diff --git "a/microsoft/grammarBERT/trainer_state.json" "b/microsoft/grammarBERT/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/microsoft/grammarBERT/trainer_state.json"
@@ -0,0 +1,9313 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.983767459418648,
+  "eval_steps": 500,
+  "global_step": 92500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999928e-05,
+      "loss": 7.2081,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999856e-05,
+      "loss": 3.5421,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999784000000004e-05,
+      "loss": 2.2759,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9999712e-05,
+      "loss": 1.8451,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9999640000000003e-05,
+      "loss": 1.6239,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9999568e-05,
+      "loss": 1.5319,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9999496e-05,
+      "loss": 1.4177,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9999424e-05,
+      "loss": 1.3957,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9999352e-05,
+      "loss": 1.3271,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.999928e-05,
+      "loss": 1.2944,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9999208e-05,
+      "loss": 1.2562,
+      "step": 660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9999136000000005e-05,
+      "loss": 1.2175,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9999064e-05,
+      "loss": 1.1996,
+      "step": 780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9998992e-05,
+      "loss": 1.1729,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.999892e-05,
+      "loss": 1.1586,
+      "step": 900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9998848000000004e-05,
+      "loss": 1.1274,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9998776e-05,
+      "loss": 1.0855,
+      "step": 1020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9998704e-05,
+      "loss": 1.1155,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9998632e-05,
+      "loss": 1.104,
+      "step": 1140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.999856e-05,
+      "loss": 1.0725,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9998488000000006e-05,
+      "loss": 1.0544,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9998416e-05,
+      "loss": 1.0552,
+      "step": 1320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9998344e-05,
+      "loss": 1.0507,
+      "step": 1380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9998272e-05,
+      "loss": 1.0456,
+      "step": 1440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9998200000000004e-05,
+      "loss": 1.027,
+      "step": 1500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9998128e-05,
+      "loss": 1.0333,
+      "step": 1560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9998056e-05,
+      "loss": 1.009,
+      "step": 1620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.999798400000001e-05,
+      "loss": 1.0066,
+      "step": 1680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9997912e-05,
+      "loss": 0.9916,
+      "step": 1740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.999784e-05,
+      "loss": 1.0084,
+      "step": 1800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9997768e-05,
+      "loss": 0.9778,
+      "step": 1860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9997696e-05,
+      "loss": 0.9835,
+      "step": 1920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9997624e-05,
+      "loss": 0.9726,
+      "step": 1980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9997552000000005e-05,
+      "loss": 0.974,
+      "step": 2040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.999748e-05,
+      "loss": 0.9648,
+      "step": 2100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9997408e-05,
+      "loss": 0.9555,
+      "step": 2160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9997336e-05,
+      "loss": 0.9506,
+      "step": 2220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9997264000000004e-05,
+      "loss": 0.9694,
+      "step": 2280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9997192e-05,
+      "loss": 0.9429,
+      "step": 2340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9997119999999997e-05,
+      "loss": 0.9526,
+      "step": 2400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9997048000000006e-05,
+      "loss": 0.9249,
+      "step": 2460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9996976e-05,
+      "loss": 0.9133,
+      "step": 2520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9996904e-05,
+      "loss": 0.9296,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9996832e-05,
+      "loss": 0.9436,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.999676e-05,
+      "loss": 0.9182,
+      "step": 2700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9996688e-05,
+      "loss": 0.8987,
+      "step": 2760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9996616000000005e-05,
+      "loss": 0.9043,
+      "step": 2820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9996544e-05,
+      "loss": 0.908,
+      "step": 2880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9996472e-05,
+      "loss": 0.8998,
+      "step": 2940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.999640000000001e-05,
+      "loss": 0.9018,
+      "step": 3000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9996328000000003e-05,
+      "loss": 0.89,
+      "step": 3060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9996256e-05,
+      "loss": 0.8844,
+      "step": 3120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9996183999999996e-05,
+      "loss": 0.9006,
+      "step": 3180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9996112000000006e-05,
+      "loss": 0.8923,
+      "step": 3240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.999604e-05,
+      "loss": 0.907,
+      "step": 3300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9995968e-05,
+      "loss": 0.8881,
+      "step": 3360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9995896e-05,
+      "loss": 0.8787,
+      "step": 3420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9995824e-05,
+      "loss": 0.8692,
+      "step": 3480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9995752e-05,
+      "loss": 0.8746,
+      "step": 3540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9995680000000004e-05,
+      "loss": 0.8755,
+      "step": 3600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9995608e-05,
+      "loss": 0.8868,
+      "step": 3660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9995536e-05,
+      "loss": 0.8726,
+      "step": 3720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.999546400000001e-05,
+      "loss": 0.8636,
+      "step": 3780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9995392e-05,
+      "loss": 0.8691,
+      "step": 3840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.999532e-05,
+      "loss": 0.862,
+      "step": 3900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9995248e-05,
+      "loss": 0.8606,
+      "step": 3960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9995176000000006e-05,
+      "loss": 0.8735,
+      "step": 4020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9995104e-05,
+      "loss": 0.8453,
+      "step": 4080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9995032e-05,
+      "loss": 0.8491,
+      "step": 4140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.999496e-05,
+      "loss": 0.8714,
+      "step": 4200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9994888000000004e-05,
+      "loss": 0.8438,
+      "step": 4260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9994816e-05,
+      "loss": 0.877,
+      "step": 4320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9994744000000004e-05,
+      "loss": 0.8526,
+      "step": 4380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.9994672e-05,
+      "loss": 0.8675,
+      "step": 4440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.9994599999999996e-05,
+      "loss": 0.8572,
+      "step": 4500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.9994528000000006e-05,
+      "loss": 0.853,
+      "step": 4560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.9994456e-05,
+      "loss": 0.8329,
+      "step": 4620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.9994384e-05,
+      "loss": 0.852,
+      "step": 4680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.9994312e-05,
+      "loss": 0.8385,
+      "step": 4740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.9994240000000005e-05,
+      "loss": 0.8431,
+      "step": 4800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.9994168e-05,
+      "loss": 0.8492,
+      "step": 4860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.9994096e-05,
+      "loss": 0.8421,
+      "step": 4920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.9994024e-05,
+      "loss": 0.8302,
+      "step": 4980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.9993952000000004e-05,
+      "loss": 0.8301,
+      "step": 5040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.999388e-05,
+      "loss": 0.8375,
+      "step": 5100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.9993808000000003e-05,
+      "loss": 0.8303,
+      "step": 5160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.9993736e-05,
+      "loss": 0.8361,
+      "step": 5220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9993663999999996e-05,
+      "loss": 0.8357,
+      "step": 5280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9993592000000006e-05,
+      "loss": 0.8418,
+      "step": 5340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.999352e-05,
+      "loss": 0.8424,
+      "step": 5400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9993448e-05,
+      "loss": 0.8374,
+      "step": 5460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.9993376e-05,
+      "loss": 0.8274,
+      "step": 5520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.9993304000000005e-05,
+      "loss": 0.8327,
+      "step": 5580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.9993232e-05,
+      "loss": 0.8032,
+      "step": 5640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.9993160000000004e-05,
+      "loss": 0.8153,
+      "step": 5700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.9993088e-05,
+      "loss": 0.8205,
+      "step": 5760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9993016000000004e-05,
+      "loss": 0.8156,
+      "step": 5820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9992944e-05,
+      "loss": 0.8177,
+      "step": 5880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9992872e-05,
+      "loss": 0.8276,
+      "step": 5940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.99928e-05,
+      "loss": 0.7939,
+      "step": 6000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9992727999999996e-05,
+      "loss": 0.8119,
+      "step": 6060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9992656000000005e-05,
+      "loss": 0.8212,
+      "step": 6120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9992584e-05,
+      "loss": 0.8212,
+      "step": 6180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9992512e-05,
+      "loss": 0.8284,
+      "step": 6240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.999244e-05,
+      "loss": 0.8143,
+      "step": 6300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9992368000000004e-05,
+      "loss": 0.8034,
+      "step": 6360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9992296e-05,
+      "loss": 0.824,
+      "step": 6420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.9992224000000004e-05,
+      "loss": 0.804,
+      "step": 6480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.9992152e-05,
+      "loss": 0.8078,
+      "step": 6540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.999208e-05,
+      "loss": 0.7899,
+      "step": 6600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9992008e-05,
+      "loss": 0.808,
+      "step": 6660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.9991936e-05,
+      "loss": 0.7898,
+      "step": 6720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.9991864e-05,
+      "loss": 0.802,
+      "step": 6780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.9991792e-05,
+      "loss": 0.798,
+      "step": 6840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.9991720000000005e-05,
+      "loss": 0.8134,
+      "step": 6900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.9991648e-05,
+      "loss": 0.8006,
+      "step": 6960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.9991576e-05,
+      "loss": 0.7945,
+      "step": 7020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.9991504e-05,
+      "loss": 0.7959,
+      "step": 7080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9991432000000004e-05,
+      "loss": 0.7983,
+      "step": 7140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.999136e-05,
+      "loss": 0.7968,
+      "step": 7200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.9991288e-05,
+      "loss": 0.7914,
+      "step": 7260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.9991216e-05,
+      "loss": 0.8027,
+      "step": 7320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.9991144e-05,
+      "loss": 0.7931,
+      "step": 7380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.9991072000000006e-05,
+      "loss": 0.7944,
+      "step": 7440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.9991e-05,
+      "loss": 0.7978,
+      "step": 7500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.9990928e-05,
+      "loss": 0.7839,
+      "step": 7560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.9990856e-05,
+      "loss": 0.803,
+      "step": 7620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.9990784000000005e-05,
+      "loss": 0.7868,
+      "step": 7680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.9990712e-05,
+      "loss": 0.7728,
+      "step": 7740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.999064e-05,
+      "loss": 0.7898,
+      "step": 7800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.9990568e-05,
+      "loss": 0.7804,
+      "step": 7860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.9990496000000003e-05,
+      "loss": 0.7929,
+      "step": 7920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.9990424e-05,
+      "loss": 0.7698,
+      "step": 7980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9990352e-05,
+      "loss": 0.79,
+      "step": 8040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.999028e-05,
+      "loss": 0.7941,
+      "step": 8100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9990208e-05,
+      "loss": 0.7694,
+      "step": 8160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9990136000000005e-05,
+      "loss": 0.7749,
+      "step": 8220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9990064e-05,
+      "loss": 0.7827,
+      "step": 8280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9989992e-05,
+      "loss": 0.7852,
+      "step": 8340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.998992000000001e-05,
+      "loss": 0.7828,
+      "step": 8400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9989848000000004e-05,
+      "loss": 0.7742,
+      "step": 8460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9989776e-05,
+      "loss": 0.7679,
+      "step": 8520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9989704e-05,
+      "loss": 0.7717,
+      "step": 8580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9989632e-05,
+      "loss": 0.7767,
+      "step": 8640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.998956e-05,
+      "loss": 0.778,
+      "step": 8700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9989488e-05,
+      "loss": 0.7703,
+      "step": 8760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9989416e-05,
+      "loss": 0.7637,
+      "step": 8820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9989344e-05,
+      "loss": 0.7665,
+      "step": 8880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9989272e-05,
+      "loss": 0.7708,
+      "step": 8940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.9989200000000005e-05,
+      "loss": 0.783,
+      "step": 9000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.9989128e-05,
+      "loss": 0.7839,
+      "step": 9060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.9989056e-05,
+      "loss": 0.7663,
+      "step": 9120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.998898400000001e-05,
+      "loss": 0.7664,
+      "step": 9180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.9988912000000004e-05,
+      "loss": 0.7694,
+      "step": 9240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.998884e-05,
+      "loss": 0.7658,
+      "step": 9300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9988767999999996e-05,
+      "loss": 0.7564,
+      "step": 9360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9988696e-05,
+      "loss": 0.7785,
+      "step": 9420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9988624e-05,
+      "loss": 0.7709,
+      "step": 9480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9988552e-05,
+      "loss": 0.7658,
+      "step": 9540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.998848e-05,
+      "loss": 0.7912,
+      "step": 9600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.9988408e-05,
+      "loss": 0.7777,
+      "step": 9660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.9988336e-05,
+      "loss": 0.7559,
+      "step": 9720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.9988264000000005e-05,
+      "loss": 0.763,
+      "step": 9780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.9988192e-05,
+      "loss": 0.7614,
+      "step": 9840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.998812e-05,
+      "loss": 0.7457,
+      "step": 9900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.998804800000001e-05,
+      "loss": 0.7683,
+      "step": 9960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9987976e-05,
+      "loss": 0.7604,
+      "step": 10020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9987904e-05,
+      "loss": 0.7683,
+      "step": 10080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.9987832e-05,
+      "loss": 0.7655,
+      "step": 10140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.9987760000000006e-05,
+      "loss": 0.762,
+      "step": 10200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.9987688e-05,
+      "loss": 0.753,
+      "step": 10260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9987616e-05,
+      "loss": 0.7625,
+      "step": 10320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9987544e-05,
+      "loss": 0.7487,
+      "step": 10380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.9987472e-05,
+      "loss": 0.7526,
+      "step": 10440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.99874e-05,
+      "loss": 0.7593,
+      "step": 10500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.9987328000000004e-05,
+      "loss": 0.7648,
+      "step": 10560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.9987256e-05,
+      "loss": 0.7574,
+      "step": 10620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.9987184e-05,
+      "loss": 0.7488,
+      "step": 10680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.9987112000000007e-05,
+      "loss": 0.7594,
+      "step": 10740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.998704e-05,
+      "loss": 0.7509,
+      "step": 10800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.9986968e-05,
+      "loss": 0.7422,
+      "step": 10860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.9986896e-05,
+      "loss": 0.7344,
+      "step": 10920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.9986824000000005e-05,
+      "loss": 0.7528,
+      "step": 10980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.9986752e-05,
+      "loss": 0.7547,
+      "step": 11040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.998668e-05,
+      "loss": 0.7416,
+      "step": 11100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.9986608e-05,
+      "loss": 0.7649,
+      "step": 11160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9986536e-05,
+      "loss": 0.7482,
+      "step": 11220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9986464e-05,
+      "loss": 0.7305,
+      "step": 11280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.9986392000000004e-05,
+      "loss": 0.7371,
+      "step": 11340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.998632e-05,
+      "loss": 0.7407,
+      "step": 11400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.9986247999999996e-05,
+      "loss": 0.728,
+      "step": 11460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.9986176000000006e-05,
+      "loss": 0.7559,
+      "step": 11520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.9986104e-05,
+      "loss": 0.7428,
+      "step": 11580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9986032e-05,
+      "loss": 0.734,
+      "step": 11640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.998596e-05,
+      "loss": 0.7402,
+      "step": 11700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9985888000000005e-05,
+      "loss": 0.7427,
+      "step": 11760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9985816e-05,
+      "loss": 0.7442,
+      "step": 11820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.9985744000000004e-05,
+      "loss": 0.7385,
+      "step": 11880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.9985672e-05,
+      "loss": 0.7491,
