diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6030 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.3535169785169785,
+  "eval_steps": 500,
+  "global_step": 500000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.2499999999999995e-06,
+      "loss": 7.4049,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2499999999999999e-05,
+      "loss": 4.7791,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.875e-05,
+      "loss": 3.8115,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4999999999999998e-05,
+      "loss": 3.3298,
+      "step": 2000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.125e-05,
+      "loss": 3.0365,
+      "step": 2500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.75e-05,
+      "loss": 2.8445,
+      "step": 3000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.375e-05,
+      "loss": 2.7101,
+      "step": 3500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999999999999996e-05,
+      "loss": 2.6043,
+      "step": 4000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.625e-05,
+      "loss": 2.5286,
+      "step": 4500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.25e-05,
+      "loss": 2.467,
+      "step": 5000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.874999999999999e-05,
+      "loss": 2.4192,
+      "step": 5500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.5e-05,
+      "loss": 2.3785,
+      "step": 6000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.124999999999998e-05,
+      "loss": 2.3456,
+      "step": 6500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.75e-05,
+      "loss": 2.3183,
+      "step": 7000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.374999999999999e-05,
+      "loss": 2.3005,
+      "step": 7500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999999999999999e-05,
+      "loss": 2.2808,
+      "step": 8000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010625,
+      "loss": 2.2722,
+      "step": 8500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001125,
+      "loss": 2.2577,
+      "step": 9000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011874999999999999,
+      "loss": 2.2479,
+      "step": 9500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000125,
+      "loss": 2.2452,
+      "step": 10000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00013125,
+      "loss": 2.2419,
+      "step": 10500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00013749999999999998,
+      "loss": 2.2345,
+      "step": 11000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014375,
+      "loss": 2.2276,
+      "step": 11500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015,
+      "loss": 2.2285,
+      "step": 12000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015625,
+      "loss": 2.2178,
+      "step": 12500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00016249999999999997,
+      "loss": 2.2214,
+      "step": 13000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00016874999999999998,
+      "loss": 2.2269,
+      "step": 13500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000175,
+      "loss": 2.2322,
+      "step": 14000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018124999999999996,
+      "loss": 2.2292,
+      "step": 14500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018749999999999998,
+      "loss": 2.2274,
+      "step": 15000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019375,
+      "loss": 2.232,
+      "step": 15500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019999999999999998,
+      "loss": 2.2316,
+      "step": 16000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00020624999999999997,
+      "loss": 2.238,
+      "step": 16500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002125,
+      "loss": 2.2426,
+      "step": 17000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00021874999999999998,
+      "loss": 2.2408,
+      "step": 17500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000225,
+      "loss": 2.2456,
+      "step": 18000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00023124999999999998,
+      "loss": 2.2473,
+      "step": 18500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00023749999999999997,
+      "loss": 2.2549,
+      "step": 19000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00024375,
+      "loss": 2.2542,
+      "step": 19500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00025,
+      "loss": 2.2604,
+      "step": 20000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00025624999999999997,
+      "loss": 2.2626,
+      "step": 20500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002625,
+      "loss": 2.2694,
+      "step": 21000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00026875,
+      "loss": 2.2724,
+      "step": 21500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00027499999999999996,
+      "loss": 2.2722,
+      "step": 22000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00028125,
+      "loss": 2.2807,
+      "step": 22500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002875,
+      "loss": 2.2846,
+      "step": 23000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029374999999999996,
+      "loss": 2.292,
+      "step": 23500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003,
+      "loss": 2.297,
+      "step": 24000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002996848739495798,
+      "loss": 2.3038,
+      "step": 24500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029936974789915966,
+      "loss": 2.2998,
+      "step": 25000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002990546218487395,
+      "loss": 2.2977,
+      "step": 25500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002987394957983193,
+      "loss": 2.2925,
+      "step": 26000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002984243697478991,
+      "loss": 2.2943,
+      "step": 26500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000298109243697479,
+      "loss": 2.2899,
+      "step": 27000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002977941176470588,
+      "loss": 2.2899,
+      "step": 27500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002974789915966386,
+      "loss": 2.2806,
+      "step": 28000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002971638655462185,
+      "loss": 2.2823,
+      "step": 28500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002968487394957983,
+      "loss": 2.2795,
+      "step": 29000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029653361344537813,
+      "loss": 2.2736,
+      "step": 29500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029621848739495795,
+      "loss": 2.2704,
+      "step": 30000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002959033613445378,
+      "loss": 2.269,
+      "step": 30500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029558823529411763,
+      "loss": 2.2601,
+      "step": 31000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029527310924369745,
+      "loss": 2.2674,
+      "step": 31500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029495798319327727,
+      "loss": 2.2575,
+      "step": 32000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002946428571428571,
+      "loss": 2.2583,
+      "step": 32500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029432773109243696,
+      "loss": 2.2494,
+      "step": 33000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002940126050420168,
+      "loss": 2.2412,
+      "step": 33500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002936974789915966,
+      "loss": 2.2453,
+      "step": 34000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002933823529411764,
+      "loss": 2.2401,
+      "step": 34500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002930672268907563,
+      "loss": 2.2375,
+      "step": 35000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002927521008403361,
+      "loss": 2.2302,
+      "step": 35500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002924369747899159,
+      "loss": 2.2275,
+      "step": 36000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002921218487394958,
+      "loss": 2.223,
+      "step": 36500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002918067226890756,
+      "loss": 2.2252,
+      "step": 37000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002914915966386554,
+      "loss": 2.2238,
+      "step": 37500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029117647058823524,
+      "loss": 2.2206,
+      "step": 38000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002908613445378151,
+      "loss": 2.2162,
+      "step": 38500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029054621848739493,
+      "loss": 2.2141,
+      "step": 39000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029023109243697475,
+      "loss": 2.2215,
+      "step": 39500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002899159663865546,
+      "loss": 2.2107,
+      "step": 40000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028960084033613444,
+      "loss": 2.2125,
+      "step": 40500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028928571428571425,
+      "loss": 2.2086,
+      "step": 41000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028897058823529407,
+      "loss": 2.2097,
+      "step": 41500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028865546218487394,
+      "loss": 2.1996,
+      "step": 42000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028834033613445376,
+      "loss": 2.1972,
+      "step": 42500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002880252100840336,
+      "loss": 2.1906,
+      "step": 43000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028771008403361345,
+      "loss": 2.2015,
+      "step": 43500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028739495798319327,
+      "loss": 2.1927,
+      "step": 44000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002870798319327731,
+      "loss": 2.1889,
+      "step": 44500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028676470588235296,
+      "loss": 2.1904,
+      "step": 45000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002864495798319328,
+      "loss": 2.1809,
+      "step": 45500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002861344537815126,
+      "loss": 2.1777,
+      "step": 46000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002858193277310924,
+      "loss": 2.1772,
+      "step": 46500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002855042016806722,
+      "loss": 2.1714,
+      "step": 47000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028518907563025204,
+      "loss": 2.171,
+      "step": 47500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002848739495798319,
+      "loss": 2.1814,
+      "step": 48000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028455882352941173,
+      "loss": 2.1713,
+      "step": 48500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028424369747899155,
+      "loss": 2.168,
+      "step": 49000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028392857142857137,
+      "loss": 2.1635,
+      "step": 49500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028361344537815124,
+      "loss": 2.1689,
+      "step": 50000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028329831932773106,
+      "loss": 2.1614,
+      "step": 50500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002829831932773109,
+      "loss": 2.1594,
+      "step": 51000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028266806722689075,
+      "loss": 2.1585,
+      "step": 51500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028235294117647056,
+      "loss": 2.1639,
+      "step": 52000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002820378151260504,
+      "loss": 2.1574,
+      "step": 52500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028172268907563025,
+      "loss": 2.1512,
+      "step": 53000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028140756302521007,
