 1.4979,
+      "step": 491200
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3873501222472652e-05,
+      "loss": 1.4854,
+      "step": 491300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3868072835919417e-05,
+      "loss": 1.4684,
+      "step": 491400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3862644449366182e-05,
+      "loss": 1.4853,
+      "step": 491500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3857216062812948e-05,
+      "loss": 1.4748,
+      "step": 491600
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3851787676259713e-05,
+      "loss": 1.4767,
+      "step": 491700
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3846359289706478e-05,
+      "loss": 1.471,
+      "step": 491800
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.384093090315324e-05,
+      "loss": 1.5002,
+      "step": 491900
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.383550251660001e-05,
+      "loss": 1.4628,
+      "step": 492000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3830074130046774e-05,
+      "loss": 1.4645,
+      "step": 492100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3824645743493536e-05,
+      "loss": 1.4713,
+      "step": 492200
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.38192173569403e-05,
+      "loss": 1.476,
+      "step": 492300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3813788970387066e-05,
+      "loss": 1.4673,
+      "step": 492400
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.380836058383383e-05,
+      "loss": 1.4811,
+      "step": 492500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3802932197280597e-05,
+      "loss": 1.4554,
+      "step": 492600
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3797503810727362e-05,
+      "loss": 1.4661,
+      "step": 492700
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3792075424174127e-05,
+      "loss": 1.45,
+      "step": 492800
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.378664703762089e-05,
+      "loss": 1.4734,
+      "step": 492900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3781218651067654e-05,
+      "loss": 1.5128,
+      "step": 493000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3775790264514423e-05,
+      "loss": 1.4665,
+      "step": 493100
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3770361877961185e-05,
+      "loss": 1.5025,
+      "step": 493200
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.376493349140795e-05,
+      "loss": 1.4727,
+      "step": 493300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3759505104854716e-05,
+      "loss": 1.4396,
+      "step": 493400
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.375407671830148e-05,
+      "loss": 1.4604,
+      "step": 493500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3748648331748246e-05,
+      "loss": 1.4847,
+      "step": 493600
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.374321994519501e-05,
+      "loss": 1.4947,
+      "step": 493700
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3737791558641777e-05,
+      "loss": 1.4925,
+      "step": 493800
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.373236317208854e-05,
+      "loss": 1.4761,
+      "step": 493900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3726934785535304e-05,
+      "loss": 1.5098,
+      "step": 494000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3721506398982072e-05,
+      "loss": 1.49,
+      "step": 494100
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3716078012428834e-05,
+      "loss": 1.5106,
+      "step": 494200
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.37106496258756e-05,
+      "loss": 1.489,
+      "step": 494300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3705221239322365e-05,
+      "loss": 1.5143,
+      "step": 494400
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.369979285276913e-05,
+      "loss": 1.4895,
+      "step": 494500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3694364466215895e-05,
+      "loss": 1.4879,
+      "step": 494600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.368893607966266e-05,
+      "loss": 1.5082,
+      "step": 494700
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3683507693109426e-05,
+      "loss": 1.4785,
+      "step": 494800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3678079306556188e-05,
+      "loss": 1.4912,
+      "step": 494900
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3672650920002953e-05,
+      "loss": 1.4702,
+      "step": 495000
+    },
+    {
+      "epoch": 2.13,
+      "eval_loss": 1.4374511241912842,
+      "eval_runtime": 17.7875,
+      "eval_samples_per_second": 562.192,
+      "eval_steps_per_second": 17.597,
+      "step": 495000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3667222533449718e-05,
+      "loss": 1.5075,
+      "step": 495100
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3661794146896483e-05,
+      "loss": 1.4957,
+      "step": 495200
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.365636576034325e-05,
+      "loss": 1.4718,
+      "step": 495300
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3650937373790014e-05,
+      "loss": 1.4745,
+      "step": 495400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.364550898723678e-05,
+      "loss": 1.4938,
+      "step": 495500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.364008060068354e-05,
+      "loss": 1.5004,
+      "step": 495600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.363465221413031e-05,
+      "loss": 1.4827,
+      "step": 495700
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3629223827577075e-05,
+      "loss": 1.5126,
+      "step": 495800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3623795441023837e-05,
+      "loss": 1.4899,
+      "step": 495900
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3618367054470602e-05,
+      "loss": 1.4931,
+      "step": 496000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3612938667917367e-05,