+      "step": 11940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.9985600000000004e-05,
+      "loss": 0.7514,
+      "step": 12000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.9985528e-05,
+      "loss": 0.7364,
+      "step": 12060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.9985456e-05,
+      "loss": 0.7375,
+      "step": 12120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.9985384e-05,
+      "loss": 0.7472,
+      "step": 12180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.9985311999999996e-05,
+      "loss": 0.7363,
+      "step": 12240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.9985240000000006e-05,
+      "loss": 0.7564,
+      "step": 12300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.9985168e-05,
+      "loss": 0.7414,
+      "step": 12360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.9985096e-05,
+      "loss": 0.7326,
+      "step": 12420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.9985024e-05,
+      "loss": 0.7518,
+      "step": 12480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9984952000000005e-05,
+      "loss": 0.7297,
+      "step": 12540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.998488e-05,
+      "loss": 0.719,
+      "step": 12600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.9984808000000004e-05,
+      "loss": 0.7326,
+      "step": 12660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.9984736e-05,
+      "loss": 0.7373,
+      "step": 12720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.9984664000000003e-05,
+      "loss": 0.7291,
+      "step": 12780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9984592000000006e-05,
+      "loss": 0.7546,
+      "step": 12840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.998452e-05,
+      "loss": 0.7359,
+      "step": 12900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.9984448e-05,
+      "loss": 0.723,
+      "step": 12960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.9984375999999995e-05,
+      "loss": 0.7463,
+      "step": 13020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9984304000000005e-05,
+      "loss": 0.7288,
+      "step": 13080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9984232e-05,
+      "loss": 0.714,
+      "step": 13140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.998416e-05,
+      "loss": 0.7358,
+      "step": 13200
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.6823992729187012,
+      "eval_runtime": 338.539,
+      "eval_samples_per_second": 295.387,
+      "eval_steps_per_second": 2.31,
+      "step": 13245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.9984088e-05,
+      "loss": 0.7322,
+      "step": 13260
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.9984016000000004e-05,
+      "loss": 0.7234,
+      "step": 13320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.9983944e-05,
+      "loss": 0.7239,
+      "step": 13380
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.9983872000000004e-05,
+      "loss": 0.7261,
+      "step": 13440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.99838e-05,
+      "loss": 0.7221,
+      "step": 13500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.9983728e-05,
+      "loss": 0.7171,
+      "step": 13560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.9983656000000006e-05,
+      "loss": 0.7191,
+      "step": 13620
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.9983584e-05,
+      "loss": 0.7295,
+      "step": 13680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.9983512e-05,
+      "loss": 0.7121,
+      "step": 13740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.9983439999999995e-05,
+      "loss": 0.7221,
+      "step": 13800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.9983368000000005e-05,
+      "loss": 0.7249,
+      "step": 13860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.9983296e-05,
+      "loss": 0.718,
+      "step": 13920
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.9983224e-05,
+      "loss": 0.7343,
+      "step": 13980
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.9983152e-05,
+      "loss": 0.7224,
+      "step": 14040
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.9983080000000004e-05,
+      "loss": 0.7246,
+      "step": 14100
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.9983008e-05,
+      "loss": 0.7299,
+      "step": 14160
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.9982936e-05,
+      "loss": 0.7202,
+      "step": 14220
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.9982864e-05,
+      "loss": 0.7202,
+      "step": 14280
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.9982792e-05,
+      "loss": 0.7131,
+      "step": 14340
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.9982720000000006e-05,
+      "loss": 0.7226,
+      "step": 14400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.9982648e-05,
+      "loss": 0.7207,
+      "step": 14460
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.9982576e-05,
+      "loss": 0.7174,
+      "step": 14520
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.9982504e-05,
+      "loss": 0.7156,
+      "step": 14580
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.9982432000000004e-05,
+      "loss": 0.7155,
+      "step": 14640
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.998236e-05,
+      "loss": 0.7232,
+      "step": 14700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.9982288e-05,
+      "loss": 0.7145,
+      "step": 14760
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.9982216e-05,
+      "loss": 0.7088,
+      "step": 14820
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.9982144e-05,
+      "loss": 0.7274,
+      "step": 14880
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.9982072e-05,
+      "loss": 0.7279,
+      "step": 14940
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.9982e-05,
+      "loss": 0.7256,
+      "step": 15000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.9981928e-05,
+      "loss": 0.7104,
+      "step": 15060
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.9981856e-05,
+      "loss": 0.7111,
+      "step": 15120
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.9981784000000005e-05,
+      "loss": 0.7323,
+      "step": 15180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.9981712e-05,
+      "loss": 0.7114,
+      "step": 15240
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.998164e-05,
+      "loss": 0.707,
+      "step": 15300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.998156800000001e-05,
+      "loss": 0.7059,
+      "step": 15360
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.9981496000000004e-05,
+      "loss": 0.7111,
+      "step": 15420
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.9981424e-05,
+      "loss": 0.6973,
+      "step": 15480
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.9981352e-05,
+      "loss": 0.7343,
+      "step": 15540
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.998128e-05,
+      "loss": 0.7219,
+      "step": 15600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.9981208e-05,
+      "loss": 0.6944,
+      "step": 15660
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.9981136e-05,
+      "loss": 0.7073,
+      "step": 15720
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.9981064e-05,
+      "loss": 0.7248,
+      "step": 15780
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.9980992e-05,
+      "loss": 0.7098,
+      "step": 15840
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.998092e-05,
+      "loss": 0.7124,
+      "step": 15900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.9980848000000005e-05,
+      "loss": 0.7129,
+      "step": 15960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.9980776e-05,
+      "loss": 0.7221,
+      "step": 16020
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.9980704e-05,
+      "loss": 0.7014,
+      "step": 16080
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.998063200000001e-05,
+      "loss": 0.7111,
+      "step": 16140
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.9980560000000004e-05,
+      "loss": 0.71,
+      "step": 16200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.9980488e-05,
+      "loss": 0.7108,
+      "step": 16260
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.9980416e-05,
+      "loss": 0.7084,
+      "step": 16320
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.9980344e-05,
+      "loss": 0.7016,
+      "step": 16380
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.9980272e-05,
+      "loss": 0.7113,
+      "step": 16440
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.99802e-05,
+      "loss": 0.7068,
+      "step": 16500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.9980128e-05,
+      "loss": 0.7147,
+      "step": 16560
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.9980056e-05,
+      "loss": 0.713,
+      "step": 16620
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.9979984e-05,
+      "loss": 0.7161,
+      "step": 16680
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.9979912000000004e-05,
+      "loss": 0.6982,
+      "step": 16740
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.997984e-05,
+      "loss": 0.7029,
+      "step": 16800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.9979768e-05,
+      "loss": 0.6942,
+      "step": 16860
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.997969600000001e-05,
+      "loss": 0.7229,
+      "step": 16920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.9979624e-05,
+      "loss": 0.7194,
+      "step": 16980
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9979552e-05,
+      "loss": 0.7047,
+      "step": 17040
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.997948e-05,
+      "loss": 0.7058,
+      "step": 17100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.9979408e-05,
+      "loss": 0.722,
+      "step": 17160
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.9979336e-05,
+      "loss": 0.7262,
+      "step": 17220
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.9979264000000005e-05,
+      "loss": 0.7159,
+      "step": 17280
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.9979192e-05,
+      "loss": 0.7058,
+      "step": 17340
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.997912e-05,
+      "loss": 0.729,
+      "step": 17400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.9979048e-05,
+      "loss": 0.713,
+      "step": 17460
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.9978976000000004e-05,
+      "loss": 0.7048,
+      "step": 17520
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.9978904e-05,
+      "loss": 0.7098,
+      "step": 17580
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.9978831999999997e-05,
+      "loss": 0.7109,
+      "step": 17640
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.9978760000000006e-05,
+      "loss": 0.6913,
+      "step": 17700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.9978688e-05,
+      "loss": 0.6954,
+      "step": 17760
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.9978616e-05,
+      "loss": 0.7038,
+      "step": 17820
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.9978544e-05,
+      "loss": 0.6868,
+      "step": 17880
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.9978472000000005e-05,
+      "loss": 0.7034,
+      "step": 17940
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.99784e-05,
+      "loss": 0.7128,
+      "step": 18000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.9978328000000005e-05,
+      "loss": 0.6995,
+      "step": 18060
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.9978256e-05,
+      "loss": 0.7158,
+      "step": 18120
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.9978184e-05,
+      "loss": 0.7059,
+      "step": 18180
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.9978112e-05,
+      "loss": 0.7026,
+      "step": 18240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.9978040000000004e-05,
+      "loss": 0.7001,
+      "step": 18300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.9977968e-05,
+      "loss": 0.7042,
+      "step": 18360
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.9977895999999996e-05,
+      "loss": 0.6947,
+      "step": 18420
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.9977824000000006e-05,
+      "loss": 0.6909,
+      "step": 18480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.9977752e-05,
+      "loss": 0.7113,
+      "step": 18540
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.997768e-05,
+      "loss": 0.7108,
+      "step": 18600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.9977608e-05,
+      "loss": 0.7088,
+      "step": 18660
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.9977536000000005e-05,
+      "loss": 0.686,
+      "step": 18720
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.9977464e-05,
+      "loss": 0.7014,
+      "step": 18780
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.9977392000000004e-05,
+      "loss": 0.7099,
+      "step": 18840
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.997732e-05,
+      "loss": 0.7062,
+      "step": 18900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.9977248e-05,
+      "loss": 0.6911,
+      "step": 18960
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.997717600000001e-05,
+      "loss": 0.6927,
+      "step": 19020
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.9977104e-05,
+      "loss": 0.6832,
+      "step": 19080
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.9977032e-05,
+      "loss": 0.6994,
+      "step": 19140
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.9976959999999996e-05,
+      "loss": 0.7017,
+      "step": 19200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.9976888000000006e-05,
+      "loss": 0.6809,
+      "step": 19260
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.9976816e-05,
+      "loss": 0.6927,
+      "step": 19320
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.9976744e-05,
+      "loss": 0.692,
+      "step": 19380
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.9976672e-05,
+      "loss": 0.7024,
+      "step": 19440
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.9976600000000004e-05,
+      "loss": 0.6924,
+      "step": 19500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.9976528e-05,
+      "loss": 0.698,
+      "step": 19560
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.9976456000000004e-05,
+      "loss": 0.712,
+      "step": 19620
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.9976384e-05,
+      "loss": 0.6885,
+      "step": 19680
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.9976311999999996e-05,
+      "loss": 0.6847,
+      "step": 19740
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.9976240000000006e-05,
+      "loss": 0.7033,
+      "step": 19800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.9976168e-05,
+      "loss": 0.6882,
+      "step": 19860
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.9976096e-05,
+      "loss": 0.6874,
+      "step": 19920
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.9976023999999995e-05,
+      "loss": 0.6991,
+      "step": 19980
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.9975952000000005e-05,
+      "loss": 0.6871,
+      "step": 20040
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.997588e-05,
+      "loss": 0.7113,
+      "step": 20100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.9975808e-05,
+      "loss": 0.6897,
+      "step": 20160
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.9975736e-05,
+      "loss": 0.6992,
+      "step": 20220
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.9975664000000004e-05,
+      "loss": 0.6903,
+      "step": 20280
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.9975592e-05,
+      "loss": 0.6822,
+      "step": 20340
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.9975520000000003e-05,
+      "loss": 0.6973,
+      "step": 20400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.9975448e-05,
+      "loss": 0.7057,
+      "step": 20460
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.9975376e-05,
+      "loss": 0.6994,
+      "step": 20520
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.9975304000000006e-05,
+      "loss": 0.7048,
+      "step": 20580
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.9975232e-05,
+      "loss": 0.6902,
+      "step": 20640
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.997516e-05,
+      "loss": 0.6974,
+      "step": 20700
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.9975088e-05,
+      "loss": 0.6899,
+      "step": 20760
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.9975016000000005e-05,
+      "loss": 0.6905,
+      "step": 20820
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.9974944e-05,
+      "loss": 0.7006,
+      "step": 20880
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.9974872e-05,
+      "loss": 0.693,
+      "step": 20940
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.99748e-05,
+      "loss": 0.6897,
+      "step": 21000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.9974728000000004e-05,
+      "loss": 0.6829,
+      "step": 21060
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.9974656e-05,
+      "loss": 0.7091,
+      "step": 21120
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.9974584e-05,
+      "loss": 0.7012,
+      "step": 21180
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.9974512e-05,
+      "loss": 0.6814,
+      "step": 21240
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.997444e-05,
+      "loss": 0.6801,
+      "step": 21300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.9974368000000005e-05,
+      "loss": 0.6956,
+      "step": 21360
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.9974296e-05,
+      "loss": 0.6945,
+      "step": 21420
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.9974224e-05,
+      "loss": 0.7011,
+      "step": 21480
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.9974152e-05,
+      "loss": 0.6843,
+      "step": 21540
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.9974080000000004e-05,
+      "loss": 0.6873,
+      "step": 21600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.9974008e-05,
+      "loss": 0.6932,
+      "step": 21660
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.9973936000000004e-05,
+      "loss": 0.6895,
+      "step": 21720
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.9973864e-05,
+      "loss": 0.7045,
+      "step": 21780
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.9973792e-05,
+      "loss": 0.6971,
+      "step": 21840
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.997372e-05,
+      "loss": 0.6963,
+      "step": 21900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.9973648e-05,
+      "loss": 0.6842,
+      "step": 21960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.9973576e-05,
+      "loss": 0.6939,
+      "step": 22020
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.9973504e-05,
+      "loss": 0.6828,
+      "step": 22080
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.9973432000000005e-05,
+      "loss": 0.688,
+      "step": 22140
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.997336e-05,
+      "loss": 0.6801,
+      "step": 22200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.9973288e-05,
+      "loss": 0.681,
+      "step": 22260
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.9973216e-05,
+      "loss": 0.6882,
+      "step": 22320
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.9973144000000004e-05,
+      "loss": 0.6953,
+      "step": 22380
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.9973072e-05,
+      "loss": 0.6971,
+      "step": 22440
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.9973e-05,
+      "loss": 0.6865,
+      "step": 22500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.9972928e-05,
+      "loss": 0.6867,
+      "step": 22560
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.9972856e-05,
+      "loss": 0.6833,
+      "step": 22620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.9972784e-05,
+      "loss": 0.6835,
+      "step": 22680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.9972712e-05,
+      "loss": 0.6931,
+      "step": 22740
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.997264e-05,
+      "loss": 0.6945,
+      "step": 22800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.9972568e-05,
+      "loss": 0.6746,
+      "step": 22860
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.9972496000000005e-05,
+      "loss": 0.6923,
+      "step": 22920
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.9972424e-05,
+      "loss": 0.6877,
+      "step": 22980
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.9972352e-05,
+      "loss": 0.6924,
+      "step": 23040
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.997228000000001e-05,
+      "loss": 0.6685,
+      "step": 23100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.9972208000000003e-05,
+      "loss": 0.6774,
+      "step": 23160
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.9972136e-05,
+      "loss": 0.6905,
+      "step": 23220
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.9972064e-05,
+      "loss": 0.6846,
+      "step": 23280
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.9971992e-05,
+      "loss": 0.6854,
+      "step": 23340
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.997192e-05,
+      "loss": 0.6903,
+      "step": 23400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.9971848000000005e-05,
+      "loss": 0.692,
+      "step": 23460
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.9971776e-05,