+      "loss": 2.1571,
+      "step": 53500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002810924369747899,
+      "loss": 2.1513,
+      "step": 54000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002807773109243697,
+      "loss": 2.1503,
+      "step": 54500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002804621848739496,
+      "loss": 2.1462,
+      "step": 55000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002801470588235294,
+      "loss": 2.1449,
+      "step": 55500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002798319327731092,
+      "loss": 2.1383,
+      "step": 56000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002795168067226891,
+      "loss": 2.1431,
+      "step": 56500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002792016806722689,
+      "loss": 2.1349,
+      "step": 57000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002788865546218487,
+      "loss": 2.1423,
+      "step": 57500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00027857142857142854,
+      "loss": 2.1375,
+      "step": 58000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002782563025210084,
+      "loss": 2.1261,
+      "step": 58500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002779411764705882,
+      "loss": 2.1277,
+      "step": 59000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00027762605042016804,
+      "loss": 2.1204,
+      "step": 59500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002773109243697479,
+      "loss": 2.1277,
+      "step": 60000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00027699579831932773,
+      "loss": 2.1246,
+      "step": 60500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027668067226890755,
+      "loss": 2.1207,
+      "step": 61000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027636554621848737,
+      "loss": 2.1161,
+      "step": 61500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002760504201680672,
+      "loss": 2.1198,
+      "step": 62000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000275735294117647,
+      "loss": 2.1122,
+      "step": 62500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00027542016806722687,
+      "loss": 2.1113,
+      "step": 63000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002751050420168067,
+      "loss": 2.1164,
+      "step": 63500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002747899159663865,
+      "loss": 2.1096,
+      "step": 64000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002744747899159664,
+      "loss": 2.106,
+      "step": 64500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002741596638655462,
+      "loss": 2.1015,
+      "step": 65000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000273844537815126,
+      "loss": 2.1041,
+      "step": 65500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027352941176470583,
+      "loss": 2.0991,
+      "step": 66000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002732142857142857,
+      "loss": 2.0956,
+      "step": 66500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002728991596638655,
+      "loss": 2.0945,
+      "step": 67000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00027258403361344534,
+      "loss": 2.0958,
+      "step": 67500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002722689075630252,
+      "loss": 2.1115,
+      "step": 68000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000271953781512605,
+      "loss": 2.1205,
+      "step": 68500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027163865546218484,
+      "loss": 2.1098,
+      "step": 69000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027132352941176466,
+      "loss": 2.1046,
+      "step": 69500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027100840336134453,
+      "loss": 2.1013,
+      "step": 70000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027069327731092435,
+      "loss": 2.0989,
+      "step": 70500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027037815126050417,
+      "loss": 2.0927,
+      "step": 71000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00027006302521008404,
+      "loss": 2.0827,
+      "step": 71500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00026974789915966386,
+      "loss": 2.0851,
+      "step": 72000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002694327731092437,
+      "loss": 2.0841,
+      "step": 72500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002691176470588235,
+      "loss": 2.0816,
+      "step": 73000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00026880252100840336,
+      "loss": 2.0761,
+      "step": 73500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002684873949579832,
+      "loss": 2.0745,
+      "step": 74000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000268172268907563,
+      "loss": 2.0762,
+      "step": 74500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00026785714285714287,
+      "loss": 2.0704,
+      "step": 75000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002675420168067227,
+      "loss": 2.0737,
+      "step": 75500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002672268907563025,
+      "loss": 2.0687,
+      "step": 76000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002669117647058823,
+      "loss": 2.0743,
+      "step": 76500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026659663865546214,
+      "loss": 2.0758,
+      "step": 77000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026628151260504196,
+      "loss": 2.0677,
+      "step": 77500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026596638655462183,
+      "loss": 2.0643,
+      "step": 78000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026565126050420165,
+      "loss": 2.0588,
+      "step": 78500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00026533613445378146,
+      "loss": 2.0676,
+      "step": 79000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026502100840336134,
+      "loss": 2.065,
+      "step": 79500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026470588235294115,
+      "loss": 2.064,
+      "step": 80000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026439075630252097,
+      "loss": 2.0579,
+      "step": 80500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002640756302521008,
+      "loss": 2.0614,
+      "step": 81000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026376050420168066,
+      "loss": 2.0664,
+      "step": 81500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002634453781512605,
+      "loss": 2.0648,
+      "step": 82000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002631302521008403,
+      "loss": 2.0564,
+      "step": 82500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00026281512605042017,
+      "loss": 2.0569,
+      "step": 83000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002625,
+      "loss": 2.0517,
+      "step": 83500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002621848739495798,
+      "loss": 2.0453,
+      "step": 84000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002618697478991596,
+      "loss": 2.0447,
+      "step": 84500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002615546218487395,
+      "loss": 2.0431,
+      "step": 85000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002612394957983193,
+      "loss": 2.0414,
+      "step": 85500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002609243697478991,
+      "loss": 2.0381,
+      "step": 86000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000260609243697479,
+      "loss": 2.0392,
+      "step": 86500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002602941176470588,
+      "loss": 2.0317,
+      "step": 87000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025997899159663863,
+      "loss": 2.0338,
+      "step": 87500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025966386554621845,
+      "loss": 2.0348,
+      "step": 88000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002593487394957983,
+      "loss": 2.0349,
+      "step": 88500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025903361344537814,
+      "loss": 2.0295,
+      "step": 89000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025871848739495796,
+      "loss": 2.0253,
+      "step": 89500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025840336134453783,
+      "loss": 2.0272,
+      "step": 90000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00025808823529411764,
+      "loss": 2.0273,
+      "step": 90500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00025777310924369746,
+      "loss": 2.0295,
+      "step": 91000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002574579831932773,
+      "loss": 2.0284,
+      "step": 91500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002571428571428571,
+      "loss": 2.0335,
+      "step": 92000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002568277310924369,
+      "loss": 2.032,
+      "step": 92500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002565126050420168,
+      "loss": 2.0256,
+      "step": 93000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002561974789915966,
+      "loss": 2.0278,
+      "step": 93500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002558823529411764,
+      "loss": 2.0226,
+      "step": 94000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002555672268907563,
+      "loss": 2.0186,
+      "step": 94500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002552521008403361,
+      "loss": 2.0111,
+      "step": 95000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025493697478991593,
+      "loss": 2.0139,
+      "step": 95500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025462184873949575,
+      "loss": 2.0092,
+      "step": 96000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002543067226890756,
+      "loss": 2.0139,
+      "step": 96500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025399159663865543,
+      "loss": 2.0017,
+      "step": 97000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00025367647058823525,
+      "loss": 2.0061,
+      "step": 97500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002533613445378151,
+      "loss": 2.0049,
+      "step": 98000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025304621848739494,
+      "loss": 2.0054,
+      "step": 98500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025273109243697476,
+      "loss": 2.0045,
+      "step": 99000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002524159663865546,
+      "loss": 2.0043,
+      "step": 99500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025210084033613445,
+      "loss": 2.0007,
+      "step": 100000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025178571428571426,
+      "loss": 2.0047,
+      "step": 100500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002514705882352941,
+      "loss": 2.0061,
+      "step": 101000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00025115546218487395,
+      "loss": 2.0006,
+      "step": 101500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00025084033613445377,
+      "loss": 1.9995,
+      "step": 102000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002505252100840336,
+      "loss": 1.9997,
+      "step": 102500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00025021008403361346,
+      "loss": 1.9942,
+      "step": 103000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002498949579831933,
+      "loss": 1.9951,
+      "step": 103500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002495798319327731,
+      "loss": 1.9872,
+      "step": 104000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002492647058823529,
+      "loss": 1.9908,
+      "step": 104500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002489495798319328,
+      "loss": 1.9933,
+      "step": 105000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002486344537815126,
+      "loss": 1.9884,