+      "loss": 1.454,
+      "step": 496100
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3607510281364133e-05,
+      "loss": 1.4802,
+      "step": 496200
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3602081894810898e-05,
+      "loss": 1.4774,
+      "step": 496300
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3596653508257663e-05,
+      "loss": 1.4746,
+      "step": 496400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.359122512170443e-05,
+      "loss": 1.5036,
+      "step": 496500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.358579673515119e-05,
+      "loss": 1.4754,
+      "step": 496600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.358036834859796e-05,
+      "loss": 1.4807,
+      "step": 496700
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3574939962044724e-05,
+      "loss": 1.4749,
+      "step": 496800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3569511575491486e-05,
+      "loss": 1.5181,
+      "step": 496900
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.356408318893825e-05,
+      "loss": 1.5014,
+      "step": 497000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3558654802385017e-05,
+      "loss": 1.5074,
+      "step": 497100
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3553226415831782e-05,
+      "loss": 1.4821,
+      "step": 497200
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3547798029278547e-05,
+      "loss": 1.4851,
+      "step": 497300
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3542369642725313e-05,
+      "loss": 1.4852,
+      "step": 497400
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3536941256172078e-05,
+      "loss": 1.4734,
+      "step": 497500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.353151286961884e-05,
+      "loss": 1.4846,
+      "step": 497600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3526084483065605e-05,
+      "loss": 1.5076,
+      "step": 497700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3520656096512374e-05,
+      "loss": 1.4697,
+      "step": 497800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3515227709959135e-05,
+      "loss": 1.476,
+      "step": 497900
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.35097993234059e-05,
+      "loss": 1.4897,
+      "step": 498000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3504370936852666e-05,
+      "loss": 1.4802,
+      "step": 498100
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.349894255029943e-05,
+      "loss": 1.4825,
+      "step": 498200
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3493514163746196e-05,
+      "loss": 1.4744,
+      "step": 498300
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3488085777192962e-05,
+      "loss": 1.462,
+      "step": 498400
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3482657390639727e-05,
+      "loss": 1.4743,
+      "step": 498500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.347722900408649e-05,
+      "loss": 1.4801,
+      "step": 498600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3471800617533254e-05,
+      "loss": 1.4718,
+      "step": 498700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3466372230980023e-05,
+      "loss": 1.4911,
+      "step": 498800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3460943844426785e-05,
+      "loss": 1.4852,
+      "step": 498900
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.345551545787355e-05,
+      "loss": 1.4648,
+      "step": 499000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3450087071320315e-05,
+      "loss": 1.4936,
+      "step": 499100
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.344465868476708e-05,
+      "loss": 1.4734,
+      "step": 499200
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3439230298213846e-05,
+      "loss": 1.4845,
+      "step": 499300
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.343380191166061e-05,
+      "loss": 1.4689,
+      "step": 499400
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3428373525107376e-05,
+      "loss": 1.4944,
+      "step": 499500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3422945138554138e-05,
+      "loss": 1.4626,
+      "step": 499600
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3417516752000903e-05,
+      "loss": 1.4614,
+      "step": 499700
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.341208836544767e-05,
+      "loss": 1.509,
+      "step": 499800
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3406659978894434e-05,
+      "loss": 1.4618,
+      "step": 499900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.34012315923412e-05,
+      "loss": 1.4663,
+      "step": 500000
+    },
+    {
+      "epoch": 2.15,
+      "eval_loss": 1.436499834060669,
+      "eval_runtime": 17.8206,
+      "eval_samples_per_second": 561.148,
+      "eval_steps_per_second": 17.564,
+      "step": 500000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3395803205787964e-05,
+      "loss": 1.4606,
+      "step": 500100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.339037481923473e-05,
+      "loss": 1.4687,
+      "step": 500200
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.338494643268149e-05,
+      "loss": 1.4966,
+      "step": 500300
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.337951804612826e-05,
+      "loss": 1.4548,
+      "step": 500400
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3374089659575026e-05,
+      "loss": 1.4535,
+      "step": 500500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3368661273021787e-05,
+      "loss": 1.4985,
+      "step": 500600
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3363232886468553e-05,
+      "loss": 1.4927,
+      "step": 500700
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3357804499915318e-05,
+      "loss": 1.4969,
+      "step": 500800
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3352376113362083e-05,
+      "loss": 1.4861,