+      "loss": 0.6843,
+      "step": 23520
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.9971704e-05,
+      "loss": 0.6771,
+      "step": 23580
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.9971632e-05,
+      "loss": 0.6741,
+      "step": 23640
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.9971560000000004e-05,
+      "loss": 0.6756,
+      "step": 23700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.9971488e-05,
+      "loss": 0.675,
+      "step": 23760
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.9971416e-05,
+      "loss": 0.6786,
+      "step": 23820
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.997134400000001e-05,
+      "loss": 0.6768,
+      "step": 23880
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.9971272e-05,
+      "loss": 0.6762,
+      "step": 23940
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.99712e-05,
+      "loss": 0.6804,
+      "step": 24000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.9971128e-05,
+      "loss": 0.6882,
+      "step": 24060
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.9971056e-05,
+      "loss": 0.6827,
+      "step": 24120
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.9970984e-05,
+      "loss": 0.6819,
+      "step": 24180
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.9970912000000005e-05,
+      "loss": 0.6813,
+      "step": 24240
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.997084e-05,
+      "loss": 0.6791,
+      "step": 24300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.9970768e-05,
+      "loss": 0.6718,
+      "step": 24360
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.9970696e-05,
+      "loss": 0.6925,
+      "step": 24420
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.9970624000000004e-05,
+      "loss": 0.6834,
+      "step": 24480
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.9970552e-05,
+      "loss": 0.6745,
+      "step": 24540
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.9970479999999996e-05,
+      "loss": 0.6744,
+      "step": 24600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.9970408000000006e-05,
+      "loss": 0.6808,
+      "step": 24660
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.9970336e-05,
+      "loss": 0.6731,
+      "step": 24720
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.9970264e-05,
+      "loss": 0.6817,
+      "step": 24780
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9970192e-05,
+      "loss": 0.671,
+      "step": 24840
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.997012e-05,
+      "loss": 0.6776,
+      "step": 24900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9970048e-05,
+      "loss": 0.6636,
+      "step": 24960
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.9969976000000005e-05,
+      "loss": 0.6903,
+      "step": 25020
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.9969904e-05,
+      "loss": 0.6851,
+      "step": 25080
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.9969832e-05,
+      "loss": 0.6763,
+      "step": 25140
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.996976000000001e-05,
+      "loss": 0.6774,
+      "step": 25200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.9969688e-05,
+      "loss": 0.6845,
+      "step": 25260
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.9969616e-05,
+      "loss": 0.6876,
+      "step": 25320
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.9969543999999996e-05,
+      "loss": 0.6632,
+      "step": 25380
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.9969472000000006e-05,
+      "loss": 0.6601,
+      "step": 25440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.99694e-05,
+      "loss": 0.6789,
+      "step": 25500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9969328e-05,
+      "loss": 0.6783,
+      "step": 25560
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9969256e-05,
+      "loss": 0.6849,
+      "step": 25620
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9969184000000005e-05,
+      "loss": 0.6644,
+      "step": 25680
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9969112e-05,
+      "loss": 0.6633,
+      "step": 25740
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.9969040000000004e-05,
+      "loss": 0.6709,
+      "step": 25800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.9968968e-05,
+      "loss": 0.6801,
+      "step": 25860
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.9968896e-05,
+      "loss": 0.6572,
+      "step": 25920
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.996882400000001e-05,
+      "loss": 0.671,
+      "step": 25980
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.9968752e-05,
+      "loss": 0.6818,
+      "step": 26040
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.996868e-05,
+      "loss": 0.6689,
+      "step": 26100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.9968608e-05,
+      "loss": 0.6863,
+      "step": 26160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.9968536000000005e-05,
+      "loss": 0.677,
+      "step": 26220
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.9968464e-05,
+      "loss": 0.6798,
+      "step": 26280
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.9968392e-05,
+      "loss": 0.6711,
+      "step": 26340
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.996832e-05,
+      "loss": 0.692,
+      "step": 26400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.9968248000000004e-05,
+      "loss": 0.6759,
+      "step": 26460
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.6304686069488525,
+      "eval_runtime": 339.6799,
+      "eval_samples_per_second": 294.395,
+      "eval_steps_per_second": 2.302,
+      "step": 26490
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.9968176e-05,
+      "loss": 0.6702,
+      "step": 26520
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.9968104000000004e-05,
+      "loss": 0.6584,
+      "step": 26580
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.9968032e-05,
+      "loss": 0.6718,
+      "step": 26640
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.9967959999999996e-05,
+      "loss": 0.6686,
+      "step": 26700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.9967888000000006e-05,
+      "loss": 0.6551,
+      "step": 26760
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.9967816e-05,
+      "loss": 0.6716,
+      "step": 26820
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.9967744e-05,
+      "loss": 0.6739,
+      "step": 26880
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.9967672e-05,
+      "loss": 0.6621,
+      "step": 26940
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.9967600000000005e-05,
+      "loss": 0.6755,
+      "step": 27000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.9967528e-05,
+      "loss": 0.6701,
+      "step": 27060
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.9967456e-05,
+      "loss": 0.67,
+      "step": 27120
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.9967384e-05,
+      "loss": 0.6553,
+      "step": 27180
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.9967312000000004e-05,
+      "loss": 0.6751,
+      "step": 27240
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.996724e-05,
+      "loss": 0.6609,
+      "step": 27300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.9967168e-05,
+      "loss": 0.6705,
+      "step": 27360
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.9967096e-05,
+      "loss": 0.6653,
+      "step": 27420
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.9967023999999996e-05,
+      "loss": 0.6738,
+      "step": 27480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.9966952000000006e-05,
+      "loss": 0.6755,
+      "step": 27540
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.996688e-05,
+      "loss": 0.6721,
+      "step": 27600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.9966808e-05,
+      "loss": 0.6522,
+      "step": 27660
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.9966736e-05,
+      "loss": 0.6683,
+      "step": 27720
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.9966664000000005e-05,
+      "loss": 0.6654,
+      "step": 27780
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.9966592e-05,
+      "loss": 0.6645,
+      "step": 27840
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9966520000000004e-05,
+      "loss": 0.6523,
+      "step": 27900
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9966448e-05,
+      "loss": 0.6684,
+      "step": 27960
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.9966376000000003e-05,
+      "loss": 0.6562,
+      "step": 28020
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.9966304e-05,
+      "loss": 0.6689,
+      "step": 28080
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.9966232e-05,
+      "loss": 0.654,
+      "step": 28140
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.996616e-05,
+      "loss": 0.6565,
+      "step": 28200
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.9966088e-05,
+      "loss": 0.6572,
+      "step": 28260
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.9966016000000005e-05,
+      "loss": 0.6693,
+      "step": 28320
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.9965944e-05,
+      "loss": 0.665,
+      "step": 28380
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.9965872e-05,
+      "loss": 0.6561,
+      "step": 28440
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.99658e-05,
+      "loss": 0.6662,
+      "step": 28500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.9965728000000004e-05,
+      "loss": 0.6771,
+      "step": 28560
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.9965656e-05,
+      "loss": 0.6525,
+      "step": 28620
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.9965584000000004e-05,
+      "loss": 0.6739,
+      "step": 28680
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.9965512e-05,
+      "loss": 0.6604,
+      "step": 28740
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.996544e-05,
+      "loss": 0.6556,
+      "step": 28800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.9965368e-05,
+      "loss": 0.6709,
+      "step": 28860
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.9965296e-05,
+      "loss": 0.6594,
+      "step": 28920
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.9965224e-05,
+      "loss": 0.6688,
+      "step": 28980
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.9965152e-05,
+      "loss": 0.6597,
+      "step": 29040
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.9965080000000005e-05,
+      "loss": 0.6599,
+      "step": 29100
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.9965008e-05,
+      "loss": 0.6638,
+      "step": 29160
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.9964936e-05,
+      "loss": 0.6642,
+      "step": 29220
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.9964864e-05,
+      "loss": 0.6624,
+      "step": 29280
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.9964792000000004e-05,
+      "loss": 0.6693,
+      "step": 29340
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.996472e-05,
+      "loss": 0.6688,
+      "step": 29400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.9964648e-05,
+      "loss": 0.6659,
+      "step": 29460
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.9964576e-05,
+      "loss": 0.6575,
+      "step": 29520
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.9964504e-05,
+      "loss": 0.6646,
+      "step": 29580
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.9964432000000006e-05,
+      "loss": 0.6665,
+      "step": 29640
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.996436e-05,
+      "loss": 0.6603,
+      "step": 29700
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.9964288e-05,
+      "loss": 0.6514,
+      "step": 29760
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.9964216e-05,
+      "loss": 0.6591,
+      "step": 29820
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.9964144000000004e-05,
+      "loss": 0.6671,
+      "step": 29880
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.9964072e-05,
+      "loss": 0.6413,
+      "step": 29940
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.9964e-05,
+      "loss": 0.6458,
+      "step": 30000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.9963928e-05,
+      "loss": 0.6527,
+      "step": 30060
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.9963856e-05,
+      "loss": 0.6715,
+      "step": 30120
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.9963784e-05,
+      "loss": 0.6625,
+      "step": 30180
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.9963712e-05,
+      "loss": 0.6449,
+      "step": 30240
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.996364e-05,
+      "loss": 0.6707,
+      "step": 30300
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.9963568e-05,
+      "loss": 0.6742,
+      "step": 30360
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.9963496000000005e-05,
+      "loss": 0.6749,
+      "step": 30420
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.9963424e-05,
+      "loss": 0.6528,
+      "step": 30480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.9963352e-05,
+      "loss": 0.662,
+      "step": 30540
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.996328000000001e-05,
+      "loss": 0.663,
+      "step": 30600
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.9963208000000004e-05,
+      "loss": 0.6633,
+      "step": 30660
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.9963136e-05,
+      "loss": 0.6746,
+      "step": 30720
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.9963064e-05,
+      "loss": 0.6711,
+      "step": 30780
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.9962992e-05,
+      "loss": 0.6673,
+      "step": 30840
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.996292e-05,
+      "loss": 0.6504,
+      "step": 30900
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.9962848e-05,
+      "loss": 0.6711,
+      "step": 30960
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.9962776e-05,
+      "loss": 0.6554,
+      "step": 31020
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.9962704e-05,
+      "loss": 0.6507,
+      "step": 31080
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.9962632e-05,
+      "loss": 0.6635,
+      "step": 31140
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.9962560000000005e-05,
+      "loss": 0.671,
+      "step": 31200
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.9962488e-05,
+      "loss": 0.6607,
+      "step": 31260
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.9962416e-05,
+      "loss": 0.673,
+      "step": 31320
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.996234400000001e-05,
+      "loss": 0.6578,
+      "step": 31380
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.9962272000000004e-05,
+      "loss": 0.6483,
+      "step": 31440
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.99622e-05,
+      "loss": 0.6557,
+      "step": 31500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.9962127999999996e-05,
+      "loss": 0.6432,
+      "step": 31560
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.9962056000000006e-05,
+      "loss": 0.654,
+      "step": 31620
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.9961984e-05,
+      "loss": 0.6546,
+      "step": 31680
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.9961912e-05,
+      "loss": 0.654,
+      "step": 31740
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.996184e-05,
+      "loss": 0.6462,
+      "step": 31800
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.9961768e-05,
+      "loss": 0.6677,
+      "step": 31860
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.9961696e-05,
+      "loss": 0.6562,
+      "step": 31920
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.9961624000000004e-05,
+      "loss": 0.6454,
+      "step": 31980
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.9961552e-05,
+      "loss": 0.66,
+      "step": 32040
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.996148e-05,
+      "loss": 0.6623,
+      "step": 32100
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.996140800000001e-05,
+      "loss": 0.6524,
+      "step": 32160
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.9961336e-05,
+      "loss": 0.655,
+      "step": 32220
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.9961264e-05,
+      "loss": 0.6531,
+      "step": 32280
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.9961192e-05,
+      "loss": 0.6634,
+      "step": 32340
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.9961120000000006e-05,
+      "loss": 0.6454,
+      "step": 32400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.9961048e-05,
+      "loss": 0.6593,
+      "step": 32460
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.9960976e-05,
+      "loss": 0.6455,
+      "step": 32520
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.9960904e-05,
+      "loss": 0.6584,
+      "step": 32580
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.9960832e-05,
+      "loss": 0.6396,
+      "step": 32640
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.996076e-05,
+      "loss": 0.6506,
+      "step": 32700
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.9960688000000004e-05,
+      "loss": 0.6572,
+      "step": 32760
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.9960616e-05,
+      "loss": 0.6574,
+      "step": 32820
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.9960543999999997e-05,
+      "loss": 0.6605,
+      "step": 32880
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.9960472000000006e-05,
+      "loss": 0.6422,
+      "step": 32940
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.99604e-05,
+      "loss": 0.6668,
+      "step": 33000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9960328e-05,
+      "loss": 0.6497,
+      "step": 33060
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9960256e-05,
+      "loss": 0.6652,
+      "step": 33120
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.9960184000000005e-05,
+      "loss": 0.6649,
+      "step": 33180
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.9960112e-05,
+      "loss": 0.6562,
+      "step": 33240
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.996004e-05,
+      "loss": 0.6517,
+      "step": 33300
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.9959968e-05,
+      "loss": 0.6489,
+      "step": 33360
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.9959896e-05,
+      "loss": 0.6579,
+      "step": 33420
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.9959824e-05,
+      "loss": 0.6519,
+      "step": 33480
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.9959752000000004e-05,
+      "loss": 0.6481,
+      "step": 33540
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.995968e-05,
+      "loss": 0.6625,
+      "step": 33600
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.9959607999999996e-05,
+      "loss": 0.6598,
+      "step": 33660
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.9959536000000006e-05,
+      "loss": 0.6423,
+      "step": 33720
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.9959464e-05,
+      "loss": 0.6534,
+      "step": 33780
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.9959392e-05,
+      "loss": 0.6484,
+      "step": 33840
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.995932e-05,
+      "loss": 0.6522,
+      "step": 33900
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.9959248000000005e-05,
+      "loss": 0.6486,
+      "step": 33960
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.9959176e-05,
+      "loss": 0.6546,
+      "step": 34020
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.9959104000000004e-05,
+      "loss": 0.6656,
+      "step": 34080
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.9959032e-05,
+      "loss": 0.6532,
+      "step": 34140
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.9958960000000004e-05,
+      "loss": 0.6557,
+      "step": 34200
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.9958888e-05,
+      "loss": 0.6442,
+      "step": 34260
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.9958816e-05,
+      "loss": 0.6521,
+      "step": 34320
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.9958744e-05,
+      "loss": 0.6526,
+      "step": 34380
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.9958671999999996e-05,
+      "loss": 0.6364,
+      "step": 34440
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.9958600000000006e-05,
+      "loss": 0.6536,
+      "step": 34500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.9958528e-05,
+      "loss": 0.6661,
+      "step": 34560
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.9958456e-05,
+      "loss": 0.6349,
+      "step": 34620
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.9958384e-05,
+      "loss": 0.6366,
+      "step": 34680
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.9958312000000004e-05,
+      "loss": 0.6632,
+      "step": 34740
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.995824e-05,
+      "loss": 0.6511,
+      "step": 34800
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.9958168000000004e-05,
+      "loss": 0.6542,
+      "step": 34860
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.9958096e-05,
+      "loss": 0.6465,
+      "step": 34920
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.9958024e-05,
+      "loss": 0.644,
+      "step": 34980
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.9957952000000006e-05,
+      "loss": 0.6445,
+      "step": 35040