+      "step": 105500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002483193277310924,
+      "loss": 1.9885,
+      "step": 106000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024800420168067224,
+      "loss": 1.9881,
+      "step": 106500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024768907563025205,
+      "loss": 1.9842,
+      "step": 107000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024737394957983187,
+      "loss": 1.9816,
+      "step": 107500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024705882352941174,
+      "loss": 1.9772,
+      "step": 108000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00024674369747899156,
+      "loss": 1.9766,
+      "step": 108500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002464285714285714,
+      "loss": 1.9795,
+      "step": 109000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024611344537815125,
+      "loss": 1.9754,
+      "step": 109500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024579831932773107,
+      "loss": 1.9751,
+      "step": 110000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002454831932773109,
+      "loss": 1.9757,
+      "step": 110500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00024516806722689076,
+      "loss": 1.9755,
+      "step": 111000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002448529411764706,
+      "loss": 1.9753,
+      "step": 111500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002445378151260504,
+      "loss": 1.9657,
+      "step": 112000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002442226890756302,
+      "loss": 1.9663,
+      "step": 112500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00024390756302521005,
+      "loss": 1.9682,
+      "step": 113000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002435924369747899,
+      "loss": 1.971,
+      "step": 113500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00024327731092436971,
+      "loss": 1.9687,
+      "step": 114000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00024296218487394956,
+      "loss": 1.9826,
+      "step": 114500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002426470588235294,
+      "loss": 1.9643,
+      "step": 115000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00024233193277310922,
+      "loss": 1.9708,
+      "step": 115500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00024201680672268907,
+      "loss": 1.9628,
+      "step": 116000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00024170168067226888,
+      "loss": 1.9633,
+      "step": 116500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00024138655462184873,
+      "loss": 1.9607,
+      "step": 117000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00024107142857142857,
+      "loss": 1.9574,
+      "step": 117500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002407563025210084,
+      "loss": 1.9611,
+      "step": 118000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00024044117647058823,
+      "loss": 1.9575,
+      "step": 118500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00024012605042016805,
+      "loss": 1.9538,
+      "step": 119000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002398109243697479,
+      "loss": 1.9573,
+      "step": 119500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00023949579831932771,
+      "loss": 1.9601,
+      "step": 120000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00023918067226890756,
+      "loss": 1.9539,
+      "step": 120500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00023886554621848735,
+      "loss": 1.9532,
+      "step": 121000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002385504201680672,
+      "loss": 1.951,
+      "step": 121500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000238235294117647,
+      "loss": 1.9554,
+      "step": 122000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00023792016806722686,
+      "loss": 1.9472,
+      "step": 122500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002376050420168067,
+      "loss": 1.9462,
+      "step": 123000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00023728991596638652,
+      "loss": 1.9524,
+      "step": 123500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00023697478991596636,
+      "loss": 1.9525,
+      "step": 124000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00023665966386554618,
+      "loss": 1.9444,
+      "step": 124500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00023634453781512602,
+      "loss": 1.9404,
+      "step": 125000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00023602941176470587,
+      "loss": 1.9431,
+      "step": 125500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00023571428571428569,
+      "loss": 1.943,
+      "step": 126000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00023539915966386553,
+      "loss": 1.9404,
+      "step": 126500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00023508403361344535,
+      "loss": 1.9447,
+      "step": 127000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002347689075630252,
+      "loss": 1.9425,
+      "step": 127500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000234453781512605,
+      "loss": 1.9376,
+      "step": 128000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00023413865546218485,
+      "loss": 1.9374,
+      "step": 128500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002338235294117647,
+      "loss": 1.9373,
+      "step": 129000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00023350840336134452,
+      "loss": 1.9364,
+      "step": 129500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00023319327731092436,
+      "loss": 1.9344,
+      "step": 130000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00023287815126050418,
+      "loss": 1.9299,
+      "step": 130500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00023256302521008402,
+      "loss": 1.9243,
+      "step": 131000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00023224789915966384,
+      "loss": 1.93,
+      "step": 131500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00023193277310924368,
+      "loss": 1.925,
+      "step": 132000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00023161764705882353,
+      "loss": 1.9256,
+      "step": 132500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00023130252100840335,
+      "loss": 1.9334,
+      "step": 133000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002309873949579832,
+      "loss": 1.9266,
+      "step": 133500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000230672268907563,
+      "loss": 1.9338,
+      "step": 134000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00023035714285714285,
+      "loss": 1.9258,
+      "step": 134500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00023004201680672267,
+      "loss": 1.9253,
+      "step": 135000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00022972689075630252,
+      "loss": 1.9259,
+      "step": 135500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002294117647058823,
+      "loss": 1.9509,
+      "step": 136000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00022909663865546215,
+      "loss": 1.9366,
+      "step": 136500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000228781512605042,
+      "loss": 1.9323,
+      "step": 137000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002284663865546218,
+      "loss": 1.9298,
+      "step": 137500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00022815126050420166,
+      "loss": 1.9611,
+      "step": 138000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00022783613445378147,
+      "loss": 1.9599,
+      "step": 138500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00022752100840336132,
+      "loss": 1.9537,
+      "step": 139000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00022720588235294114,
+      "loss": 1.9308,
+      "step": 139500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00022689075630252098,
+      "loss": 1.9208,
+      "step": 140000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00022657563025210083,
+      "loss": 1.9233,
+      "step": 140500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00022626050420168064,
+      "loss": 1.9233,
+      "step": 141000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002259453781512605,
+      "loss": 1.927,
+      "step": 141500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002256302521008403,
+      "loss": 1.9331,
+      "step": 142000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00022531512605042015,
+      "loss": 1.9358,
+      "step": 142500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000225,
+      "loss": 1.9415,
+      "step": 143000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002246848739495798,
+      "loss": 1.9427,
+      "step": 143500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00022436974789915966,
+      "loss": 1.9216,
+      "step": 144000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00022405462184873947,
+      "loss": 1.9196,
+      "step": 144500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00022373949579831932,
+      "loss": 1.9109,
+      "step": 145000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00022342436974789914,
+      "loss": 1.9166,
+      "step": 145500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00022310924369747898,
+      "loss": 1.9231,
+      "step": 146000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00022279411764705882,
+      "loss": 1.9091,
+      "step": 146500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00022247899159663864,
+      "loss": 1.9096,
+      "step": 147000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00022216386554621849,
+      "loss": 1.9107,
+      "step": 147500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002218487394957983,
+      "loss": 1.908,
+      "step": 148000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00022153361344537815,
+      "loss": 1.9052,
+      "step": 148500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00022121848739495797,
+      "loss": 1.9058,
+      "step": 149000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002209033613445378,
+      "loss": 1.9043,
+      "step": 149500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00022058823529411765,
+      "loss": 1.9049,
+      "step": 150000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00022027310924369745,
+      "loss": 1.9028,
+      "step": 150500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002199579831932773,
+      "loss": 1.9034,
+      "step": 151000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002196428571428571,
+      "loss": 1.8976,
+      "step": 151500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021932773109243695,
+      "loss": 1.8957,
+      "step": 152000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021901260504201677,
+      "loss": 1.895,
+      "step": 152500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00021869747899159661,
+      "loss": 1.8944,
+      "step": 153000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00021838235294117643,
+      "loss": 1.8904,
+      "step": 153500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00021806722689075628,
+      "loss": 1.8939,
+      "step": 154000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00021775210084033612,
+      "loss": 1.8921,
+      "step": 154500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00021743697478991594,
+      "loss": 1.8873,
+      "step": 155000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00021712184873949578,
+      "loss": 1.8907,
+      "step": 155500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002168067226890756,
+      "loss": 1.8898,
+      "step": 156000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00021649159663865544,
+      "loss": 1.8826,
+      "step": 156500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00021617647058823526,
+      "loss": 1.8896,
+      "step": 157000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002158613445378151,