+      "step": 500900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.334694772680885e-05,
+      "loss": 1.4743,
+      "step": 501000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3341519340255614e-05,
+      "loss": 1.507,
+      "step": 501100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.333609095370238e-05,
+      "loss": 1.5005,
+      "step": 501200
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.333066256714914e-05,
+      "loss": 1.4792,
+      "step": 501300
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.332523418059591e-05,
+      "loss": 1.4758,
+      "step": 501400
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3319805794042675e-05,
+      "loss": 1.5028,
+      "step": 501500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3314377407489437e-05,
+      "loss": 1.4661,
+      "step": 501600
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3308949020936202e-05,
+      "loss": 1.4981,
+      "step": 501700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3303520634382967e-05,
+      "loss": 1.4711,
+      "step": 501800
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3298092247829732e-05,
+      "loss": 1.4494,
+      "step": 501900
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3292663861276498e-05,
+      "loss": 1.4624,
+      "step": 502000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3287235474723263e-05,
+      "loss": 1.49,
+      "step": 502100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3281807088170028e-05,
+      "loss": 1.4928,
+      "step": 502200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.327637870161679e-05,
+      "loss": 1.5082,
+      "step": 502300
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3270950315063555e-05,
+      "loss": 1.4798,
+      "step": 502400
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3265521928510324e-05,
+      "loss": 1.4794,
+      "step": 502500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3260093541957086e-05,
+      "loss": 1.5046,
+      "step": 502600
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.325466515540385e-05,
+      "loss": 1.4967,
+      "step": 502700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3249236768850616e-05,
+      "loss": 1.5038,
+      "step": 502800
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.324380838229738e-05,
+      "loss": 1.5092,
+      "step": 502900
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3238379995744147e-05,
+      "loss": 1.4548,
+      "step": 503000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3232951609190912e-05,
+      "loss": 1.4837,
+      "step": 503100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3227523222637677e-05,
+      "loss": 1.4739,
+      "step": 503200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.322209483608444e-05,
+      "loss": 1.489,
+      "step": 503300
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3216666449531205e-05,
+      "loss": 1.4906,
+      "step": 503400
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3211238062977973e-05,
+      "loss": 1.5106,
+      "step": 503500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3205809676424735e-05,
+      "loss": 1.4605,
+      "step": 503600
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.32003812898715e-05,
+      "loss": 1.4719,
+      "step": 503700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3194952903318266e-05,
+      "loss": 1.4849,
+      "step": 503800
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.318952451676503e-05,
+      "loss": 1.4882,
+      "step": 503900
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3184096130211793e-05,
+      "loss": 1.4845,
+      "step": 504000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.317866774365856e-05,
+      "loss": 1.4624,
+      "step": 504100
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3173239357105327e-05,
+      "loss": 1.4871,
+      "step": 504200
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.316781097055209e-05,
+      "loss": 1.4711,
+      "step": 504300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3162382583998854e-05,
+      "loss": 1.4815,
+      "step": 504400
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.315695419744562e-05,
+      "loss": 1.4643,
+      "step": 504500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3151525810892384e-05,
+      "loss": 1.4835,
+      "step": 504600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.314609742433915e-05,
+      "loss": 1.4565,
+      "step": 504700
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3140669037785915e-05,
+      "loss": 1.4739,
+      "step": 504800
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.313524065123268e-05,
+      "loss": 1.4988,
+      "step": 504900
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3129812264679442e-05,
+      "loss": 1.4915,
+      "step": 505000
+    },
+    {
+      "epoch": 2.17,
+      "eval_loss": 1.4346122741699219,
+      "eval_runtime": 17.8459,
+      "eval_samples_per_second": 560.352,
+      "eval_steps_per_second": 17.539,
+      "step": 505000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.312438387812621e-05,
+      "loss": 1.5072,
+      "step": 505100
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3118955491572976e-05,
+      "loss": 1.5014,
+      "step": 505200
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3113527105019738e-05,
+      "loss": 1.461,
+      "step": 505300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3108098718466503e-05,
+      "loss": 1.4723,
+      "step": 505400
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.310267033191327e-05,
+      "loss": 1.4914,
+      "step": 505500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3097241945360034e-05,
+      "loss": 1.4882,
+      "step": 505600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.30918135588068e-05,
+      "loss": 1.4902,
+      "step": 505700