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.995788e-05,
+      "loss": 0.643,
+      "step": 35100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.9957808e-05,
+      "loss": 0.652,
+      "step": 35160
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.9957735999999995e-05,
+      "loss": 0.6375,
+      "step": 35220
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.9957664000000005e-05,
+      "loss": 0.642,
+      "step": 35280
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.9957592e-05,
+      "loss": 0.6422,
+      "step": 35340
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.995752e-05,
+      "loss": 0.6552,
+      "step": 35400
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.9957448e-05,
+      "loss": 0.6568,
+      "step": 35460
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.9957376000000004e-05,
+      "loss": 0.6429,
+      "step": 35520
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.9957304e-05,
+      "loss": 0.6411,
+      "step": 35580
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.9957232000000003e-05,
+      "loss": 0.6687,
+      "step": 35640
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.995716e-05,
+      "loss": 0.6598,
+      "step": 35700
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.9957088e-05,
+      "loss": 0.6478,
+      "step": 35760
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.9957016000000006e-05,
+      "loss": 0.6422,
+      "step": 35820
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.9956944e-05,
+      "loss": 0.6482,
+      "step": 35880
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.9956872e-05,
+      "loss": 0.655,
+      "step": 35940
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.99568e-05,
+      "loss": 0.6461,
+      "step": 36000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.9956728000000005e-05,
+      "loss": 0.655,
+      "step": 36060
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.9956656e-05,
+      "loss": 0.6643,
+      "step": 36120
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.9956584e-05,
+      "loss": 0.6349,
+      "step": 36180
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.9956512e-05,
+      "loss": 0.6512,
+      "step": 36240
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.9956440000000004e-05,
+      "loss": 0.6477,
+      "step": 36300
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.9956368e-05,
+      "loss": 0.6471,
+      "step": 36360
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.9956296e-05,
+      "loss": 0.6455,
+      "step": 36420
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.9956224e-05,
+      "loss": 0.6387,
+      "step": 36480
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.9956152e-05,
+      "loss": 0.6482,
+      "step": 36540
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.9956080000000006e-05,
+      "loss": 0.6553,
+      "step": 36600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.9956008e-05,
+      "loss": 0.6485,
+      "step": 36660
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.9955936e-05,
+      "loss": 0.6458,
+      "step": 36720
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.995586400000001e-05,
+      "loss": 0.641,
+      "step": 36780
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.9955792000000004e-05,
+      "loss": 0.6466,
+      "step": 36840
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.995572e-05,
+      "loss": 0.6493,
+      "step": 36900
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.9955648e-05,
+      "loss": 0.6465,
+      "step": 36960
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.9955576e-05,
+      "loss": 0.6345,
+      "step": 37020
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.9955504e-05,
+      "loss": 0.6472,
+      "step": 37080
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.9955432e-05,
+      "loss": 0.6526,
+      "step": 37140
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.995536e-05,
+      "loss": 0.6605,
+      "step": 37200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.9955288e-05,
+      "loss": 0.6597,
+      "step": 37260
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.9955216e-05,
+      "loss": 0.6428,
+      "step": 37320
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.9955144000000005e-05,
+      "loss": 0.6491,
+      "step": 37380
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.9955072e-05,
+      "loss": 0.6408,
+      "step": 37440
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.9955e-05,
+      "loss": 0.6314,
+      "step": 37500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.995492800000001e-05,
+      "loss": 0.6494,
+      "step": 37560
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.9954856000000004e-05,
+      "loss": 0.6493,
+      "step": 37620
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.9954784e-05,
+      "loss": 0.6515,
+      "step": 37680
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.9954711999999997e-05,
+      "loss": 0.6491,
+      "step": 37740
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.995464e-05,
+      "loss": 0.6523,
+      "step": 37800
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.9954568e-05,
+      "loss": 0.6524,
+      "step": 37860
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.9954496e-05,
+      "loss": 0.6483,
+      "step": 37920
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.9954424e-05,
+      "loss": 0.6337,
+      "step": 37980
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.9954352e-05,
+      "loss": 0.6486,
+      "step": 38040
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.995428e-05,
+      "loss": 0.6491,
+      "step": 38100
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.9954208000000005e-05,
+      "loss": 0.649,
+      "step": 38160
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.9954136e-05,
+      "loss": 0.6485,
+      "step": 38220
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.9954064e-05,
+      "loss": 0.6528,
+      "step": 38280
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.995399200000001e-05,
+      "loss": 0.6455,
+      "step": 38340
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.9953920000000003e-05,
+      "loss": 0.6312,
+      "step": 38400
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.9953848e-05,
+      "loss": 0.6569,
+      "step": 38460
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.9953776e-05,
+      "loss": 0.6436,
+      "step": 38520
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.9953704e-05,
+      "loss": 0.6427,
+      "step": 38580
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.9953632e-05,
+      "loss": 0.639,
+      "step": 38640
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.995356e-05,
+      "loss": 0.6542,
+      "step": 38700
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.9953488e-05,
+      "loss": 0.6524,
+      "step": 38760
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.9953416e-05,
+      "loss": 0.6515,
+      "step": 38820
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.9953344e-05,
+      "loss": 0.6283,
+      "step": 38880
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.9953272000000004e-05,
+      "loss": 0.656,
+      "step": 38940
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.99532e-05,
+      "loss": 0.6476,
+      "step": 39000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.9953128e-05,
+      "loss": 0.651,
+      "step": 39060
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.995305600000001e-05,
+      "loss": 0.6431,
+      "step": 39120
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.9952984e-05,
+      "loss": 0.6448,
+      "step": 39180
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.9952912e-05,
+      "loss": 0.6377,
+      "step": 39240
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.995284e-05,
+      "loss": 0.6469,
+      "step": 39300
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.9952768000000006e-05,
+      "loss": 0.6399,
+      "step": 39360
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.9952696e-05,
+      "loss": 0.6365,
+      "step": 39420
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.9952624000000005e-05,
+      "loss": 0.6346,
+      "step": 39480
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.9952552e-05,
+      "loss": 0.647,
+      "step": 39540
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.995248e-05,
+      "loss": 0.6414,
+      "step": 39600
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.9952408e-05,
+      "loss": 0.6308,
+      "step": 39660
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.9952336000000004e-05,
+      "loss": 0.6505,
+      "step": 39720
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.60313481092453,
+      "eval_runtime": 338.5244,
+      "eval_samples_per_second": 295.4,
+      "eval_steps_per_second": 2.31,
+      "step": 39735
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.9952264e-05,
+      "loss": 0.6396,
+      "step": 39780
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.9952191999999996e-05,
+      "loss": 0.6311,
+      "step": 39840
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.9952120000000006e-05,
+      "loss": 0.6478,
+      "step": 39900
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.9952048e-05,
+      "loss": 0.6335,
+      "step": 39960
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.9951976e-05,
+      "loss": 0.6318,
+      "step": 40020
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.9951904e-05,
+      "loss": 0.6384,
+      "step": 40080
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.9951832000000005e-05,
+      "loss": 0.633,
+      "step": 40140
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.995176e-05,
+      "loss": 0.6345,
+      "step": 40200
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.9951688000000005e-05,
+      "loss": 0.6392,
+      "step": 40260
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.9951616e-05,
+      "loss": 0.6314,
+      "step": 40320
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.9951544e-05,
+      "loss": 0.64,
+      "step": 40380
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.9951472e-05,
+      "loss": 0.6352,
+      "step": 40440
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.99514e-05,
+      "loss": 0.6519,
+      "step": 40500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.9951328e-05,
+      "loss": 0.6145,
+      "step": 40560
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.9951255999999996e-05,
+      "loss": 0.6442,
+      "step": 40620
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.9951184000000006e-05,
+      "loss": 0.6402,
+      "step": 40680
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.9951112e-05,
+      "loss": 0.6464,
+      "step": 40740
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.995104e-05,
+      "loss": 0.6344,
+      "step": 40800
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.9950968e-05,
+      "loss": 0.6261,
+      "step": 40860
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.9950896000000005e-05,
+      "loss": 0.6335,
+      "step": 40920
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.9950824e-05,
+      "loss": 0.6308,
+      "step": 40980
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.9950752000000004e-05,
+      "loss": 0.629,
+      "step": 41040
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.995068e-05,
+      "loss": 0.6299,
+      "step": 41100
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.9950608e-05,
+      "loss": 0.6239,
+      "step": 41160
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.995053600000001e-05,
+      "loss": 0.6467,
+      "step": 41220
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.9950464e-05,
+      "loss": 0.634,
+      "step": 41280
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.9950392e-05,
+      "loss": 0.6406,
+      "step": 41340
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.9950319999999996e-05,
+      "loss": 0.634,
+      "step": 41400
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.9950248000000005e-05,
+      "loss": 0.6471,
+      "step": 41460
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.9950176e-05,
+      "loss": 0.6598,
+      "step": 41520
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.9950104e-05,
+      "loss": 0.6457,
+      "step": 41580
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.9950032e-05,
+      "loss": 0.6438,
+      "step": 41640
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.9949960000000004e-05,
+      "loss": 0.6257,
+      "step": 41700
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.9949888e-05,
+      "loss": 0.6234,
+      "step": 41760
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.9949816000000004e-05,
+      "loss": 0.6368,
+      "step": 41820
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.9949744e-05,
+      "loss": 0.6355,
+      "step": 41880
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.9949672e-05,
+      "loss": 0.6212,
+      "step": 41940
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.9949600000000006e-05,
+      "loss": 0.6424,
+      "step": 42000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.9949528e-05,
+      "loss": 0.6374,
+      "step": 42060
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.9949456e-05,
+      "loss": 0.6323,
+      "step": 42120
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.9949383999999995e-05,
+      "loss": 0.6326,
+      "step": 42180
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.9949312000000005e-05,
+      "loss": 0.6331,
+      "step": 42240
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.994924e-05,
+      "loss": 0.6299,
+      "step": 42300
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.9949168e-05,
+      "loss": 0.6315,
+      "step": 42360
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.9949096e-05,
+      "loss": 0.641,
+      "step": 42420
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9949024000000004e-05,
+      "loss": 0.6422,
+      "step": 42480
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9948952e-05,
+      "loss": 0.6289,
+      "step": 42540
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.994888e-05,
+      "loss": 0.6355,
+      "step": 42600
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9948808e-05,
+      "loss": 0.6364,
+      "step": 42660
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.9948736e-05,
+      "loss": 0.6311,
+      "step": 42720
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.9948664000000006e-05,
+      "loss": 0.6352,
+      "step": 42780
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.9948592e-05,
+      "loss": 0.6299,
+      "step": 42840
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.994852e-05,
+      "loss": 0.6311,
+      "step": 42900
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.9948448e-05,
+      "loss": 0.6367,
+      "step": 42960
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.9948376000000005e-05,
+      "loss": 0.6342,
+      "step": 43020
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.9948304e-05,
+      "loss": 0.6399,
+      "step": 43080
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.9948232e-05,
+      "loss": 0.6335,
+      "step": 43140
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.994816e-05,
+      "loss": 0.6391,
+      "step": 43200
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.9948088000000003e-05,
+      "loss": 0.6321,
+      "step": 43260
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.9948016e-05,
+      "loss": 0.6329,
+      "step": 43320
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.9947944e-05,
+      "loss": 0.6365,
+      "step": 43380
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.9947872e-05,
+      "loss": 0.6435,
+      "step": 43440
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.99478e-05,
+      "loss": 0.6156,
+      "step": 43500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.9947728000000005e-05,
+      "loss": 0.629,
+      "step": 43560
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.9947656e-05,
+      "loss": 0.6275,
+      "step": 43620
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.9947584e-05,
+      "loss": 0.6492,
+      "step": 43680
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.9947512e-05,
+      "loss": 0.6258,
+      "step": 43740
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.9947440000000004e-05,
+      "loss": 0.6333,
+      "step": 43800
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.9947368e-05,
+      "loss": 0.641,
+      "step": 43860
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.9947296000000004e-05,
+      "loss": 0.6242,
+      "step": 43920
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.9947224e-05,
+      "loss": 0.6332,
+      "step": 43980
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.9947152e-05,
+      "loss": 0.647,
+      "step": 44040
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.994708e-05,
+      "loss": 0.6403,
+      "step": 44100
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.9947008e-05,
+      "loss": 0.6428,
+      "step": 44160
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.9946936e-05,
+      "loss": 0.6369,
+      "step": 44220
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.9946864e-05,
+      "loss": 0.6345,
+      "step": 44280
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.9946792000000005e-05,
+      "loss": 0.6235,
+      "step": 44340
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.994672e-05,
+      "loss": 0.6261,
+      "step": 44400
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.9946648e-05,
+      "loss": 0.6231,
+      "step": 44460
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.9946576e-05,
+      "loss": 0.6348,
+      "step": 44520
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.9946504000000004e-05,
+      "loss": 0.6262,
+      "step": 44580
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.9946432e-05,
+      "loss": 0.6325,
+      "step": 44640
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.994636e-05,
+      "loss": 0.6308,
+      "step": 44700
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.9946288e-05,
+      "loss": 0.6292,
+      "step": 44760
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.9946216e-05,
+      "loss": 0.6365,
+      "step": 44820
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.9946144e-05,
+      "loss": 0.6075,
+      "step": 44880
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.9946072e-05,
+      "loss": 0.6316,
+      "step": 44940
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.9946e-05,
+      "loss": 0.6216,
+      "step": 45000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.9945928e-05,
+      "loss": 0.624,
+      "step": 45060
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.9945856000000004e-05,
+      "loss": 0.6371,
+      "step": 45120
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.9945784e-05,
+      "loss": 0.6215,
+      "step": 45180
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.9945712e-05,
+      "loss": 0.6304,
+      "step": 45240
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.994564000000001e-05,
+      "loss": 0.6337,
+      "step": 45300
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.9945568e-05,
+      "loss": 0.6418,
+      "step": 45360
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.9945496e-05,
+      "loss": 0.6377,
+      "step": 45420
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.9945424e-05,
+      "loss": 0.6416,
+      "step": 45480
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.9945352e-05,
+      "loss": 0.617,
+      "step": 45540
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.994528e-05,
+      "loss": 0.6221,
+      "step": 45600
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5.9945208000000005e-05,
+      "loss": 0.6255,
+      "step": 45660
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5.9945136e-05,
+      "loss": 0.6387,
+      "step": 45720
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5.9945064e-05,
+      "loss": 0.6311,
+      "step": 45780
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5.9944992e-05,
+      "loss": 0.6332,
+      "step": 45840
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5.9944920000000004e-05,
+      "loss": 0.6233,
+      "step": 45900
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5.9944848e-05,
+      "loss": 0.6395,
+      "step": 45960
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5.9944776e-05,
+      "loss": 0.6296,
+      "step": 46020
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5.9944704000000007e-05,
+      "loss": 0.6276,
+      "step": 46080
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5.9944632e-05,
+      "loss": 0.636,
+      "step": 46140
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5.994456e-05,
+      "loss": 0.6169,
+      "step": 46200
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5.9944488e-05,
+      "loss": 0.6233,
+      "step": 46260
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.9944416e-05,
+      "loss": 0.6232,
+      "step": 46320
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.9944344e-05,
+      "loss": 0.6362,
+      "step": 46380
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.9944272000000005e-05,
+      "loss": 0.6218,
+      "step": 46440
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.99442e-05,
+      "loss": 0.6356,
+      "step": 46500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.9944128e-05,
+      "loss": 0.6118,
+      "step": 46560
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.9944056e-05,
+      "loss": 0.6275,