+      "loss": 1.8836,
+      "step": 157500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00021554621848739495,
+      "loss": 1.9006,
+      "step": 158000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00021523109243697477,
+      "loss": 1.8949,
+      "step": 158500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002149159663865546,
+      "loss": 1.8931,
+      "step": 159000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00021460084033613443,
+      "loss": 1.8886,
+      "step": 159500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00021428571428571427,
+      "loss": 1.892,
+      "step": 160000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002139705882352941,
+      "loss": 1.8935,
+      "step": 160500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00021365546218487394,
+      "loss": 1.8977,
+      "step": 161000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00021334033613445378,
+      "loss": 1.8891,
+      "step": 161500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002130252100840336,
+      "loss": 1.8846,
+      "step": 162000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00021271008403361344,
+      "loss": 1.8816,
+      "step": 162500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00021239495798319326,
+      "loss": 1.8791,
+      "step": 163000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002120798319327731,
+      "loss": 1.9027,
+      "step": 163500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00021176470588235295,
+      "loss": 1.8873,
+      "step": 164000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00021144957983193277,
+      "loss": 1.8795,
+      "step": 164500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002111344537815126,
+      "loss": 1.8775,
+      "step": 165000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002108193277310924,
+      "loss": 1.8799,
+      "step": 165500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00021050420168067225,
+      "loss": 1.8771,
+      "step": 166000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00021018907563025206,
+      "loss": 1.8797,
+      "step": 166500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002098739495798319,
+      "loss": 1.8759,
+      "step": 167000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00020955882352941173,
+      "loss": 1.8742,
+      "step": 167500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00020924369747899157,
+      "loss": 1.8691,
+      "step": 168000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002089285714285714,
+      "loss": 1.8702,
+      "step": 168500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00020861344537815123,
+      "loss": 1.8689,
+      "step": 169000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00020829831932773108,
+      "loss": 1.8701,
+      "step": 169500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002079831932773109,
+      "loss": 1.8695,
+      "step": 170000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00020766806722689074,
+      "loss": 1.8706,
+      "step": 170500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00020735294117647056,
+      "loss": 1.8935,
+      "step": 171000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002070378151260504,
+      "loss": 1.9075,
+      "step": 171500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00020672268907563025,
+      "loss": 1.8929,
+      "step": 172000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00020640756302521006,
+      "loss": 1.8795,
+      "step": 172500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002060924369747899,
+      "loss": 1.8845,
+      "step": 173000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00020577731092436973,
+      "loss": 1.8762,
+      "step": 173500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00020546218487394957,
+      "loss": 1.8745,
+      "step": 174000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002051470588235294,
+      "loss": 1.8751,
+      "step": 174500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00020483193277310923,
+      "loss": 1.8746,
+      "step": 175000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00020451680672268908,
+      "loss": 1.8614,
+      "step": 175500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002042016806722689,
+      "loss": 1.8634,
+      "step": 176000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00020388655462184874,
+      "loss": 1.8711,
+      "step": 176500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00020357142857142856,
+      "loss": 1.8648,
+      "step": 177000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002032563025210084,
+      "loss": 1.8615,
+      "step": 177500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00020294117647058822,
+      "loss": 1.8661,
+      "step": 178000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00020262605042016806,
+      "loss": 1.8616,
+      "step": 178500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002023109243697479,
+      "loss": 1.8601,
+      "step": 179000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020199579831932772,
+      "loss": 1.865,
+      "step": 179500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020168067226890757,
+      "loss": 1.8664,
+      "step": 180000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020136554621848736,
+      "loss": 1.8562,
+      "step": 180500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002010504201680672,
+      "loss": 1.8589,
+      "step": 181000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020073529411764702,
+      "loss": 1.8583,
+      "step": 181500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020042016806722687,
+      "loss": 1.8623,
+      "step": 182000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00020010504201680668,
+      "loss": 1.8561,
+      "step": 182500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019978991596638653,
+      "loss": 1.8537,
+      "step": 183000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019947478991596637,
+      "loss": 1.8622,
+      "step": 183500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001991596638655462,
+      "loss": 1.8568,
+      "step": 184000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019884453781512603,
+      "loss": 1.8543,
+      "step": 184500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019852941176470585,
+      "loss": 1.8528,
+      "step": 185000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001982142857142857,
+      "loss": 1.8509,
+      "step": 185500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001978991596638655,
+      "loss": 1.8489,
+      "step": 186000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019758403361344536,
+      "loss": 1.8483,
+      "step": 186500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001972689075630252,
+      "loss": 1.8487,
+      "step": 187000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019695378151260502,
+      "loss": 1.8416,
+      "step": 187500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019663865546218486,
+      "loss": 1.8497,
+      "step": 188000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019632352941176468,
+      "loss": 1.849,
+      "step": 188500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019600840336134453,
+      "loss": 1.8466,
+      "step": 189000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019569327731092434,
+      "loss": 1.8456,
+      "step": 189500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001953781512605042,
+      "loss": 1.8453,
+      "step": 190000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019506302521008403,
+      "loss": 1.8418,
+      "step": 190500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019474789915966385,
+      "loss": 1.8429,
+      "step": 191000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001944327731092437,
+      "loss": 1.8453,
+      "step": 191500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001941176470588235,
+      "loss": 1.8458,
+      "step": 192000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019380252100840336,
+      "loss": 1.842,
+      "step": 192500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001934873949579832,
+      "loss": 1.8381,
+      "step": 193000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019317226890756302,
+      "loss": 1.8374,
+      "step": 193500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019285714285714286,
+      "loss": 1.8359,
+      "step": 194000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019254201680672268,
+      "loss": 1.8336,
+      "step": 194500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019222689075630253,
+      "loss": 1.8365,
+      "step": 195000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019191176470588232,
+      "loss": 1.8391,
+      "step": 195500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019159663865546216,
+      "loss": 1.8431,
+      "step": 196000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019128151260504198,
+      "loss": 1.8361,
+      "step": 196500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019096638655462182,
+      "loss": 1.8583,
+      "step": 197000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019065126050420164,
+      "loss": 1.8656,
+      "step": 197500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019033613445378148,
+      "loss": 1.8508,
+      "step": 198000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019002100840336133,
+      "loss": 1.8533,
+      "step": 198500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018970588235294115,
+      "loss": 1.8416,
+      "step": 199000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000189390756302521,
+      "loss": 1.8503,
+      "step": 199500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001890756302521008,
+      "loss": 1.844,
+      "step": 200000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018876050420168065,
+      "loss": 1.8389,
+      "step": 200500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001884453781512605,
+      "loss": 1.836,
+      "step": 201000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018813025210084031,
+      "loss": 1.8354,
+      "step": 201500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018781512605042016,
+      "loss": 1.833,
+      "step": 202000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018749999999999998,
+      "loss": 1.8308,
+      "step": 202500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018718487394957982,
+      "loss": 1.8314,
+      "step": 203000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018686974789915964,
+      "loss": 1.83,
+      "step": 203500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018655462184873948,
+      "loss": 1.8318,
+      "step": 204000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018623949579831933,
+      "loss": 1.8313,
+      "step": 204500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018592436974789915,
+      "loss": 1.8234,
+      "step": 205000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000185609243697479,
+      "loss": 1.8286,
+      "step": 205500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001852941176470588,
+      "loss": 1.8246,
+      "step": 206000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018497899159663865,
+      "loss": 1.8213,
+      "step": 206500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018466386554621847,
+      "loss": 1.8237,
+      "step": 207000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018434873949579831,
+      "loss": 1.8247,
+      "step": 207500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018403361344537816,
+      "loss": 1.8215,
+      "step": 208000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018371848739495798,
+      "loss": 1.8174,
+      "step": 208500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018340336134453782,
+      "loss": 1.8215,