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3086385172253564e-05,
+      "loss": 1.4811,
+      "step": 505800
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.308095678570033e-05,
+      "loss": 1.4857,
+      "step": 505900
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.307552839914709e-05,
+      "loss": 1.5034,
+      "step": 506000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3070100012593857e-05,
+      "loss": 1.4841,
+      "step": 506100
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3064671626040625e-05,
+      "loss": 1.4934,
+      "step": 506200
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3059243239487387e-05,
+      "loss": 1.4803,
+      "step": 506300
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3053814852934152e-05,
+      "loss": 1.4701,
+      "step": 506400
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3048386466380918e-05,
+      "loss": 1.4797,
+      "step": 506500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3042958079827683e-05,
+      "loss": 1.4941,
+      "step": 506600
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3037529693274448e-05,
+      "loss": 1.4571,
+      "step": 506700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3032101306721213e-05,
+      "loss": 1.4694,
+      "step": 506800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.302667292016798e-05,
+      "loss": 1.4953,
+      "step": 506900
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.302124453361474e-05,
+      "loss": 1.5125,
+      "step": 507000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3015816147061506e-05,
+      "loss": 1.5192,
+      "step": 507100
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3010387760508274e-05,
+      "loss": 1.4959,
+      "step": 507200
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3004959373955036e-05,
+      "loss": 1.4939,
+      "step": 507300
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.29995309874018e-05,
+      "loss": 1.5133,
+      "step": 507400
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2994102600848567e-05,
+      "loss": 1.4837,
+      "step": 507500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2988674214295332e-05,
+      "loss": 1.4846,
+      "step": 507600
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2983245827742097e-05,
+      "loss": 1.502,
+      "step": 507700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2977817441188863e-05,
+      "loss": 1.5174,
+      "step": 507800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2972389054635628e-05,
+      "loss": 1.4953,
+      "step": 507900
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.296696066808239e-05,
+      "loss": 1.4986,
+      "step": 508000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2961532281529155e-05,
+      "loss": 1.4764,
+      "step": 508100
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.295610389497592e-05,
+      "loss": 1.4775,
+      "step": 508200
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2950675508422686e-05,
+      "loss": 1.4733,
+      "step": 508300
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.294524712186945e-05,
+      "loss": 1.4912,
+      "step": 508400
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2939818735316216e-05,
+      "loss": 1.4955,
+      "step": 508500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.293439034876298e-05,
+      "loss": 1.466,
+      "step": 508600
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2928961962209743e-05,
+      "loss": 1.5112,
+      "step": 508700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2923533575656512e-05,
+      "loss": 1.4897,
+      "step": 508800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2918105189103277e-05,
+      "loss": 1.5011,
+      "step": 508900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.291267680255004e-05,
+      "loss": 1.4853,
+      "step": 509000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2907248415996804e-05,
+      "loss": 1.4533,
+      "step": 509100
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.290182002944357e-05,
+      "loss": 1.4893,
+      "step": 509200
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2896391642890335e-05,
+      "loss": 1.5081,
+      "step": 509300
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.28909632563371e-05,
+      "loss": 1.4702,
+      "step": 509400
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2885534869783865e-05,
+      "loss": 1.5152,
+      "step": 509500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.288010648323063e-05,
+      "loss": 1.4632,
+      "step": 509600
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2874678096677392e-05,
+      "loss": 1.4534,
+      "step": 509700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.286924971012416e-05,
+      "loss": 1.484,
+      "step": 509800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2863821323570926e-05,
+      "loss": 1.464,
+      "step": 509900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2858392937017688e-05,
+      "loss": 1.4934,
+      "step": 510000
+    },
+    {
+      "epoch": 2.19,
+      "eval_loss": 1.4337053298950195,
+      "eval_runtime": 17.8118,
+      "eval_samples_per_second": 561.427,
+      "eval_steps_per_second": 17.573,
+      "step": 510000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2852964550464454e-05,
+      "loss": 1.4795,
+      "step": 510100
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.284753616391122e-05,
+      "loss": 1.4797,
+      "step": 510200
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2842107777357984e-05,
+      "loss": 1.4538,
+      "step": 510300
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.283667939080475e-05,
+      "loss": 1.4941,
+      "step": 510400
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2831251004251515e-05,
+      "loss": 1.4898,
+      "step": 510500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2825822617698276e-05,
+      "loss": 1.4943,