+      "step": 46620
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.9943984000000004e-05,
+      "loss": 0.6309,
+      "step": 46680
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.9943912e-05,
+      "loss": 0.6195,
+      "step": 46740
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.9943839999999996e-05,
+      "loss": 0.6202,
+      "step": 46800
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9943768000000006e-05,
+      "loss": 0.6276,
+      "step": 46860
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9943696e-05,
+      "loss": 0.6316,
+      "step": 46920
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9943624e-05,
+      "loss": 0.6294,
+      "step": 46980
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9943552e-05,
+      "loss": 0.6335,
+      "step": 47040
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.9943480000000005e-05,
+      "loss": 0.6182,
+      "step": 47100
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.9943408e-05,
+      "loss": 0.629,
+      "step": 47160
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.9943336000000004e-05,
+      "loss": 0.6342,
+      "step": 47220
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.9943264e-05,
+      "loss": 0.6322,
+      "step": 47280
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.9943192e-05,
+      "loss": 0.6268,
+      "step": 47340
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.994312000000001e-05,
+      "loss": 0.6227,
+      "step": 47400
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.9943048e-05,
+      "loss": 0.6227,
+      "step": 47460
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.9942976e-05,
+      "loss": 0.6187,
+      "step": 47520
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.9942903999999996e-05,
+      "loss": 0.6237,
+      "step": 47580
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.9942832000000006e-05,
+      "loss": 0.6254,
+      "step": 47640
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.994276e-05,
+      "loss": 0.6272,
+      "step": 47700
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.9942688e-05,
+      "loss": 0.6162,
+      "step": 47760
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.9942616e-05,
+      "loss": 0.6334,
+      "step": 47820
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.9942544000000005e-05,
+      "loss": 0.6321,
+      "step": 47880
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.9942472e-05,
+      "loss": 0.6285,
+      "step": 47940
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.9942400000000004e-05,
+      "loss": 0.631,
+      "step": 48000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.9942328e-05,
+      "loss": 0.6264,
+      "step": 48060
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.9942255999999997e-05,
+      "loss": 0.6396,
+      "step": 48120
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.9942184000000006e-05,
+      "loss": 0.6345,
+      "step": 48180
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.9942112e-05,
+      "loss": 0.6193,
+      "step": 48240
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.994204e-05,
+      "loss": 0.612,
+      "step": 48300
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.9941968e-05,
+      "loss": 0.6058,
+      "step": 48360
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.9941896000000005e-05,
+      "loss": 0.6218,
+      "step": 48420
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.9941824e-05,
+      "loss": 0.6287,
+      "step": 48480
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.9941752e-05,
+      "loss": 0.6241,
+      "step": 48540
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.994168e-05,
+      "loss": 0.62,
+      "step": 48600
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.9941608000000004e-05,
+      "loss": 0.6175,
+      "step": 48660
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.9941536e-05,
+      "loss": 0.6195,
+      "step": 48720
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.9941464000000004e-05,
+      "loss": 0.6185,
+      "step": 48780
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.9941392e-05,
+      "loss": 0.626,
+      "step": 48840
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.9941319999999996e-05,
+      "loss": 0.6155,
+      "step": 48900
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.9941248000000006e-05,
+      "loss": 0.6159,
+      "step": 48960
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.9941176e-05,
+      "loss": 0.6263,
+      "step": 49020
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.9941104e-05,
+      "loss": 0.621,
+      "step": 49080
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.9941032e-05,
+      "loss": 0.6268,
+      "step": 49140
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.9940960000000005e-05,
+      "loss": 0.6225,
+      "step": 49200
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.9940888e-05,
+      "loss": 0.6362,
+      "step": 49260
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.9940816e-05,
+      "loss": 0.6103,
+      "step": 49320
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.9940744e-05,
+      "loss": 0.6185,
+      "step": 49380
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.9940672000000004e-05,
+      "loss": 0.6218,
+      "step": 49440
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.99406e-05,
+      "loss": 0.6188,
+      "step": 49500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.9940528e-05,
+      "loss": 0.6314,
+      "step": 49560
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.9940456e-05,
+      "loss": 0.6121,
+      "step": 49620
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.9940384e-05,
+      "loss": 0.6189,
+      "step": 49680
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 5.9940312000000006e-05,
+      "loss": 0.6254,
+      "step": 49740
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 5.994024e-05,
+      "loss": 0.62,
+      "step": 49800
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 5.9940168e-05,
+      "loss": 0.6318,
+      "step": 49860
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5.9940096e-05,
+      "loss": 0.6232,
+      "step": 49920
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5.9940024000000004e-05,
+      "loss": 0.6145,
+      "step": 49980
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 5.9939952e-05,
+      "loss": 0.6267,
+      "step": 50040
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 5.9939880000000004e-05,
+      "loss": 0.6204,
+      "step": 50100
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 5.9939808e-05,
+      "loss": 0.62,
+      "step": 50160
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 5.9939736e-05,
+      "loss": 0.6281,
+      "step": 50220
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5.9939664e-05,
+      "loss": 0.6223,
+      "step": 50280
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5.9939592e-05,
+      "loss": 0.6308,
+      "step": 50340
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5.993952e-05,
+      "loss": 0.637,
+      "step": 50400
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5.9939448e-05,
+      "loss": 0.6139,
+      "step": 50460
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5.9939376000000005e-05,
+      "loss": 0.6219,
+      "step": 50520
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5.9939304e-05,
+      "loss": 0.6225,
+      "step": 50580
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5.9939232e-05,
+      "loss": 0.623,
+      "step": 50640
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.993916e-05,
+      "loss": 0.6196,
+      "step": 50700
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.9939088000000004e-05,
+      "loss": 0.6198,
+      "step": 50760
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.9939016e-05,
+      "loss": 0.6133,
+      "step": 50820
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.9938944000000003e-05,
+      "loss": 0.63,
+      "step": 50880
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.9938872e-05,
+      "loss": 0.6242,
+      "step": 50940
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.99388e-05,
+      "loss": 0.6344,
+      "step": 51000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.9938728e-05,
+      "loss": 0.6119,
+      "step": 51060
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.9938656e-05,
+      "loss": 0.6137,
+      "step": 51120
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.9938584e-05,
+      "loss": 0.6223,
+      "step": 51180
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.9938512e-05,
+      "loss": 0.6276,
+      "step": 51240
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.9938440000000005e-05,
+      "loss": 0.6314,
+      "step": 51300
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.9938368e-05,
+      "loss": 0.6172,
+      "step": 51360
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.9938296e-05,
+      "loss": 0.621,
+      "step": 51420
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.9938224e-05,
+      "loss": 0.6131,
+      "step": 51480
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.9938152000000004e-05,
+      "loss": 0.6138,
+      "step": 51540
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.993808e-05,
+      "loss": 0.619,
+      "step": 51600
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.9938008e-05,
+      "loss": 0.6062,
+      "step": 51660
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.9937936e-05,
+      "loss": 0.6168,
+      "step": 51720
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.9937864e-05,
+      "loss": 0.625,
+      "step": 51780
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.9937792000000006e-05,
+      "loss": 0.6193,
+      "step": 51840
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.993772e-05,
+      "loss": 0.61,
+      "step": 51900
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.9937648e-05,
+      "loss": 0.6218,
+      "step": 51960
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.9937576e-05,
+      "loss": 0.6248,
+      "step": 52020
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.9937504000000004e-05,
+      "loss": 0.6125,
+      "step": 52080
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.9937432e-05,
+      "loss": 0.631,
+      "step": 52140
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.993736e-05,
+      "loss": 0.6121,
+      "step": 52200
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.9937288e-05,
+      "loss": 0.6254,
+      "step": 52260
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.9937216e-05,
+      "loss": 0.6245,
+      "step": 52320
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.9937144e-05,
+      "loss": 0.6332,
+      "step": 52380
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.9937072e-05,
+      "loss": 0.6206,
+      "step": 52440
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.9937e-05,
+      "loss": 0.6181,
+      "step": 52500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.9936928e-05,
+      "loss": 0.6265,
+      "step": 52560
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.9936856000000005e-05,
+      "loss": 0.622,
+      "step": 52620
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.9936784e-05,
+      "loss": 0.628,
+      "step": 52680
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.9936712e-05,
+      "loss": 0.6213,
+      "step": 52740
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.993664000000001e-05,
+      "loss": 0.6229,
+      "step": 52800
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.9936568000000004e-05,
+      "loss": 0.6177,
+      "step": 52860
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.9936496e-05,
+      "loss": 0.6188,
+      "step": 52920
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.9936423999999997e-05,
+      "loss": 0.617,
+      "step": 52980
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 0.5853202939033508,
+      "eval_runtime": 337.718,
+      "eval_samples_per_second": 296.105,
+      "eval_steps_per_second": 2.316,
+      "step": 52980
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.9936352000000006e-05,
+      "loss": 0.6146,
+      "step": 53040
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.993628e-05,
+      "loss": 0.6101,
+      "step": 53100
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.9936208e-05,
+      "loss": 0.6137,
+      "step": 53160
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.9936136e-05,
+      "loss": 0.627,
+      "step": 53220
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.9936064e-05,
+      "loss": 0.6259,
+      "step": 53280
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.9935992e-05,
+      "loss": 0.6176,
+      "step": 53340
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.9935920000000005e-05,
+      "loss": 0.6107,
+      "step": 53400
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.9935848e-05,
+      "loss": 0.6131,
+      "step": 53460
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.9935776e-05,
+      "loss": 0.6066,
+      "step": 53520
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.993570400000001e-05,
+      "loss": 0.6153,
+      "step": 53580
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.9935632000000003e-05,
+      "loss": 0.6267,
+      "step": 53640
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.993556e-05,
+      "loss": 0.6112,
+      "step": 53700
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.9935487999999996e-05,
+      "loss": 0.6191,
+      "step": 53760
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.9935416000000006e-05,
+      "loss": 0.6286,
+      "step": 53820
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.9935344e-05,
+      "loss": 0.6203,
+      "step": 53880
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.9935272e-05,
+      "loss": 0.6165,
+      "step": 53940
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5.99352e-05,
+      "loss": 0.6054,
+      "step": 54000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5.9935128e-05,
+      "loss": 0.617,
+      "step": 54060
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.9935056e-05,
+      "loss": 0.6207,
+      "step": 54120
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.9934984000000004e-05,
+      "loss": 0.6132,
+      "step": 54180
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.9934912e-05,
+      "loss": 0.6157,
+      "step": 54240
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.993484e-05,
+      "loss": 0.6141,
+      "step": 54300
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.993476800000001e-05,
+      "loss": 0.6167,
+      "step": 54360
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.9934696e-05,
+      "loss": 0.611,
+      "step": 54420
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.9934624e-05,
+      "loss": 0.6081,
+      "step": 54480
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.9934552e-05,
+      "loss": 0.6263,
+      "step": 54540
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.9934480000000006e-05,
+      "loss": 0.6162,
+      "step": 54600
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5.9934408e-05,
+      "loss": 0.6106,
+      "step": 54660
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5.9934336e-05,
+      "loss": 0.62,
+      "step": 54720
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5.9934264e-05,
+      "loss": 0.6054,
+      "step": 54780
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5.9934192e-05,
+      "loss": 0.6121,
+      "step": 54840
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5.993412e-05,
+      "loss": 0.6164,
+      "step": 54900
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5.9934048000000004e-05,
+      "loss": 0.6188,
+      "step": 54960
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5.9933976e-05,
+      "loss": 0.61,
+      "step": 55020
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.9933903999999996e-05,
+      "loss": 0.6161,
+      "step": 55080
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.9933832000000006e-05,
+      "loss": 0.6067,
+      "step": 55140
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5.993376e-05,
+      "loss": 0.6,
+      "step": 55200
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5.9933688e-05,
+      "loss": 0.6144,
+      "step": 55260
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5.9933616e-05,
+      "loss": 0.6085,
+      "step": 55320
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5.9933544000000005e-05,
+      "loss": 0.6278,
+      "step": 55380
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.9933472e-05,
+      "loss": 0.6108,
+      "step": 55440
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.99334e-05,
+      "loss": 0.6203,
+      "step": 55500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.9933328e-05,
+      "loss": 0.6107,
+      "step": 55560
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5.9933256000000004e-05,
+      "loss": 0.6105,
+      "step": 55620
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5.9933184e-05,
+      "loss": 0.6239,
+      "step": 55680
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.9933112000000003e-05,
+      "loss": 0.6145,
+      "step": 55740
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.993304e-05,
+      "loss": 0.6135,
+      "step": 55800
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.9932967999999996e-05,
+      "loss": 0.6032,
+      "step": 55860
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.9932896000000006e-05,
+      "loss": 0.6047,
+      "step": 55920
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.9932824e-05,
+      "loss": 0.6104,
+      "step": 55980
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.9932752e-05,
+      "loss": 0.6119,
+      "step": 56040
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.993268e-05,
+      "loss": 0.6132,
+      "step": 56100
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.9932608000000005e-05,
+      "loss": 0.6061,
+      "step": 56160
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.9932536e-05,
+      "loss": 0.6274,
+      "step": 56220
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.9932464000000004e-05,
+      "loss": 0.6112,
+      "step": 56280
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.9932392e-05,
+      "loss": 0.6236,
+      "step": 56340
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.9932320000000004e-05,
+      "loss": 0.6004,
+      "step": 56400
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.9932248e-05,
+      "loss": 0.6065,
+      "step": 56460
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.9932176e-05,
+      "loss": 0.6039,
+      "step": 56520
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.9932104e-05,
+      "loss": 0.608,
+      "step": 56580
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.9932031999999996e-05,
+      "loss": 0.6132,
+      "step": 56640
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.9931960000000005e-05,
+      "loss": 0.6049,
+      "step": 56700
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.9931888e-05,
+      "loss": 0.6215,
+      "step": 56760
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.9931816e-05,
+      "loss": 0.623,
+      "step": 56820
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.9931744e-05,
+      "loss": 0.6257,
+      "step": 56880
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.9931672000000004e-05,
+      "loss": 0.5961,
+      "step": 56940
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.99316e-05,
+      "loss": 0.6128,
+      "step": 57000
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.9931528000000004e-05,
+      "loss": 0.6245,
+      "step": 57060
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.9931456e-05,
+      "loss": 0.6192,
+      "step": 57120
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.9931384e-05,
+      "loss": 0.6096,
+      "step": 57180
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.9931312000000006e-05,
+      "loss": 0.6255,
+      "step": 57240
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.993124e-05,
+      "loss": 0.6119,
+      "step": 57300
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.9931168e-05,
+      "loss": 0.6037,
+      "step": 57360
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.9931095999999995e-05,
+      "loss": 0.606,
+      "step": 57420
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.9931024000000005e-05,
+      "loss": 0.6059,
+      "step": 57480
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.9930952e-05,
+      "loss": 0.614,
+      "step": 57540
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.993088e-05,
+      "loss": 0.6036,
+      "step": 57600
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.9930808e-05,
+      "loss": 0.6214,
+      "step": 57660
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.9930736000000004e-05,
+      "loss": 0.6081,
+      "step": 57720
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.9930664e-05,
+      "loss": 0.6236,
+      "step": 57780
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.9930592e-05,
+      "loss": 0.6011,
+      "step": 57840
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.993052e-05,
+      "loss": 0.6212,
+      "step": 57900
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.9930448e-05,
+      "loss": 0.6054,
+      "step": 57960
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.9930376000000006e-05,
+      "loss": 0.5944,
+      "step": 58020
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.9930304e-05,
+      "loss": 0.6127,
+      "step": 58080
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.9930232e-05,