+      "step": 209000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018308823529411764,
+      "loss": 1.8168,
+      "step": 209500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018277310924369746,
+      "loss": 1.8186,
+      "step": 210000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018245798319327727,
+      "loss": 1.8229,
+      "step": 210500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018214285714285712,
+      "loss": 1.8165,
+      "step": 211000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018182773109243693,
+      "loss": 1.8173,
+      "step": 211500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018151260504201678,
+      "loss": 1.8115,
+      "step": 212000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018119747899159662,
+      "loss": 1.8157,
+      "step": 212500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018088235294117644,
+      "loss": 1.813,
+      "step": 213000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018056722689075629,
+      "loss": 1.8126,
+      "step": 213500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001802521008403361,
+      "loss": 1.808,
+      "step": 214000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017993697478991595,
+      "loss": 1.8116,
+      "step": 214500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017962184873949577,
+      "loss": 1.8165,
+      "step": 215000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001793067226890756,
+      "loss": 1.807,
+      "step": 215500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017899159663865545,
+      "loss": 1.8051,
+      "step": 216000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017867647058823527,
+      "loss": 1.8081,
+      "step": 216500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017836134453781512,
+      "loss": 1.8084,
+      "step": 217000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017804621848739493,
+      "loss": 1.811,
+      "step": 217500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017773109243697478,
+      "loss": 1.8068,
+      "step": 218000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001774159663865546,
+      "loss": 1.805,
+      "step": 218500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017710084033613444,
+      "loss": 1.8042,
+      "step": 219000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017678571428571428,
+      "loss": 1.8004,
+      "step": 219500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001764705882352941,
+      "loss": 1.8028,
+      "step": 220000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017615546218487395,
+      "loss": 1.8111,
+      "step": 220500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017584033613445376,
+      "loss": 1.8076,
+      "step": 221000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001755252100840336,
+      "loss": 1.8017,
+      "step": 221500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017521008403361345,
+      "loss": 1.802,
+      "step": 222000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017489495798319327,
+      "loss": 1.803,
+      "step": 222500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017457983193277312,
+      "loss": 1.8023,
+      "step": 223000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00017426470588235293,
+      "loss": 1.7978,
+      "step": 223500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00017394957983193278,
+      "loss": 1.7959,
+      "step": 224000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001736344537815126,
+      "loss": 1.7971,
+      "step": 224500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001733193277310924,
+      "loss": 1.801,
+      "step": 225000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00017300420168067223,
+      "loss": 1.7987,
+      "step": 225500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00017268907563025207,
+      "loss": 1.7975,
+      "step": 226000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001723739495798319,
+      "loss": 1.7997,
+      "step": 226500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00017205882352941174,
+      "loss": 1.7993,
+      "step": 227000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017174369747899158,
+      "loss": 1.8021,
+      "step": 227500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001714285714285714,
+      "loss": 1.795,
+      "step": 228000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017111344537815124,
+      "loss": 1.7957,
+      "step": 228500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017079831932773106,
+      "loss": 1.7967,
+      "step": 229000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001704831932773109,
+      "loss": 1.7951,
+      "step": 229500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017016806722689075,
+      "loss": 1.7953,
+      "step": 230000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016985294117647057,
+      "loss": 1.7938,
+      "step": 230500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001695378151260504,
+      "loss": 1.7925,
+      "step": 231000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016922268907563023,
+      "loss": 1.7944,
+      "step": 231500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016890756302521007,
+      "loss": 1.7897,
+      "step": 232000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001685924369747899,
+      "loss": 1.7929,
+      "step": 232500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016827731092436974,
+      "loss": 1.7916,
+      "step": 233000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016796218487394958,
+      "loss": 1.7885,
+      "step": 233500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001676470588235294,
+      "loss": 1.7971,
+      "step": 234000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016733193277310924,
+      "loss": 1.7871,
+      "step": 234500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00016701680672268906,
+      "loss": 1.7862,
+      "step": 235000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001667016806722689,
+      "loss": 1.7858,
+      "step": 235500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00016638655462184872,
+      "loss": 1.7859,
+      "step": 236000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00016607142857142857,
+      "loss": 1.7864,
+      "step": 236500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001657563025210084,
+      "loss": 1.7869,
+      "step": 237000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00016544117647058823,
+      "loss": 1.7816,
+      "step": 237500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00016512605042016807,
+      "loss": 1.7865,
+      "step": 238000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001648109243697479,
+      "loss": 1.7867,
+      "step": 238500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016449579831932773,
+      "loss": 1.7868,
+      "step": 239000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016418067226890755,
+      "loss": 1.7795,
+      "step": 239500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016386554621848737,
+      "loss": 1.7833,
+      "step": 240000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016355042016806719,
+      "loss": 1.7885,
+      "step": 240500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016323529411764703,
+      "loss": 1.7886,
+      "step": 241000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016292016806722688,
+      "loss": 1.7865,
+      "step": 241500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001626050420168067,
+      "loss": 1.787,
+      "step": 242000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016228991596638654,
+      "loss": 1.7791,
+      "step": 242500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016197478991596635,
+      "loss": 1.7786,
+      "step": 243000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001616596638655462,
+      "loss": 1.7819,
+      "step": 243500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016134453781512602,
+      "loss": 1.7766,
+      "step": 244000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016102941176470586,
+      "loss": 1.7766,
+      "step": 244500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001607142857142857,
+      "loss": 1.7809,
+      "step": 245000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016039915966386552,
+      "loss": 1.7708,
+      "step": 245500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00016008403361344537,
+      "loss": 1.7713,
+      "step": 246000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015976890756302519,
+      "loss": 1.7773,
+      "step": 246500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015945378151260503,
+      "loss": 1.7746,
+      "step": 247000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015913865546218485,
+      "loss": 1.7722,
+      "step": 247500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001588235294117647,
+      "loss": 1.7707,
+      "step": 248000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015850840336134454,
+      "loss": 1.7785,
+      "step": 248500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015819327731092435,
+      "loss": 1.7745,
+      "step": 249000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001578781512605042,
+      "loss": 1.7746,
+      "step": 249500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015756302521008402,
+      "loss": 1.7706,
+      "step": 250000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015724789915966386,
+      "loss": 1.7689,
+      "step": 250500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001569327731092437,
+      "loss": 1.7702,
+      "step": 251000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015661764705882352,
+      "loss": 1.7716,
+      "step": 251500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015630252100840337,
+      "loss": 1.7709,
+      "step": 252000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015598739495798318,
+      "loss": 1.7689,
+      "step": 252500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015567226890756303,
+      "loss": 1.7709,
+      "step": 253000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015535714285714285,
+      "loss": 1.7666,
+      "step": 253500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001550420168067227,
+      "loss": 1.768,
+      "step": 254000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015472689075630254,
+      "loss": 1.7646,
+      "step": 254500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015441176470588233,
+      "loss": 1.7624,
+      "step": 255000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015409663865546214,
+      "loss": 1.7639,
+      "step": 255500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000153781512605042,
+      "loss": 1.7684,
+      "step": 256000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015346638655462183,
+      "loss": 1.7621,
+      "step": 256500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015315126050420165,
+      "loss": 1.7632,
+      "step": 257000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001528361344537815,
+      "loss": 1.7617,
+      "step": 257500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001525210084033613,
+      "loss": 1.7616,
+      "step": 258000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015220588235294116,
+      "loss": 1.7607,
+      "step": 258500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000151890756302521,
+      "loss": 1.7652,
+      "step": 259000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015157563025210082,
+      "loss": 1.7628,
+      "step": 259500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015126050420168066,
+      "loss": 1.7605,
+      "step": 260000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015094537815126048,
+      "loss": 1.7624,
+      "step": 260500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015063025210084032,
+      "loss": 1.7642,