+      "step": 510600
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2820394231145042e-05,
+      "loss": 1.4766,
+      "step": 510700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2814965844591807e-05,
+      "loss": 1.5113,
+      "step": 510800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2809537458038576e-05,
+      "loss": 1.487,
+      "step": 510900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2804109071485337e-05,
+      "loss": 1.4824,
+      "step": 511000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2798680684932103e-05,
+      "loss": 1.4604,
+      "step": 511100
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2793252298378868e-05,
+      "loss": 1.4889,
+      "step": 511200
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.278782391182563e-05,
+      "loss": 1.5088,
+      "step": 511300
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.27823955252724e-05,
+      "loss": 1.4742,
+      "step": 511400
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2776967138719164e-05,
+      "loss": 1.4663,
+      "step": 511500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2771538752165926e-05,
+      "loss": 1.4459,
+      "step": 511600
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.276611036561269e-05,
+      "loss": 1.4881,
+      "step": 511700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2760681979059456e-05,
+      "loss": 1.4614,
+      "step": 511800
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2755253592506225e-05,
+      "loss": 1.4643,
+      "step": 511900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2749825205952987e-05,
+      "loss": 1.4864,
+      "step": 512000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2744396819399752e-05,
+      "loss": 1.4967,
+      "step": 512100
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2738968432846517e-05,
+      "loss": 1.4663,
+      "step": 512200
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.273354004629328e-05,
+      "loss": 1.5019,
+      "step": 512300
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2728111659740048e-05,
+      "loss": 1.4702,
+      "step": 512400
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2722683273186813e-05,
+      "loss": 1.5025,
+      "step": 512500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2717254886633575e-05,
+      "loss": 1.496,
+      "step": 512600
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.271182650008034e-05,
+      "loss": 1.4837,
+      "step": 512700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2706398113527105e-05,
+      "loss": 1.4755,
+      "step": 512800
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.270096972697387e-05,
+      "loss": 1.4758,
+      "step": 512900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2695541340420636e-05,
+      "loss": 1.4985,
+      "step": 513000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.26901129538674e-05,
+      "loss": 1.4931,
+      "step": 513100
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2684684567314166e-05,
+      "loss": 1.4566,
+      "step": 513200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.267925618076093e-05,
+      "loss": 1.4659,
+      "step": 513300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2673827794207694e-05,
+      "loss": 1.4651,
+      "step": 513400
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2668399407654462e-05,
+      "loss": 1.4844,
+      "step": 513500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2662971021101224e-05,
+      "loss": 1.4773,
+      "step": 513600
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.265754263454799e-05,
+      "loss": 1.4805,
+      "step": 513700
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2652114247994755e-05,
+      "loss": 1.4656,
+      "step": 513800
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.264668586144152e-05,
+      "loss": 1.4908,
+      "step": 513900
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2641257474888285e-05,
+      "loss": 1.4835,
+      "step": 514000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.263582908833505e-05,
+      "loss": 1.4758,
+      "step": 514100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2630400701781816e-05,
+      "loss": 1.4812,
+      "step": 514200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2624972315228578e-05,
+      "loss": 1.4898,
+      "step": 514300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2619543928675343e-05,
+      "loss": 1.4905,
+      "step": 514400
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.261411554212211e-05,
+      "loss": 1.4645,
+      "step": 514500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2608687155568873e-05,
+      "loss": 1.4528,
+      "step": 514600
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.260325876901564e-05,
+      "loss": 1.4878,
+      "step": 514700
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2597830382462404e-05,
+      "loss": 1.4721,
+      "step": 514800
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.259240199590917e-05,
+      "loss": 1.5166,
+      "step": 514900
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.258697360935593e-05,
+      "loss": 1.4518,
+      "step": 515000
+    },
+    {
+      "epoch": 2.21,
+      "eval_loss": 1.4345492124557495,
+      "eval_runtime": 17.7486,
+      "eval_samples_per_second": 563.424,
+      "eval_steps_per_second": 17.635,
+      "step": 515000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.25815452228027e-05,
+      "loss": 1.5096,
+      "step": 515100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2576116836249465e-05,
+      "loss": 1.5049,
+      "step": 515200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2570688449696227e-05,
+      "loss": 1.4933,
+      "step": 515300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2565260063142992e-05,
+      "loss": 1.4921,
+      "step": 515400