+      "loss": 0.6217,
+      "step": 58140
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.993016e-05,
+      "loss": 0.6095,
+      "step": 58200
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.9930088000000005e-05,
+      "loss": 0.6231,
+      "step": 58260
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.9930016e-05,
+      "loss": 0.6194,
+      "step": 58320
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.9929944e-05,
+      "loss": 0.6093,
+      "step": 58380
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.9929872e-05,
+      "loss": 0.6183,
+      "step": 58440
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.9929800000000003e-05,
+      "loss": 0.6171,
+      "step": 58500
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.9929728e-05,
+      "loss": 0.625,
+      "step": 58560
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.9929656e-05,
+      "loss": 0.6203,
+      "step": 58620
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.9929584e-05,
+      "loss": 0.6107,
+      "step": 58680
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.9929512e-05,
+      "loss": 0.5989,
+      "step": 58740
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.9929440000000005e-05,
+      "loss": 0.5933,
+      "step": 58800
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.9929368e-05,
+      "loss": 0.6191,
+      "step": 58860
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.9929296e-05,
+      "loss": 0.6147,
+      "step": 58920
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.992922400000001e-05,
+      "loss": 0.6249,
+      "step": 58980
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.9929152000000004e-05,
+      "loss": 0.6148,
+      "step": 59040
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.992908e-05,
+      "loss": 0.6036,
+      "step": 59100
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.9929008e-05,
+      "loss": 0.6136,
+      "step": 59160
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.9928936e-05,
+      "loss": 0.6098,
+      "step": 59220
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.9928864e-05,
+      "loss": 0.6076,
+      "step": 59280
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.9928792e-05,
+      "loss": 0.619,
+      "step": 59340
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.992872e-05,
+      "loss": 0.6106,
+      "step": 59400
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.9928648e-05,
+      "loss": 0.6103,
+      "step": 59460
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.9928576e-05,
+      "loss": 0.6068,
+      "step": 59520
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.9928504000000005e-05,
+      "loss": 0.6042,
+      "step": 59580
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.9928432e-05,
+      "loss": 0.611,
+      "step": 59640
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.992836e-05,
+      "loss": 0.6143,
+      "step": 59700
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.992828800000001e-05,
+      "loss": 0.6106,
+      "step": 59760
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.9928216000000004e-05,
+      "loss": 0.6212,
+      "step": 59820
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.9928144e-05,
+      "loss": 0.6157,
+      "step": 59880
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.9928071999999996e-05,
+      "loss": 0.6149,
+      "step": 59940
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.9928e-05,
+      "loss": 0.6099,
+      "step": 60000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.9927928e-05,
+      "loss": 0.6193,
+      "step": 60060
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.9927856e-05,
+      "loss": 0.6131,
+      "step": 60120
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.9927784e-05,
+      "loss": 0.6095,
+      "step": 60180
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.9927712e-05,
+      "loss": 0.6089,
+      "step": 60240
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.992764e-05,
+      "loss": 0.6218,
+      "step": 60300
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.9927568000000005e-05,
+      "loss": 0.6144,
+      "step": 60360
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.9927496e-05,
+      "loss": 0.6125,
+      "step": 60420
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.9927424e-05,
+      "loss": 0.6134,
+      "step": 60480
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.992735200000001e-05,
+      "loss": 0.6046,
+      "step": 60540
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.992728e-05,
+      "loss": 0.5963,
+      "step": 60600
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.9927208e-05,
+      "loss": 0.6006,
+      "step": 60660
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.9927136e-05,
+      "loss": 0.6011,
+      "step": 60720
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.9927064000000006e-05,
+      "loss": 0.6008,
+      "step": 60780
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.9926992e-05,
+      "loss": 0.6021,
+      "step": 60840
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.992692e-05,
+      "loss": 0.6111,
+      "step": 60900
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.9926848e-05,
+      "loss": 0.606,
+      "step": 60960
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.9926776e-05,
+      "loss": 0.6035,
+      "step": 61020
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.9926704e-05,
+      "loss": 0.6108,
+      "step": 61080
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.9926632000000004e-05,
+      "loss": 0.6027,
+      "step": 61140
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.992656e-05,
+      "loss": 0.6077,
+      "step": 61200
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.9926488e-05,
+      "loss": 0.6051,
+      "step": 61260
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.9926416000000007e-05,
+      "loss": 0.6091,
+      "step": 61320
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.9926344e-05,
+      "loss": 0.5969,
+      "step": 61380
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.9926272e-05,
+      "loss": 0.6132,
+      "step": 61440
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.99262e-05,
+      "loss": 0.6272,
+      "step": 61500
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.9926128000000005e-05,
+      "loss": 0.6204,
+      "step": 61560
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.9926056e-05,
+      "loss": 0.6178,
+      "step": 61620
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.9925984000000005e-05,
+      "loss": 0.6102,
+      "step": 61680
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.9925912e-05,
+      "loss": 0.6125,
+      "step": 61740
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.992584e-05,
+      "loss": 0.6055,
+      "step": 61800
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.9925768e-05,
+      "loss": 0.5978,
+      "step": 61860
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.9925696000000004e-05,
+      "loss": 0.6069,
+      "step": 61920
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.9925624e-05,
+      "loss": 0.6087,
+      "step": 61980
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.9925551999999996e-05,
+      "loss": 0.6163,
+      "step": 62040
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.9925480000000006e-05,
+      "loss": 0.6218,
+      "step": 62100
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.9925408e-05,
+      "loss": 0.6101,
+      "step": 62160
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.9925336e-05,
+      "loss": 0.6122,
+      "step": 62220
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.9925264e-05,
+      "loss": 0.6148,
+      "step": 62280
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.9925192000000005e-05,
+      "loss": 0.6066,
+      "step": 62340
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.992512e-05,
+      "loss": 0.6138,
+      "step": 62400
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.9925048000000004e-05,
+      "loss": 0.6071,
+      "step": 62460
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.9924976e-05,
+      "loss": 0.6175,
+      "step": 62520
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.9924904e-05,
+      "loss": 0.6083,
+      "step": 62580
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.9924832e-05,
+      "loss": 0.6173,
+      "step": 62640
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.992476e-05,
+      "loss": 0.6068,
+      "step": 62700
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.9924688e-05,
+      "loss": 0.6134,
+      "step": 62760
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.9924615999999996e-05,
+      "loss": 0.6055,
+      "step": 62820
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.9924544000000006e-05,
+      "loss": 0.6122,
+      "step": 62880
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.9924472e-05,
+      "loss": 0.6018,
+      "step": 62940
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.99244e-05,
+      "loss": 0.5996,
+      "step": 63000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.9924328e-05,
+      "loss": 0.6241,
+      "step": 63060
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.9924256000000005e-05,
+      "loss": 0.5985,
+      "step": 63120
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.9924184e-05,
+      "loss": 0.6062,
+      "step": 63180
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.9924112000000004e-05,
+      "loss": 0.5962,
+      "step": 63240
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.992404e-05,
+      "loss": 0.5966,
+      "step": 63300
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.9923968000000003e-05,
+      "loss": 0.5923,
+      "step": 63360
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.9923896000000006e-05,
+      "loss": 0.6207,
+      "step": 63420
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.9923824e-05,
+      "loss": 0.61,
+      "step": 63480
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.9923752e-05,
+      "loss": 0.6009,
+      "step": 63540
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.9923679999999995e-05,
+      "loss": 0.6125,
+      "step": 63600
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.9923608000000005e-05,
+      "loss": 0.5973,
+      "step": 63660
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.9923536e-05,
+      "loss": 0.6041,
+      "step": 63720
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.9923464e-05,
+      "loss": 0.6064,
+      "step": 63780
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.9923392e-05,
+      "loss": 0.5993,
+      "step": 63840
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.9923320000000004e-05,
+      "loss": 0.6038,
+      "step": 63900
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.9923248e-05,
+      "loss": 0.6001,
+      "step": 63960
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.9923176000000004e-05,
+      "loss": 0.6085,
+      "step": 64020
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.9923104e-05,
+      "loss": 0.6034,
+      "step": 64080
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.9923032e-05,
+      "loss": 0.6153,
+      "step": 64140
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.9922960000000006e-05,
+      "loss": 0.5958,
+      "step": 64200
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.9922888e-05,
+      "loss": 0.6085,
+      "step": 64260
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.9922816e-05,
+      "loss": 0.6181,
+      "step": 64320
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.9922743999999995e-05,
+      "loss": 0.6082,
+      "step": 64380
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.9922672000000005e-05,
+      "loss": 0.6192,
+      "step": 64440
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.99226e-05,
+      "loss": 0.5981,
+      "step": 64500
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.9922528e-05,
+      "loss": 0.6103,
+      "step": 64560
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.9922456e-05,
+      "loss": 0.6098,
+      "step": 64620
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.9922384000000004e-05,
+      "loss": 0.6172,
+      "step": 64680
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.9922312e-05,
+      "loss": 0.5967,
+      "step": 64740
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.992224e-05,
+      "loss": 0.6232,
+      "step": 64800
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.9922168e-05,
+      "loss": 0.6128,
+      "step": 64860
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.9922096e-05,
+      "loss": 0.6088,
+      "step": 64920
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.9922024000000006e-05,
+      "loss": 0.6025,
+      "step": 64980
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.9921952e-05,
+      "loss": 0.6145,
+      "step": 65040
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.992188e-05,
+      "loss": 0.6092,
+      "step": 65100
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.9921808e-05,
+      "loss": 0.6028,
+      "step": 65160
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.9921736000000004e-05,
+      "loss": 0.5969,
+      "step": 65220
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.9921664e-05,
+      "loss": 0.594,
+      "step": 65280
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.9921592e-05,
+      "loss": 0.6074,
+      "step": 65340
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.992152e-05,
+      "loss": 0.6072,
+      "step": 65400
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.9921448e-05,
+      "loss": 0.6098,
+      "step": 65460
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.9921376e-05,
+      "loss": 0.6043,
+      "step": 65520
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.9921304e-05,
+      "loss": 0.6058,
+      "step": 65580
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.9921232e-05,
+      "loss": 0.6024,
+      "step": 65640
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.992116e-05,
+      "loss": 0.6055,
+      "step": 65700
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.9921088000000005e-05,
+      "loss": 0.606,
+      "step": 65760
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.9921016e-05,
+      "loss": 0.6016,
+      "step": 65820
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.9920944e-05,
+      "loss": 0.61,
+      "step": 65880
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.9920872e-05,
+      "loss": 0.6045,
+      "step": 65940
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.9920800000000004e-05,
+      "loss": 0.6068,
+      "step": 66000
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.9920728e-05,
+      "loss": 0.6096,
+      "step": 66060
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.9920656000000003e-05,
+      "loss": 0.6188,
+      "step": 66120
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.9920584e-05,
+      "loss": 0.6049,
+      "step": 66180
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 0.5711187720298767,
+      "eval_runtime": 338.4117,
+      "eval_samples_per_second": 295.498,
+      "eval_steps_per_second": 2.311,
+      "step": 66225
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.9920512e-05,
+      "loss": 0.5971,
+      "step": 66240
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.992044e-05,
+      "loss": 0.6038,
+      "step": 66300
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.9920368e-05,
+      "loss": 0.6169,
+      "step": 66360
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.9920296e-05,
+      "loss": 0.5918,
+      "step": 66420
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.9920224e-05,
+      "loss": 0.5963,
+      "step": 66480
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.9920152000000005e-05,
+      "loss": 0.5814,
+      "step": 66540
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.992008e-05,
+      "loss": 0.5906,
+      "step": 66600
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.9920008e-05,
+      "loss": 0.6049,
+      "step": 66660
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 5.991993600000001e-05,
+      "loss": 0.5876,
+      "step": 66720
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 5.9919864000000004e-05,
+      "loss": 0.6058,
+      "step": 66780
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 5.9919792e-05,
+      "loss": 0.5979,
+      "step": 66840
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 5.991972e-05,
+      "loss": 0.6152,
+      "step": 66900
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 5.9919648e-05,
+      "loss": 0.5958,
+      "step": 66960
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 5.9919576e-05,
+      "loss": 0.5936,
+      "step": 67020
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 5.9919504e-05,
+      "loss": 0.6238,
+      "step": 67080
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 5.9919432e-05,
+      "loss": 0.5909,
+      "step": 67140
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 5.991936e-05,
+      "loss": 0.5929,
+      "step": 67200
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 5.9919288e-05,
+      "loss": 0.5922,
+      "step": 67260
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 5.9919216000000004e-05,
+      "loss": 0.5987,
+      "step": 67320
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 5.9919144e-05,
+      "loss": 0.6068,
+      "step": 67380
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 5.9919072e-05,
+      "loss": 0.6013,
+      "step": 67440
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 5.991900000000001e-05,
+      "loss": 0.594,
+      "step": 67500
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 5.9918928e-05,
+      "loss": 0.6135,
+      "step": 67560
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 5.9918856e-05,
+      "loss": 0.5962,
+      "step": 67620
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 5.9918784e-05,
+      "loss": 0.5978,
+      "step": 67680
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 5.9918712e-05,
+      "loss": 0.6028,
+      "step": 67740
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 5.991864e-05,
+      "loss": 0.598,
+      "step": 67800
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 5.9918568000000005e-05,
+      "loss": 0.5986,
+      "step": 67860
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 5.9918496e-05,
+      "loss": 0.5942,
+      "step": 67920
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 5.9918424e-05,
+      "loss": 0.5942,
+      "step": 67980
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 5.9918352e-05,
+      "loss": 0.5838,
+      "step": 68040
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 5.9918280000000004e-05,
+      "loss": 0.5969,
+      "step": 68100
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 5.9918208e-05,
+      "loss": 0.5888,
+      "step": 68160
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 5.9918135999999997e-05,
+      "loss": 0.6135,
+      "step": 68220
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 5.9918064000000006e-05,
+      "loss": 0.5873,
+      "step": 68280
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 5.9917992e-05,
+      "loss": 0.5871,
+      "step": 68340
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 5.991792e-05,
+      "loss": 0.6008,
+      "step": 68400
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 5.9917848e-05,
+      "loss": 0.5999,
+      "step": 68460
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 5.9917776000000005e-05,
+      "loss": 0.5977,
+      "step": 68520
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 5.9917704e-05,
+      "loss": 0.6002,
+      "step": 68580
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 5.9917632000000005e-05,
+      "loss": 0.6012,
+      "step": 68640
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 5.991756e-05,
+      "loss": 0.5976,
+      "step": 68700
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 5.9917488e-05,
+      "loss": 0.5938,
+      "step": 68760
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 5.9917416e-05,
+      "loss": 0.6128,
+      "step": 68820
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 5.9917344000000004e-05,
+      "loss": 0.604,
+      "step": 68880
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 5.9917272e-05,
+      "loss": 0.6021,
+      "step": 68940
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 5.9917199999999996e-05,
+      "loss": 0.5859,
+      "step": 69000
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 5.9917128000000006e-05,
+      "loss": 0.5943,
+      "step": 69060
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 5.9917056e-05,
+      "loss": 0.5914,
+      "step": 69120
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 5.9916984e-05,
+      "loss": 0.6021,
+      "step": 69180
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 5.9916912e-05,
+      "loss": 0.5977,
+      "step": 69240
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 5.9916840000000005e-05,
+      "loss": 0.5947,
+      "step": 69300
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 5.9916768e-05,
+      "loss": 0.601,
+      "step": 69360
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 5.9916696000000004e-05,
+      "loss": 0.6036,
+      "step": 69420
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 5.9916624e-05,
+      "loss": 0.5976,
+      "step": 69480
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 5.9916552e-05,
+      "loss": 0.6009,
+      "step": 69540
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 5.991648000000001e-05,
+      "loss": 0.6,