+      "step": 261000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015031512605042014,
+      "loss": 1.7577,
+      "step": 261500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015,
+      "loss": 1.7586,
+      "step": 262000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014968487394957983,
+      "loss": 1.7582,
+      "step": 262500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00014936974789915965,
+      "loss": 1.7578,
+      "step": 263000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001490546218487395,
+      "loss": 1.7638,
+      "step": 263500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001487394957983193,
+      "loss": 1.7529,
+      "step": 264000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014842436974789916,
+      "loss": 1.7571,
+      "step": 264500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014810924369747897,
+      "loss": 1.7603,
+      "step": 265000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014779411764705882,
+      "loss": 1.7539,
+      "step": 265500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014747899159663863,
+      "loss": 1.7537,
+      "step": 266000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014716386554621848,
+      "loss": 1.7532,
+      "step": 266500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001468487394957983,
+      "loss": 1.7529,
+      "step": 267000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00014653361344537814,
+      "loss": 1.7504,
+      "step": 267500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014621848739495796,
+      "loss": 1.7513,
+      "step": 268000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001459033613445378,
+      "loss": 1.7492,
+      "step": 268500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014558823529411762,
+      "loss": 1.7476,
+      "step": 269000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014527310924369747,
+      "loss": 1.7557,
+      "step": 269500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001449579831932773,
+      "loss": 1.7506,
+      "step": 270000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014464285714285713,
+      "loss": 1.7497,
+      "step": 270500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014432773109243697,
+      "loss": 1.7499,
+      "step": 271000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001440126050420168,
+      "loss": 1.7459,
+      "step": 271500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014369747899159663,
+      "loss": 1.7445,
+      "step": 272000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014338235294117648,
+      "loss": 1.7445,
+      "step": 272500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001430672268907563,
+      "loss": 1.739,
+      "step": 273000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001427521008403361,
+      "loss": 1.7409,
+      "step": 273500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014243697478991596,
+      "loss": 1.7429,
+      "step": 274000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014212184873949578,
+      "loss": 1.741,
+      "step": 274500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014180672268907562,
+      "loss": 1.7431,
+      "step": 275000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014149159663865544,
+      "loss": 1.745,
+      "step": 275500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014117647058823528,
+      "loss": 1.7412,
+      "step": 276000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014086134453781513,
+      "loss": 1.7372,
+      "step": 276500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014054621848739494,
+      "loss": 1.7369,
+      "step": 277000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001402310924369748,
+      "loss": 1.7373,
+      "step": 277500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001399159663865546,
+      "loss": 1.7404,
+      "step": 278000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00013960084033613445,
+      "loss": 1.7424,
+      "step": 278500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013928571428571427,
+      "loss": 1.7359,
+      "step": 279000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001389705882352941,
+      "loss": 1.7376,
+      "step": 279500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013865546218487396,
+      "loss": 1.7362,
+      "step": 280000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013834033613445377,
+      "loss": 1.7405,
+      "step": 280500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001380252100840336,
+      "loss": 1.7375,
+      "step": 281000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013771008403361344,
+      "loss": 1.7355,
+      "step": 281500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013739495798319325,
+      "loss": 1.733,
+      "step": 282000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001370798319327731,
+      "loss": 1.7353,
+      "step": 282500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013676470588235292,
+      "loss": 1.7307,
+      "step": 283000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013644957983193276,
+      "loss": 1.7362,
+      "step": 283500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001361344537815126,
+      "loss": 1.7338,
+      "step": 284000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013581932773109242,
+      "loss": 1.732,
+      "step": 284500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013550420168067227,
+      "loss": 1.7284,
+      "step": 285000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013518907563025208,
+      "loss": 1.7307,
+      "step": 285500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013487394957983193,
+      "loss": 1.7303,
+      "step": 286000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013455882352941175,
+      "loss": 1.7277,
+      "step": 286500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001342436974789916,
+      "loss": 1.7332,
+      "step": 287000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013392857142857144,
+      "loss": 1.7255,
+      "step": 287500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013361344537815125,
+      "loss": 1.7242,
+      "step": 288000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013329831932773107,
+      "loss": 1.7248,
+      "step": 288500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013298319327731091,
+      "loss": 1.731,
+      "step": 289000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00013266806722689073,
+      "loss": 1.7324,
+      "step": 289500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013235294117647058,
+      "loss": 1.7314,
+      "step": 290000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001320378151260504,
+      "loss": 1.7325,
+      "step": 290500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013172268907563024,
+      "loss": 1.7259,
+      "step": 291000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013140756302521008,
+      "loss": 1.7271,
+      "step": 291500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001310924369747899,
+      "loss": 1.7257,
+      "step": 292000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013077731092436975,
+      "loss": 1.7267,
+      "step": 292500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00013046218487394956,
+      "loss": 1.7262,
+      "step": 293000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001301470588235294,
+      "loss": 1.7245,
+      "step": 293500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012983193277310922,
+      "loss": 1.7252,
+      "step": 294000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012951680672268907,
+      "loss": 1.7217,
+      "step": 294500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012920168067226891,
+      "loss": 1.7193,
+      "step": 295000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012888655462184873,
+      "loss": 1.7184,
+      "step": 295500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012857142857142855,
+      "loss": 1.7205,
+      "step": 296000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001282563025210084,
+      "loss": 1.7195,
+      "step": 296500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001279411764705882,
+      "loss": 1.7195,
+      "step": 297000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012762605042016806,
+      "loss": 1.7126,
+      "step": 297500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012731092436974787,
+      "loss": 1.7158,
+      "step": 298000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012699579831932772,
+      "loss": 1.7137,
+      "step": 298500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012668067226890756,
+      "loss": 1.7154,
+      "step": 299000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012636554621848738,
+      "loss": 1.7192,
+      "step": 299500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012605042016806722,
+      "loss": 1.7155,
+      "step": 300000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012573529411764704,
+      "loss": 1.7106,
+      "step": 300500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012542016806722689,
+      "loss": 1.7127,
+      "step": 301000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012510504201680673,
+      "loss": 1.7103,
+      "step": 301500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012478991596638655,
+      "loss": 1.7125,
+      "step": 302000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001244747899159664,
+      "loss": 1.7141,
+      "step": 302500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001241596638655462,
+      "loss": 1.7122,
+      "step": 303000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012384453781512603,
+      "loss": 1.7121,
+      "step": 303500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012352941176470587,
+      "loss": 1.7106,
+      "step": 304000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001232142857142857,
+      "loss": 1.708,
+      "step": 304500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012289915966386553,
+      "loss": 1.7086,
+      "step": 305000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012258403361344538,
+      "loss": 1.7096,
+      "step": 305500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001222689075630252,
+      "loss": 1.7094,
+      "step": 306000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012195378151260503,
+      "loss": 1.7103,
+      "step": 306500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012163865546218486,
+      "loss": 1.7038,
+      "step": 307000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001213235294117647,
+      "loss": 1.707,
+      "step": 307500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012100840336134453,
+      "loss": 1.7094,
+      "step": 308000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012069327731092436,
+      "loss": 1.7078,
+      "step": 308500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001203781512605042,
+      "loss": 1.7043,
+      "step": 309000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00012006302521008403,
+      "loss": 1.7079,
+      "step": 309500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011974789915966386,
+      "loss": 1.7025,
+      "step": 310000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011943277310924367,
+      "loss": 1.7065,
+      "step": 310500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001191176470588235,
+      "loss": 1.7047,
+      "step": 311000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011880252100840335,
+      "loss": 1.6999,
+      "step": 311500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011848739495798318,
+      "loss": 1.7018,
+      "step": 312000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011817226890756301,
+      "loss": 1.7037,
+      "step": 312500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011785714285714284,
+      "loss": 1.6972,