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2559831676589757e-05,
+      "loss": 1.4608,
+      "step": 515500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2554403290036523e-05,
+      "loss": 1.496,
+      "step": 515600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2548974903483288e-05,
+      "loss": 1.5036,
+      "step": 515700
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2543546516930053e-05,
+      "loss": 1.4806,
+      "step": 515800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.253811813037682e-05,
+      "loss": 1.4941,
+      "step": 515900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.253268974382358e-05,
+      "loss": 1.4829,
+      "step": 516000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.252726135727035e-05,
+      "loss": 1.4574,
+      "step": 516100
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2521832970717114e-05,
+      "loss": 1.4756,
+      "step": 516200
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2516404584163876e-05,
+      "loss": 1.5002,
+      "step": 516300
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.251097619761064e-05,
+      "loss": 1.4989,
+      "step": 516400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2505547811057407e-05,
+      "loss": 1.4672,
+      "step": 516500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2500119424504172e-05,
+      "loss": 1.478,
+      "step": 516600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2494691037950937e-05,
+      "loss": 1.4855,
+      "step": 516700
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2489262651397702e-05,
+      "loss": 1.4786,
+      "step": 516800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2483834264844468e-05,
+      "loss": 1.4896,
+      "step": 516900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.247840587829123e-05,
+      "loss": 1.4775,
+      "step": 517000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2472977491737995e-05,
+      "loss": 1.4999,
+      "step": 517100
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2467549105184763e-05,
+      "loss": 1.4808,
+      "step": 517200
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2462120718631525e-05,
+      "loss": 1.491,
+      "step": 517300
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.245669233207829e-05,
+      "loss": 1.4508,
+      "step": 517400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2451263945525056e-05,
+      "loss": 1.4689,
+      "step": 517500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.244583555897182e-05,
+      "loss": 1.4916,
+      "step": 517600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2440407172418586e-05,
+      "loss": 1.4849,
+      "step": 517700
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.243497878586535e-05,
+      "loss": 1.4747,
+      "step": 517800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2429550399312117e-05,
+      "loss": 1.4743,
+      "step": 517900
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.242412201275888e-05,
+      "loss": 1.5091,
+      "step": 518000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2418693626205644e-05,
+      "loss": 1.4748,
+      "step": 518100
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2413265239652413e-05,
+      "loss": 1.5079,
+      "step": 518200
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2407836853099175e-05,
+      "loss": 1.5001,
+      "step": 518300
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.240240846654594e-05,
+      "loss": 1.51,
+      "step": 518400
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2396980079992705e-05,
+      "loss": 1.4986,
+      "step": 518500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.239155169343947e-05,
+      "loss": 1.4622,
+      "step": 518600
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2386123306886236e-05,
+      "loss": 1.4945,
+      "step": 518700
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2380694920333e-05,
+      "loss": 1.5062,
+      "step": 518800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2375266533779766e-05,
+      "loss": 1.4986,
+      "step": 518900
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2369838147226528e-05,
+      "loss": 1.4779,
+      "step": 519000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2364409760673293e-05,
+      "loss": 1.4839,
+      "step": 519100
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2358981374120062e-05,
+      "loss": 1.511,
+      "step": 519200
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2353552987566824e-05,
+      "loss": 1.5019,
+      "step": 519300
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.234812460101359e-05,
+      "loss": 1.462,
+      "step": 519400
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2342696214460354e-05,
+      "loss": 1.4597,
+      "step": 519500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.233726782790712e-05,
+      "loss": 1.4734,
+      "step": 519600
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.233183944135388e-05,
+      "loss": 1.4984,
+      "step": 519700
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.232641105480065e-05,
+      "loss": 1.4915,
+      "step": 519800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2320982668247415e-05,
+      "loss": 1.4425,
+      "step": 519900
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2315554281694177e-05,
+      "loss": 1.4798,
+      "step": 520000
+    },
+    {
+      "epoch": 2.23,
+      "eval_loss": 1.4345911741256714,
+      "eval_runtime": 17.8258,
+      "eval_samples_per_second": 560.985,
+      "eval_steps_per_second": 17.559,
+      "step": 520000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2310125895140943e-05,
+      "loss": 1.4789,
+      "step": 520100
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2304697508587708e-05,
+      "loss": 1.5009,
+      "step": 520200
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2299269122034473e-05,
+      "loss": 1.4765,