+      "step": 69600
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 5.9916408e-05,
+      "loss": 0.5974,
+      "step": 69660
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 5.9916336e-05,
+      "loss": 0.5969,
+      "step": 69720
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 5.9916263999999996e-05,
+      "loss": 0.5927,
+      "step": 69780
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 5.9916192000000006e-05,
+      "loss": 0.5953,
+      "step": 69840
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 5.991612e-05,
+      "loss": 0.5962,
+      "step": 69900
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 5.9916048e-05,
+      "loss": 0.6006,
+      "step": 69960
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 5.9915976e-05,
+      "loss": 0.6094,
+      "step": 70020
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 5.9915904000000004e-05,
+      "loss": 0.6063,
+      "step": 70080
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 5.9915832e-05,
+      "loss": 0.6011,
+      "step": 70140
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 5.9915760000000004e-05,
+      "loss": 0.6211,
+      "step": 70200
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 5.9915688e-05,
+      "loss": 0.5883,
+      "step": 70260
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 5.9915615999999996e-05,
+      "loss": 0.5907,
+      "step": 70320
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 5.9915544000000006e-05,
+      "loss": 0.5849,
+      "step": 70380
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 5.9915472e-05,
+      "loss": 0.5953,
+      "step": 70440
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 5.99154e-05,
+      "loss": 0.6026,
+      "step": 70500
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 5.9915328e-05,
+      "loss": 0.5977,
+      "step": 70560
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 5.9915256000000005e-05,
+      "loss": 0.5908,
+      "step": 70620
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 5.9915184e-05,
+      "loss": 0.5916,
+      "step": 70680
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 5.9915112e-05,
+      "loss": 0.6072,
+      "step": 70740
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.991504e-05,
+      "loss": 0.6063,
+      "step": 70800
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.9914968000000004e-05,
+      "loss": 0.6131,
+      "step": 70860
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.9914896e-05,
+      "loss": 0.606,
+      "step": 70920
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.9914824000000003e-05,
+      "loss": 0.6085,
+      "step": 70980
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.9914752e-05,
+      "loss": 0.5879,
+      "step": 71040
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.991468e-05,
+      "loss": 0.6012,
+      "step": 71100
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.9914608000000006e-05,
+      "loss": 0.5996,
+      "step": 71160
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.9914536e-05,
+      "loss": 0.609,
+      "step": 71220
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.9914464e-05,
+      "loss": 0.5972,
+      "step": 71280
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.9914392e-05,
+      "loss": 0.5933,
+      "step": 71340
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.9914320000000005e-05,
+      "loss": 0.6043,
+      "step": 71400
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.9914248e-05,
+      "loss": 0.5974,
+      "step": 71460
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.9914176e-05,
+      "loss": 0.6019,
+      "step": 71520
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.9914104e-05,
+      "loss": 0.5936,
+      "step": 71580
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 5.9914032000000004e-05,
+      "loss": 0.6024,
+      "step": 71640
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 5.991396e-05,
+      "loss": 0.5902,
+      "step": 71700
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 5.9913888e-05,
+      "loss": 0.5972,
+      "step": 71760
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 5.9913816e-05,
+      "loss": 0.6076,
+      "step": 71820
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 5.9913744e-05,
+      "loss": 0.5947,
+      "step": 71880
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 5.9913672000000005e-05,
+      "loss": 0.6034,
+      "step": 71940
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 5.99136e-05,
+      "loss": 0.578,
+      "step": 72000
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 5.9913528e-05,
+      "loss": 0.5964,
+      "step": 72060
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 5.9913456e-05,
+      "loss": 0.6042,
+      "step": 72120
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 5.9913384000000004e-05,
+      "loss": 0.5839,
+      "step": 72180
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 5.9913312e-05,
+      "loss": 0.592,
+      "step": 72240
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 5.9913240000000004e-05,
+      "loss": 0.5976,
+      "step": 72300
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 5.9913168e-05,
+      "loss": 0.6012,
+      "step": 72360
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 5.9913096e-05,
+      "loss": 0.608,
+      "step": 72420
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 5.9913024e-05,
+      "loss": 0.5921,
+      "step": 72480
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 5.9912952e-05,
+      "loss": 0.5931,
+      "step": 72540
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 5.991288e-05,
+      "loss": 0.5975,
+      "step": 72600
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 5.9912808e-05,
+      "loss": 0.5962,
+      "step": 72660
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 5.9912736000000005e-05,
+      "loss": 0.5797,
+      "step": 72720
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 5.9912664e-05,
+      "loss": 0.6069,
+      "step": 72780
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 5.9912592e-05,
+      "loss": 0.6101,
+      "step": 72840
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 5.991252e-05,
+      "loss": 0.5948,
+      "step": 72900
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 5.9912448000000004e-05,
+      "loss": 0.585,
+      "step": 72960
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 5.9912376e-05,
+      "loss": 0.5822,
+      "step": 73020
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 5.9912304e-05,
+      "loss": 0.5928,
+      "step": 73080
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 5.9912232e-05,
+      "loss": 0.5937,
+      "step": 73140
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 5.991216e-05,
+      "loss": 0.5935,
+      "step": 73200
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 5.9912088e-05,
+      "loss": 0.6086,
+      "step": 73260
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 5.9912016e-05,
+      "loss": 0.5861,
+      "step": 73320
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 5.9911944e-05,
+      "loss": 0.6044,
+      "step": 73380
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 5.9911872e-05,
+      "loss": 0.5965,
+      "step": 73440
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 5.9911800000000005e-05,
+      "loss": 0.6038,
+      "step": 73500
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 5.9911728e-05,
+      "loss": 0.6069,
+      "step": 73560
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 5.9911656e-05,
+      "loss": 0.5932,
+      "step": 73620
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 5.9911584e-05,
+      "loss": 0.5989,
+      "step": 73680
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 5.9911512000000003e-05,
+      "loss": 0.5797,
+      "step": 73740
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 5.991144e-05,
+      "loss": 0.5895,
+      "step": 73800
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 5.9911368e-05,
+      "loss": 0.5978,
+      "step": 73860
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 5.9911296e-05,
+      "loss": 0.6056,
+      "step": 73920
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 5.9911224e-05,
+      "loss": 0.5696,
+      "step": 73980
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 5.9911152000000005e-05,
+      "loss": 0.6003,
+      "step": 74040
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 5.991108e-05,
+      "loss": 0.5911,
+      "step": 74100
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 5.9911008e-05,
+      "loss": 0.5946,
+      "step": 74160
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 5.9910936e-05,
+      "loss": 0.5974,
+      "step": 74220
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 5.9910864000000004e-05,
+      "loss": 0.5955,
+      "step": 74280
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 5.9910792e-05,
+      "loss": 0.6124,
+      "step": 74340
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 5.991072e-05,
+      "loss": 0.5952,
+      "step": 74400
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 5.991064800000001e-05,
+      "loss": 0.5927,
+      "step": 74460
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 5.9910576e-05,
+      "loss": 0.5917,
+      "step": 74520
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 5.9910504e-05,
+      "loss": 0.592,
+      "step": 74580
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 5.9910432e-05,
+      "loss": 0.5965,
+      "step": 74640
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 5.991036e-05,
+      "loss": 0.6002,
+      "step": 74700
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 5.9910288e-05,
+      "loss": 0.5959,
+      "step": 74760
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 5.9910216000000005e-05,
+      "loss": 0.5998,
+      "step": 74820
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 5.9910144e-05,
+      "loss": 0.581,
+      "step": 74880
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 5.9910072e-05,
+      "loss": 0.5939,
+      "step": 74940
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 5.991000000000001e-05,
+      "loss": 0.5856,
+      "step": 75000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 5.9909928000000004e-05,
+      "loss": 0.5912,
+      "step": 75060
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 5.9909856e-05,
+      "loss": 0.6011,
+      "step": 75120
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 5.9909783999999996e-05,
+      "loss": 0.6042,
+      "step": 75180
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 5.9909712000000006e-05,
+      "loss": 0.6024,
+      "step": 75240
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 5.990964e-05,
+      "loss": 0.5982,
+      "step": 75300
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 5.9909568e-05,
+      "loss": 0.5922,
+      "step": 75360
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 5.9909496e-05,
+      "loss": 0.5983,
+      "step": 75420
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 5.9909424e-05,
+      "loss": 0.5912,
+      "step": 75480
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 5.9909352e-05,
+      "loss": 0.5958,
+      "step": 75540
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 5.9909280000000005e-05,
+      "loss": 0.5978,
+      "step": 75600
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 5.9909208e-05,
+      "loss": 0.5982,
+      "step": 75660
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 5.9909136e-05,
+      "loss": 0.5973,
+      "step": 75720
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 5.990906400000001e-05,
+      "loss": 0.5918,
+      "step": 75780
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 5.9908992e-05,
+      "loss": 0.5985,
+      "step": 75840
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 5.990892e-05,
+      "loss": 0.6067,
+      "step": 75900
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 5.9908847999999996e-05,
+      "loss": 0.5987,
+      "step": 75960
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 5.9908776000000006e-05,
+      "loss": 0.5952,
+      "step": 76020
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 5.9908704e-05,
+      "loss": 0.5845,
+      "step": 76080
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 5.9908632e-05,
+      "loss": 0.5931,
+      "step": 76140
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 5.990856e-05,
+      "loss": 0.5919,
+      "step": 76200
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 5.9908488e-05,
+      "loss": 0.598,
+      "step": 76260
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 5.9908416e-05,
+      "loss": 0.5918,
+      "step": 76320
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 5.9908344000000004e-05,
+      "loss": 0.6009,
+      "step": 76380
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 5.9908272e-05,
+      "loss": 0.5986,
+      "step": 76440
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 5.99082e-05,
+      "loss": 0.5821,
+      "step": 76500
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 5.9908128000000007e-05,
+      "loss": 0.5853,
+      "step": 76560
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 5.9908056e-05,
+      "loss": 0.5927,
+      "step": 76620
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 5.9907984e-05,
+      "loss": 0.5981,
+      "step": 76680
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 5.9907912e-05,
+      "loss": 0.5942,
+      "step": 76740
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.9907840000000005e-05,
+      "loss": 0.602,
+      "step": 76800
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.9907768e-05,
+      "loss": 0.6061,
+      "step": 76860
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.9907696e-05,
+      "loss": 0.5987,
+      "step": 76920
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.9907624e-05,
+      "loss": 0.5855,
+      "step": 76980
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 5.9907552000000004e-05,
+      "loss": 0.6024,
+      "step": 77040
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 5.990748e-05,
+      "loss": 0.6012,
+      "step": 77100
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 5.9907408000000004e-05,
+      "loss": 0.5963,
+      "step": 77160
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 5.9907336e-05,
+      "loss": 0.589,
+      "step": 77220
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 5.9907263999999996e-05,
+      "loss": 0.5946,
+      "step": 77280
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 5.9907192000000006e-05,
+      "loss": 0.5841,
+      "step": 77340
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 5.990712e-05,
+      "loss": 0.5998,
+      "step": 77400
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 5.9907048e-05,
+      "loss": 0.595,
+      "step": 77460
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 5.9906976e-05,
+      "loss": 0.5847,
+      "step": 77520
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 5.9906904000000005e-05,
+      "loss": 0.5911,
+      "step": 77580
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 5.9906832e-05,
+      "loss": 0.5893,
+      "step": 77640
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 5.990676e-05,
+      "loss": 0.5755,
+      "step": 77700
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 5.9906688e-05,
+      "loss": 0.5896,
+      "step": 77760
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 5.9906616000000004e-05,
+      "loss": 0.5859,
+      "step": 77820
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 5.9906544e-05,
+      "loss": 0.6098,
+      "step": 77880
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 5.9906472e-05,
+      "loss": 0.5881,
+      "step": 77940
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 5.99064e-05,
+      "loss": 0.5928,
+      "step": 78000
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 5.9906327999999996e-05,
+      "loss": 0.5871,
+      "step": 78060
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 5.9906256000000006e-05,
+      "loss": 0.5964,
+      "step": 78120
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 5.9906184e-05,
+      "loss": 0.6032,
+      "step": 78180
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 5.9906112e-05,
+      "loss": 0.601,
+      "step": 78240
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 5.990604e-05,
+      "loss": 0.5932,
+      "step": 78300
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 5.9905968000000005e-05,
+      "loss": 0.591,
+      "step": 78360
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 5.9905896e-05,
+      "loss": 0.5946,
+      "step": 78420
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 5.9905824000000004e-05,
+      "loss": 0.6095,
+      "step": 78480
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 5.9905752e-05,
+      "loss": 0.6016,
+      "step": 78540
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 5.9905680000000003e-05,
+      "loss": 0.6,
+      "step": 78600
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 5.9905608e-05,
+      "loss": 0.6051,
+      "step": 78660
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 5.9905536e-05,
+      "loss": 0.5969,
+      "step": 78720
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 5.9905464e-05,
+      "loss": 0.598,
+      "step": 78780
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 5.9905391999999995e-05,
+      "loss": 0.5902,
+      "step": 78840
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 5.9905320000000005e-05,
+      "loss": 0.5953,
+      "step": 78900
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 5.9905248e-05,
+      "loss": 0.594,
+      "step": 78960
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 5.9905176e-05,
+      "loss": 0.5871,
+      "step": 79020
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 5.9905104e-05,
+      "loss": 0.6082,
+      "step": 79080
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 5.9905032000000004e-05,
+      "loss": 0.5992,
+      "step": 79140
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 5.990496e-05,
+      "loss": 0.595,
+      "step": 79200
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 5.9904888000000004e-05,
+      "loss": 0.5921,
+      "step": 79260
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 5.9904816e-05,
+      "loss": 0.5906,
+      "step": 79320
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 5.9904744e-05,
+      "loss": 0.5945,
+      "step": 79380
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 5.9904672000000006e-05,
+      "loss": 0.6001,
+      "step": 79440
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 0.5644345283508301,
+      "eval_runtime": 338.3121,
+      "eval_samples_per_second": 295.585,
+      "eval_steps_per_second": 2.311,
+      "step": 79470
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 5.99046e-05,
+      "loss": 0.5921,
+      "step": 79500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 5.9904528e-05,
+      "loss": 0.59,
+      "step": 79560
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 5.9904456e-05,
+      "loss": 0.5994,
+      "step": 79620
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.9904384000000005e-05,
+      "loss": 0.5955,
+      "step": 79680
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.9904312e-05,
+      "loss": 0.5916,
+      "step": 79740
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 5.990424e-05,
+      "loss": 0.5788,
+      "step": 79800
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.9904168e-05,
+      "loss": 0.5976,
+      "step": 79860
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 5.9904096000000004e-05,
+      "loss": 0.5906,
+      "step": 79920
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.9904024e-05,
+      "loss": 0.5782,
+      "step": 79980
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 5.9903952e-05,
+      "loss": 0.6055,
+      "step": 80040
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.990388e-05,
+      "loss": 0.5741,
+      "step": 80100
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 5.9903808e-05,
+      "loss": 0.5896,
+      "step": 80160
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.9903736000000006e-05,
+      "loss": 0.575,
+      "step": 80220
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 5.9903664e-05,
+      "loss": 0.579,
+      "step": 80280
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.9903592e-05,
+      "loss": 0.5835,
+      "step": 80340
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.990352e-05,
+      "loss": 0.5945,
+      "step": 80400
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 5.9903448000000004e-05,
+      "loss": 0.5817,
+      "step": 80460
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.9903376e-05,
+      "loss": 0.5845,
+      "step": 80520
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 5.9903304e-05,
+      "loss": 0.5876,
+      "step": 80580
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.9903232e-05,
+      "loss": 0.5875,
+      "step": 80640
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 5.990316e-05,
+      "loss": 0.5869,
+      "step": 80700
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.9903088e-05,
+      "loss": 0.5873,
+      "step": 80760
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 5.9903016e-05,
+      "loss": 0.5796,
+      "step": 80820
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.9902944e-05,
+      "loss": 0.6041,
+      "step": 80880
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 5.9902872e-05,
+      "loss": 0.587,
+      "step": 80940
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.9902800000000005e-05,