+      "step": 313000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011754201680672267,
+      "loss": 1.6998,
+      "step": 313500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001172268907563025,
+      "loss": 1.7012,
+      "step": 314000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011691176470588235,
+      "loss": 1.6978,
+      "step": 314500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011659663865546218,
+      "loss": 1.7013,
+      "step": 315000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011628151260504201,
+      "loss": 1.6982,
+      "step": 315500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011596638655462184,
+      "loss": 1.699,
+      "step": 316000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011565126050420167,
+      "loss": 1.6956,
+      "step": 316500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001153361344537815,
+      "loss": 1.7124,
+      "step": 317000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011502100840336134,
+      "loss": 1.7101,
+      "step": 317500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011470588235294115,
+      "loss": 1.7024,
+      "step": 318000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000114390756302521,
+      "loss": 1.7011,
+      "step": 318500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011407563025210083,
+      "loss": 1.7009,
+      "step": 319000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011376050420168066,
+      "loss": 1.6989,
+      "step": 319500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011344537815126049,
+      "loss": 1.7002,
+      "step": 320000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011313025210084032,
+      "loss": 1.6989,
+      "step": 320500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011281512605042015,
+      "loss": 1.6977,
+      "step": 321000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001125,
+      "loss": 1.6991,
+      "step": 321500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011218487394957983,
+      "loss": 1.6965,
+      "step": 322000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011186974789915966,
+      "loss": 1.6934,
+      "step": 322500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011155462184873949,
+      "loss": 1.6943,
+      "step": 323000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011123949579831932,
+      "loss": 1.6894,
+      "step": 323500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011092436974789915,
+      "loss": 1.6906,
+      "step": 324000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011060924369747898,
+      "loss": 1.7128,
+      "step": 324500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011029411764705883,
+      "loss": 1.6963,
+      "step": 325000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00010997899159663864,
+      "loss": 1.7206,
+      "step": 325500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00010966386554621848,
+      "loss": 1.7105,
+      "step": 326000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00010934873949579831,
+      "loss": 1.6997,
+      "step": 326500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00010903361344537814,
+      "loss": 1.7016,
+      "step": 327000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00010871848739495797,
+      "loss": 1.7026,
+      "step": 327500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001084033613445378,
+      "loss": 1.6942,
+      "step": 328000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00010808823529411763,
+      "loss": 1.6964,
+      "step": 328500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00010777310924369748,
+      "loss": 1.696,
+      "step": 329000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001074579831932773,
+      "loss": 1.694,
+      "step": 329500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00010714285714285714,
+      "loss": 1.6878,
+      "step": 330000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00010682773109243697,
+      "loss": 1.6921,
+      "step": 330500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001065126050420168,
+      "loss": 1.6871,
+      "step": 331000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00010619747899159663,
+      "loss": 1.6846,
+      "step": 331500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00010588235294117647,
+      "loss": 1.6895,
+      "step": 332000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001055672268907563,
+      "loss": 1.6855,
+      "step": 332500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00010525210084033612,
+      "loss": 1.6807,
+      "step": 333000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00010493697478991595,
+      "loss": 1.6864,
+      "step": 333500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00010462184873949579,
+      "loss": 1.681,
+      "step": 334000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00010430672268907562,
+      "loss": 1.6913,
+      "step": 334500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00010399159663865545,
+      "loss": 1.6789,
+      "step": 335000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00010367647058823528,
+      "loss": 1.6939,
+      "step": 335500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00010336134453781512,
+      "loss": 1.7741,
+      "step": 336000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00010304621848739495,
+      "loss": 1.7619,
+      "step": 336500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00010273109243697478,
+      "loss": 1.7091,
+      "step": 337000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00010241596638655462,
+      "loss": 1.7116,
+      "step": 337500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00010210084033613445,
+      "loss": 1.7243,
+      "step": 338000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00010178571428571428,
+      "loss": 1.7255,
+      "step": 338500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00010147058823529411,
+      "loss": 1.715,
+      "step": 339000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00010115546218487395,
+      "loss": 1.6924,
+      "step": 339500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00010084033613445378,
+      "loss": 1.6824,
+      "step": 340000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001005252100840336,
+      "loss": 1.6885,
+      "step": 340500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00010021008403361343,
+      "loss": 1.6848,
+      "step": 341000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.989495798319326e-05,
+      "loss": 1.6831,
+      "step": 341500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.95798319327731e-05,
+      "loss": 1.6836,
+      "step": 342000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.926470588235293e-05,
+      "loss": 1.6765,
+      "step": 342500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.894957983193276e-05,
+      "loss": 1.6775,
+      "step": 343000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.86344537815126e-05,
+      "loss": 1.6854,
+      "step": 343500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.831932773109243e-05,
+      "loss": 1.6851,
+      "step": 344000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.800420168067226e-05,
+      "loss": 1.6761,
+      "step": 344500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.76890756302521e-05,
+      "loss": 1.6693,
+      "step": 345000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.737394957983193e-05,
+      "loss": 1.6734,
+      "step": 345500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.705882352941176e-05,
+      "loss": 1.6774,
+      "step": 346000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.67436974789916e-05,
+      "loss": 1.6749,
+      "step": 346500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.642857142857143e-05,
+      "loss": 1.6789,
+      "step": 347000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.611344537815126e-05,
+      "loss": 1.6758,
+      "step": 347500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.579831932773108e-05,
+      "loss": 1.6753,
+      "step": 348000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.548319327731091e-05,
+      "loss": 1.6737,
+      "step": 348500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.516806722689074e-05,
+      "loss": 1.6723,
+      "step": 349000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.485294117647057e-05,
+      "loss": 1.6752,
+      "step": 349500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.45378151260504e-05,
+      "loss": 1.6706,
+      "step": 350000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.422268907563025e-05,
+      "loss": 1.669,
+      "step": 350500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.390756302521008e-05,
+      "loss": 1.6694,
+      "step": 351000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.359243697478991e-05,
+      "loss": 1.6677,
+      "step": 351500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.327731092436974e-05,
+      "loss": 1.6709,
+      "step": 352000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.296218487394957e-05,
+      "loss": 1.6645,
+      "step": 352500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.26470588235294e-05,
+      "loss": 1.6648,
+      "step": 353000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.233193277310923e-05,
+      "loss": 1.6717,
+      "step": 353500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.201680672268908e-05,
+      "loss": 1.6614,
+      "step": 354000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.170168067226891e-05,
+      "loss": 1.664,
+      "step": 354500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.138655462184873e-05,
+      "loss": 1.6681,
+      "step": 355000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.107142857142856e-05,
+      "loss": 1.6683,
+      "step": 355500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.075630252100839e-05,
+      "loss": 1.6639,
+      "step": 356000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.044117647058822e-05,
+      "loss": 1.6637,
+      "step": 356500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.012605042016805e-05,
+      "loss": 1.6576,
+      "step": 357000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.981092436974788e-05,
+      "loss": 1.6616,
+      "step": 357500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.949579831932773e-05,
+      "loss": 1.6604,
+      "step": 358000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.918067226890756e-05,
+      "loss": 1.6611,
+      "step": 358500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.886554621848739e-05,
+      "loss": 1.6597,
+      "step": 359000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.855042016806722e-05,
+      "loss": 1.6613,
+      "step": 359500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.823529411764705e-05,
+      "loss": 1.6588,
+      "step": 360000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.792016806722688e-05,
+      "loss": 1.6573,
+      "step": 360500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.760504201680673e-05,
+      "loss": 1.6587,
+      "step": 361000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.728991596638656e-05,
+      "loss": 1.6581,
+      "step": 361500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.697478991596639e-05,
+      "loss": 1.6531,
+      "step": 362000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.66596638655462e-05,
+      "loss": 1.6542,
+      "step": 362500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.634453781512604e-05,
+      "loss": 1.6545,
+      "step": 363000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.602941176470587e-05,
+      "loss": 1.6519,
+      "step": 363500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.57142857142857e-05,
+      "loss": 1.6557,
+      "step": 364000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.539915966386553e-05,
+      "loss": 1.6518,