+      "step": 520300
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.229384073548124e-05,
+      "loss": 1.4855,
+      "step": 520400
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2288412348928004e-05,
+      "loss": 1.4905,
+      "step": 520500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.228298396237477e-05,
+      "loss": 1.5065,
+      "step": 520600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.227755557582153e-05,
+      "loss": 1.4894,
+      "step": 520700
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.22721271892683e-05,
+      "loss": 1.4793,
+      "step": 520800
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2266698802715065e-05,
+      "loss": 1.475,
+      "step": 520900
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2261270416161827e-05,
+      "loss": 1.4647,
+      "step": 521000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2255842029608592e-05,
+      "loss": 1.4542,
+      "step": 521100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2250413643055357e-05,
+      "loss": 1.4955,
+      "step": 521200
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2244985256502122e-05,
+      "loss": 1.4918,
+      "step": 521300
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2239556869948888e-05,
+      "loss": 1.5104,
+      "step": 521400
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2234128483395653e-05,
+      "loss": 1.5074,
+      "step": 521500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2228700096842418e-05,
+      "loss": 1.4817,
+      "step": 521600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.222327171028918e-05,
+      "loss": 1.4727,
+      "step": 521700
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2217843323735945e-05,
+      "loss": 1.4676,
+      "step": 521800
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2212414937182714e-05,
+      "loss": 1.4566,
+      "step": 521900
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2206986550629476e-05,
+      "loss": 1.4915,
+      "step": 522000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.220155816407624e-05,
+      "loss": 1.4709,
+      "step": 522100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2196129777523006e-05,
+      "loss": 1.4749,
+      "step": 522200
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.219070139096977e-05,
+      "loss": 1.4818,
+      "step": 522300
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2185273004416537e-05,
+      "loss": 1.4536,
+      "step": 522400
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2179844617863302e-05,
+      "loss": 1.4939,
+      "step": 522500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2174416231310067e-05,
+      "loss": 1.4868,
+      "step": 522600
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.216898784475683e-05,
+      "loss": 1.4932,
+      "step": 522700
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2163559458203594e-05,
+      "loss": 1.4582,
+      "step": 522800
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2158131071650363e-05,
+      "loss": 1.4579,
+      "step": 522900
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2152702685097125e-05,
+      "loss": 1.477,
+      "step": 523000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.214727429854389e-05,
+      "loss": 1.4633,
+      "step": 523100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2141845911990656e-05,
+      "loss": 1.4759,
+      "step": 523200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.213641752543742e-05,
+      "loss": 1.4887,
+      "step": 523300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2130989138884186e-05,
+      "loss": 1.4862,
+      "step": 523400
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.212556075233095e-05,
+      "loss": 1.4889,
+      "step": 523500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2120132365777717e-05,
+      "loss": 1.4924,
+      "step": 523600
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.211470397922448e-05,
+      "loss": 1.4896,
+      "step": 523700
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2109275592671244e-05,
+      "loss": 1.4904,
+      "step": 523800
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.210384720611801e-05,
+      "loss": 1.4934,
+      "step": 523900
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2098418819564774e-05,
+      "loss": 1.5086,
+      "step": 524000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.209299043301154e-05,
+      "loss": 1.4682,
+      "step": 524100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2087562046458305e-05,
+      "loss": 1.4608,
+      "step": 524200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.208213365990507e-05,
+      "loss": 1.4936,
+      "step": 524300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2076705273351832e-05,
+      "loss": 1.4867,
+      "step": 524400
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.20712768867986e-05,
+      "loss": 1.472,
+      "step": 524500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2065848500245366e-05,
+      "loss": 1.4607,
+      "step": 524600
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2060420113692128e-05,
+      "loss": 1.4941,
+      "step": 524700
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.2054991727138893e-05,
+      "loss": 1.4715,
+      "step": 524800
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.2049563340585658e-05,
+      "loss": 1.4758,
+      "step": 524900
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.2044134954032424e-05,
+      "loss": 1.4918,
+      "step": 525000
+    },
+    {
+      "epoch": 2.26,
+      "eval_loss": 1.4338562488555908,
+      "eval_runtime": 17.7905,
+      "eval_samples_per_second": 562.097,
+      "eval_steps_per_second": 17.594,
+      "step": 525000
+    }
+  ],
+  "max_steps": 931084,
+  "num_train_epochs": 4,
+  "total_flos": 1.0159882796191924e+19,
+  "trial_name": null,
+  "trial_params": null
+}