+      "loss": 0.5826,
+      "step": 81000
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.9902728e-05,
+      "loss": 0.5953,
+      "step": 81060
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 5.9902656e-05,
+      "loss": 0.5851,
+      "step": 81120
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.990258400000001e-05,
+      "loss": 0.5919,
+      "step": 81180
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 5.9902512000000004e-05,
+      "loss": 0.5829,
+      "step": 81240
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.990244e-05,
+      "loss": 0.5911,
+      "step": 81300
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 5.9902368e-05,
+      "loss": 0.5872,
+      "step": 81360
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.9902296e-05,
+      "loss": 0.5879,
+      "step": 81420
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 5.9902224e-05,
+      "loss": 0.5873,
+      "step": 81480
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.9902152e-05,
+      "loss": 0.5865,
+      "step": 81540
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 5.990208e-05,
+      "loss": 0.5723,
+      "step": 81600
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.9902008e-05,
+      "loss": 0.5865,
+      "step": 81660
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.9901936e-05,
+      "loss": 0.5861,
+      "step": 81720
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 5.9901864000000005e-05,
+      "loss": 0.5925,
+      "step": 81780
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.9901792e-05,
+      "loss": 0.5928,
+      "step": 81840
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 5.990172e-05,
+      "loss": 0.5826,
+      "step": 81900
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.990164800000001e-05,
+      "loss": 0.5909,
+      "step": 81960
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 5.9901576000000004e-05,
+      "loss": 0.5895,
+      "step": 82020
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.9901504e-05,
+      "loss": 0.5725,
+      "step": 82080
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 5.9901431999999996e-05,
+      "loss": 0.5894,
+      "step": 82140
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.9901360000000006e-05,
+      "loss": 0.5825,
+      "step": 82200
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 5.9901288e-05,
+      "loss": 0.5783,
+      "step": 82260
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.9901216e-05,
+      "loss": 0.5806,
+      "step": 82320
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.9901144e-05,
+      "loss": 0.5812,
+      "step": 82380
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 5.9901072e-05,
+      "loss": 0.593,
+      "step": 82440
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.9901e-05,
+      "loss": 0.5773,
+      "step": 82500
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 5.9900928000000004e-05,
+      "loss": 0.581,
+      "step": 82560
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.9900856e-05,
+      "loss": 0.5819,
+      "step": 82620
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.9900784e-05,
+      "loss": 0.5973,
+      "step": 82680
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.990071200000001e-05,
+      "loss": 0.5941,
+      "step": 82740
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.990064e-05,
+      "loss": 0.5874,
+      "step": 82800
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.9900568e-05,
+      "loss": 0.5889,
+      "step": 82860
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.9900496e-05,
+      "loss": 0.5878,
+      "step": 82920
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.9900424000000006e-05,
+      "loss": 0.5899,
+      "step": 82980
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.9900352e-05,
+      "loss": 0.5884,
+      "step": 83040
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.990028e-05,
+      "loss": 0.5962,
+      "step": 83100
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.9900208e-05,
+      "loss": 0.5812,
+      "step": 83160
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.9900136e-05,
+      "loss": 0.5924,
+      "step": 83220
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.9900064e-05,
+      "loss": 0.5824,
+      "step": 83280
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.9899992000000004e-05,
+      "loss": 0.5817,
+      "step": 83340
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.989992e-05,
+      "loss": 0.5796,
+      "step": 83400
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.9899847999999997e-05,
+      "loss": 0.6034,
+      "step": 83460
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.9899776000000006e-05,
+      "loss": 0.5809,
+      "step": 83520
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.9899704e-05,
+      "loss": 0.5832,
+      "step": 83580
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.9899632e-05,
+      "loss": 0.582,
+      "step": 83640
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.989956e-05,
+      "loss": 0.57,
+      "step": 83700
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.9899488000000005e-05,
+      "loss": 0.5765,
+      "step": 83760
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.9899416e-05,
+      "loss": 0.5828,
+      "step": 83820
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.9899344000000005e-05,
+      "loss": 0.5766,
+      "step": 83880
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.9899272e-05,
+      "loss": 0.5856,
+      "step": 83940
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.98992e-05,
+      "loss": 0.592,
+      "step": 84000
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.9899128e-05,
+      "loss": 0.5809,
+      "step": 84060
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.9899056000000004e-05,
+      "loss": 0.5756,
+      "step": 84120
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.9898984e-05,
+      "loss": 0.5894,
+      "step": 84180
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.9898911999999996e-05,
+      "loss": 0.584,
+      "step": 84240
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.9898840000000006e-05,
+      "loss": 0.5866,
+      "step": 84300
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.9898768e-05,
+      "loss": 0.5815,
+      "step": 84360
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 5.9898696e-05,
+      "loss": 0.5796,
+      "step": 84420
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.9898624e-05,
+      "loss": 0.5779,
+      "step": 84480
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 5.9898552000000005e-05,
+      "loss": 0.5908,
+      "step": 84540
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.989848e-05,
+      "loss": 0.5897,
+      "step": 84600
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 5.9898408000000004e-05,
+      "loss": 0.5884,
+      "step": 84660
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.9898336e-05,
+      "loss": 0.5725,
+      "step": 84720
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 5.9898264000000004e-05,
+      "loss": 0.5918,
+      "step": 84780
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.9898192e-05,
+      "loss": 0.585,
+      "step": 84840
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.989812e-05,
+      "loss": 0.5804,
+      "step": 84900
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 5.9898048e-05,
+      "loss": 0.5777,
+      "step": 84960
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.9897975999999996e-05,
+      "loss": 0.5838,
+      "step": 85020
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.9897904000000006e-05,
+      "loss": 0.5902,
+      "step": 85080
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.9897832e-05,
+      "loss": 0.5763,
+      "step": 85140
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.989776e-05,
+      "loss": 0.5702,
+      "step": 85200
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.9897688e-05,
+      "loss": 0.5807,
+      "step": 85260
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.9897616000000004e-05,
+      "loss": 0.5957,
+      "step": 85320
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.9897544e-05,
+      "loss": 0.5953,
+      "step": 85380
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.9897472000000004e-05,
+      "loss": 0.5845,
+      "step": 85440
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.98974e-05,
+      "loss": 0.6025,
+      "step": 85500
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.9897328e-05,
+      "loss": 0.5909,
+      "step": 85560
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.9897256000000006e-05,
+      "loss": 0.6031,
+      "step": 85620
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.9897184e-05,
+      "loss": 0.5765,
+      "step": 85680
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.9897112e-05,
+      "loss": 0.5887,
+      "step": 85740
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.9897039999999995e-05,
+      "loss": 0.5918,
+      "step": 85800
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.9896968000000005e-05,
+      "loss": 0.5826,
+      "step": 85860
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.9896896e-05,
+      "loss": 0.5864,
+      "step": 85920
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.9896824e-05,
+      "loss": 0.5921,
+      "step": 85980
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.9896752e-05,
+      "loss": 0.5878,
+      "step": 86040
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.9896680000000004e-05,
+      "loss": 0.5893,
+      "step": 86100
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.9896608e-05,
+      "loss": 0.5741,
+      "step": 86160
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.9896536000000003e-05,
+      "loss": 0.5796,
+      "step": 86220
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.9896464e-05,
+      "loss": 0.5714,
+      "step": 86280
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.9896392e-05,
+      "loss": 0.5863,
+      "step": 86340
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.9896320000000006e-05,
+      "loss": 0.5824,
+      "step": 86400
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.9896248e-05,
+      "loss": 0.5816,
+      "step": 86460
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.9896176e-05,
+      "loss": 0.5969,
+      "step": 86520
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.9896103999999995e-05,
+      "loss": 0.5828,
+      "step": 86580
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.9896032000000005e-05,
+      "loss": 0.5777,
+      "step": 86640
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.989596e-05,
+      "loss": 0.5836,
+      "step": 86700
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.9895888e-05,
+      "loss": 0.5998,
+      "step": 86760
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.9895816e-05,
+      "loss": 0.5851,
+      "step": 86820
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.9895744000000004e-05,
+      "loss": 0.5884,
+      "step": 86880
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.9895672e-05,
+      "loss": 0.5761,
+      "step": 86940
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.98956e-05,
+      "loss": 0.5775,
+      "step": 87000
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.9895528e-05,
+      "loss": 0.5872,
+      "step": 87060
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.9895456e-05,
+      "loss": 0.5801,
+      "step": 87120
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.9895384000000006e-05,
+      "loss": 0.5934,
+      "step": 87180
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.9895312e-05,
+      "loss": 0.584,
+      "step": 87240
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.989524e-05,
+      "loss": 0.5738,
+      "step": 87300
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.9895168e-05,
+      "loss": 0.5793,
+      "step": 87360
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.9895096000000004e-05,
+      "loss": 0.5901,
+      "step": 87420
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.9895024e-05,
+      "loss": 0.5848,
+      "step": 87480
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.9894952e-05,
+      "loss": 0.5762,
+      "step": 87540
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.989488e-05,
+      "loss": 0.5841,
+      "step": 87600
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.9894808e-05,
+      "loss": 0.581,
+      "step": 87660
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.9894736e-05,
+      "loss": 0.5923,
+      "step": 87720
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.9894664e-05,
+      "loss": 0.5986,
+      "step": 87780
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.9894592e-05,
+      "loss": 0.5811,
+      "step": 87840
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.989452e-05,
+      "loss": 0.5936,
+      "step": 87900
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.9894448000000005e-05,
+      "loss": 0.5842,
+      "step": 87960
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.9894376e-05,
+      "loss": 0.586,
+      "step": 88020
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.9894304e-05,
+      "loss": 0.5939,
+      "step": 88080
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.989423200000001e-05,
+      "loss": 0.5847,
+      "step": 88140
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.9894160000000004e-05,
+      "loss": 0.5964,
+      "step": 88200
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.9894088e-05,
+      "loss": 0.5861,
+      "step": 88260
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.9894016e-05,
+      "loss": 0.5808,
+      "step": 88320
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.9893944e-05,
+      "loss": 0.5997,
+      "step": 88380
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.9893872e-05,
+      "loss": 0.5815,
+      "step": 88440
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.98938e-05,
+      "loss": 0.5903,
+      "step": 88500
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.9893728e-05,
+      "loss": 0.59,
+      "step": 88560
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.9893656e-05,
+      "loss": 0.5801,
+      "step": 88620
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.9893584e-05,
+      "loss": 0.5801,
+      "step": 88680
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.9893512000000005e-05,
+      "loss": 0.5811,
+      "step": 88740
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.989344e-05,
+      "loss": 0.591,
+      "step": 88800
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.9893368e-05,
+      "loss": 0.5894,
+      "step": 88860
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.989329600000001e-05,
+      "loss": 0.5805,
+      "step": 88920
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.9893224000000003e-05,
+      "loss": 0.5849,
+      "step": 88980
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.9893152e-05,
+      "loss": 0.5835,
+      "step": 89040
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.989308e-05,
+      "loss": 0.5938,
+      "step": 89100
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.9893008e-05,
+      "loss": 0.5937,
+      "step": 89160
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.9892936e-05,
+      "loss": 0.5864,
+      "step": 89220
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.9892864e-05,
+      "loss": 0.5814,
+      "step": 89280
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.9892792e-05,
+      "loss": 0.5774,
+      "step": 89340
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.989272e-05,
+      "loss": 0.5917,
+      "step": 89400
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.9892648e-05,
+      "loss": 0.5861,
+      "step": 89460
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.9892576000000004e-05,
+      "loss": 0.5781,
+      "step": 89520
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.9892504e-05,
+      "loss": 0.5817,
+      "step": 89580
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.9892432e-05,
+      "loss": 0.5847,
+      "step": 89640
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.989236000000001e-05,
+      "loss": 0.5865,
+      "step": 89700
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.9892288e-05,
+      "loss": 0.5813,
+      "step": 89760
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 5.9892216e-05,
+      "loss": 0.5787,
+      "step": 89820
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.9892144e-05,
+      "loss": 0.5754,
+      "step": 89880
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 5.9892072e-05,
+      "loss": 0.5839,
+      "step": 89940
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.9892e-05,
+      "loss": 0.5826,
+      "step": 90000
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.9891928000000005e-05,
+      "loss": 0.5811,
+      "step": 90060
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 5.9891856e-05,
+      "loss": 0.5754,
+      "step": 90120
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.9891784e-05,
+      "loss": 0.5925,
+      "step": 90180
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 5.9891712e-05,
+      "loss": 0.5707,
+      "step": 90240
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.9891640000000004e-05,
+      "loss": 0.577,
+      "step": 90300
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 5.9891568e-05,
+      "loss": 0.5724,
+      "step": 90360
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.9891495999999996e-05,
+      "loss": 0.5719,
+      "step": 90420
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 5.9891424000000006e-05,
+      "loss": 0.574,
+      "step": 90480
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.9891352e-05,
+      "loss": 0.5835,
+      "step": 90540
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.989128e-05,
+      "loss": 0.5839,
+      "step": 90600
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 5.9891208e-05,
+      "loss": 0.5915,
+      "step": 90660
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.9891136000000005e-05,
+      "loss": 0.5833,
+      "step": 90720
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 5.9891064e-05,
+      "loss": 0.5849,
+      "step": 90780
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.9890992000000005e-05,
+      "loss": 0.5737,
+      "step": 90840
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 5.989092e-05,
+      "loss": 0.5922,
+      "step": 90900
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.9890848e-05,
+      "loss": 0.583,
+      "step": 90960
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 5.9890776e-05,
+      "loss": 0.581,
+      "step": 91020
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.9890704e-05,
+      "loss": 0.5743,
+      "step": 91080
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 5.9890632e-05,
+      "loss": 0.5791,
+      "step": 91140
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.9890559999999996e-05,
+      "loss": 0.5735,
+      "step": 91200
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.9890488000000006e-05,
+      "loss": 0.5844,
+      "step": 91260
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 5.9890416e-05,
+      "loss": 0.5857,
+      "step": 91320
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.9890344e-05,
+      "loss": 0.586,
+      "step": 91380
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 5.9890272e-05,
+      "loss": 0.586,
+      "step": 91440
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.9890200000000005e-05,
+      "loss": 0.5881,
+      "step": 91500
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 5.9890128e-05,
+      "loss": 0.5847,
+      "step": 91560
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.9890056000000004e-05,
+      "loss": 0.5821,
+      "step": 91620
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 5.9889984e-05,
+      "loss": 0.5704,
+      "step": 91680
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.9889912e-05,
+      "loss": 0.5919,
+      "step": 91740
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.988984000000001e-05,
+      "loss": 0.5823,
+      "step": 91800
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.9889768e-05,
+      "loss": 0.5871,
+      "step": 91860
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.9889696e-05,
+      "loss": 0.5815,
+      "step": 91920
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 5.9889623999999996e-05,
+      "loss": 0.5844,
+      "step": 91980
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.9889552000000005e-05,
+      "loss": 0.5819,
+      "step": 92040
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 5.988948e-05,
+      "loss": 0.5846,
+      "step": 92100
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.9889408e-05,
+      "loss": 0.5786,
+      "step": 92160
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 5.9889336e-05,
+      "loss": 0.589,
+      "step": 92220
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.9889264000000004e-05,
+      "loss": 0.5776,
+      "step": 92280
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 5.9889192e-05,
+      "loss": 0.5901,
+      "step": 92340
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.9889120000000004e-05,
+      "loss": 0.5847,
+      "step": 92400
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 5.9889048e-05,
+      "loss": 0.597,
+      "step": 92460
+    }
+  ],
+  "logging_steps": 60,
+  "max_steps": 50000000,
+  "num_train_epochs": 3776,
+  "save_steps": 500,
+  "total_flos": 2.045499936537987e+18,
+  "trial_name": null,
+  "trial_params": null
+}