+      "step": 364500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.508403361344537e-05,
+      "loss": 1.6531,
+      "step": 365000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.47689075630252e-05,
+      "loss": 1.6481,
+      "step": 365500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.445378151260504e-05,
+      "loss": 1.6475,
+      "step": 366000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.413865546218487e-05,
+      "loss": 1.6491,
+      "step": 366500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.38235294117647e-05,
+      "loss": 1.6556,
+      "step": 367000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.350840336134453e-05,
+      "loss": 1.6472,
+      "step": 367500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.319327731092436e-05,
+      "loss": 1.6477,
+      "step": 368000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.28781512605042e-05,
+      "loss": 1.6496,
+      "step": 368500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.256302521008404e-05,
+      "loss": 1.6479,
+      "step": 369000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.224789915966387e-05,
+      "loss": 1.6492,
+      "step": 369500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.193277310924368e-05,
+      "loss": 1.6443,
+      "step": 370000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.161764705882352e-05,
+      "loss": 1.6443,
+      "step": 370500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.130252100840335e-05,
+      "loss": 1.6462,
+      "step": 371000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.098739495798318e-05,
+      "loss": 1.6454,
+      "step": 371500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.067226890756301e-05,
+      "loss": 1.6416,
+      "step": 372000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.035714285714285e-05,
+      "loss": 1.6433,
+      "step": 372500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.004201680672268e-05,
+      "loss": 1.6447,
+      "step": 373000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.972689075630251e-05,
+      "loss": 1.6454,
+      "step": 373500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.941176470588235e-05,
+      "loss": 1.6402,
+      "step": 374000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.909663865546218e-05,
+      "loss": 1.642,
+      "step": 374500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.878151260504201e-05,
+      "loss": 1.6401,
+      "step": 375000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.846638655462185e-05,
+      "loss": 1.6446,
+      "step": 375500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.815126050420168e-05,
+      "loss": 1.6374,
+      "step": 376000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.783613445378151e-05,
+      "loss": 1.6425,
+      "step": 376500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.752100840336135e-05,
+      "loss": 1.6418,
+      "step": 377000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.720588235294116e-05,
+      "loss": 1.6407,
+      "step": 377500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.6890756302521e-05,
+      "loss": 1.6373,
+      "step": 378000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.657563025210082e-05,
+      "loss": 1.6397,
+      "step": 378500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.626050420168066e-05,
+      "loss": 1.6408,
+      "step": 379000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.59453781512605e-05,
+      "loss": 1.6407,
+      "step": 379500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.563025210084033e-05,
+      "loss": 1.6397,
+      "step": 380000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.531512605042016e-05,
+      "loss": 1.6406,
+      "step": 380500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.5e-05,
+      "loss": 1.6399,
+      "step": 381000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.468487394957982e-05,
+      "loss": 1.6403,
+      "step": 381500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.436974789915966e-05,
+      "loss": 1.6394,
+      "step": 382000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.405462184873949e-05,
+      "loss": 1.6377,
+      "step": 382500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.373949579831932e-05,
+      "loss": 1.6365,
+      "step": 383000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.342436974789915e-05,
+      "loss": 1.6329,
+      "step": 383500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.310924369747898e-05,
+      "loss": 1.6361,
+      "step": 384000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.279411764705881e-05,
+      "loss": 1.6325,
+      "step": 384500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.247899159663865e-05,
+      "loss": 1.6347,
+      "step": 385000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.216386554621849e-05,
+      "loss": 1.6364,
+      "step": 385500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.184873949579832e-05,
+      "loss": 1.6293,
+      "step": 386000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.153361344537815e-05,
+      "loss": 1.6306,
+      "step": 386500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.121848739495798e-05,
+      "loss": 1.6308,
+      "step": 387000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.090336134453781e-05,
+      "loss": 1.6315,
+      "step": 387500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.058823529411764e-05,
+      "loss": 1.6326,
+      "step": 388000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.027310924369747e-05,
+      "loss": 1.6296,
+      "step": 388500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.99579831932773e-05,
+      "loss": 1.6332,
+      "step": 389000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.964285714285713e-05,
+      "loss": 1.6337,
+      "step": 389500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.932773109243698e-05,
+      "loss": 1.6279,
+      "step": 390000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.90126050420168e-05,
+      "loss": 1.6296,
+      "step": 390500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.869747899159663e-05,
+      "loss": 1.6244,
+      "step": 391000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.838235294117646e-05,
+      "loss": 1.6323,
+      "step": 391500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.80672268907563e-05,
+      "loss": 1.63,
+      "step": 392000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.775210084033613e-05,
+      "loss": 1.6253,
+      "step": 392500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.743697478991596e-05,
+      "loss": 1.623,
+      "step": 393000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.71218487394958e-05,
+      "loss": 1.6291,
+      "step": 393500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.680672268907563e-05,
+      "loss": 1.6264,
+      "step": 394000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.649159663865546e-05,
+      "loss": 1.6278,
+      "step": 394500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.617647058823529e-05,
+      "loss": 1.6274,
+      "step": 395000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.586134453781512e-05,
+      "loss": 1.6254,
+      "step": 395500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.554621848739495e-05,
+      "loss": 1.6237,
+      "step": 396000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.523109243697478e-05,
+      "loss": 1.6232,
+      "step": 396500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.491596638655461e-05,
+      "loss": 1.6204,
+      "step": 397000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.460084033613446e-05,
+      "loss": 1.6243,
+      "step": 397500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.428571428571427e-05,
+      "loss": 1.6234,
+      "step": 398000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.39705882352941e-05,
+      "loss": 1.6237,
+      "step": 398500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.365546218487394e-05,
+      "loss": 1.6239,
+      "step": 399000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.334033613445378e-05,
+      "loss": 1.6215,
+      "step": 399500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.302521008403361e-05,
+      "loss": 1.6229,
+      "step": 400000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.271008403361344e-05,
+      "loss": 1.6197,
+      "step": 400500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.239495798319327e-05,
+      "loss": 1.6183,
+      "step": 401000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.20798319327731e-05,
+      "loss": 1.6206,
+      "step": 401500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.176470588235294e-05,
+      "loss": 1.6182,
+      "step": 402000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.144957983193277e-05,
+      "loss": 1.621,
+      "step": 402500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.11344537815126e-05,
+      "loss": 1.6204,
+      "step": 403000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.081932773109243e-05,
+      "loss": 1.6213,
+      "step": 403500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.0504201680672267e-05,
+      "loss": 1.6187,
+      "step": 404000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.01890756302521e-05,
+      "loss": 1.617,
+      "step": 404500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.987394957983193e-05,
+      "loss": 1.6145,
+      "step": 405000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.955882352941175e-05,
+      "loss": 1.6146,
+      "step": 405500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.924369747899159e-05,
+      "loss": 1.615,
+      "step": 406000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.892857142857142e-05,
+      "loss": 1.6138,
+      "step": 406500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.861344537815125e-05,
+      "loss": 1.6158,
+      "step": 407000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.829831932773109e-05,
+      "loss": 1.6149,
+      "step": 407500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.798319327731092e-05,
+      "loss": 1.6198,
+      "step": 408000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.766806722689075e-05,
+      "loss": 1.6117,
+      "step": 408500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.7352941176470576e-05,
+      "loss": 1.6141,
+      "step": 409000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.7037815126050414e-05,
+      "loss": 1.6113,
+      "step": 409500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.6722689075630245e-05,
+      "loss": 1.6126,
+      "step": 410000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.6407563025210076e-05,
+      "loss": 1.612,
+      "step": 410500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.6092436974789914e-05,
+      "loss": 1.6067,
+      "step": 411000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.5777310924369745e-05,
+      "loss": 1.6128,
+      "step": 411500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.5462184873949576e-05,
+      "loss": 1.6148,
+      "step": 412000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.5147058823529414e-05,
+      "loss": 1.61,
+      "step": 412500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.483193277310924e-05,
+      "loss": 1.6094,
+      "step": 413000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.451680672268907e-05,
+      "loss": 1.6108,
+      "step": 413500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.42016806722689e-05,
+      "loss": 1.6029,
+      "step": 414000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.388655462184874e-05,
+      "loss": 1.608,
+      "step": 414500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.357142857142857e-05,
+      "loss": 1.6046,
+      "step": 415000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.32563025210084e-05,
+      "loss": 1.6096,
+      "step": 415500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.294117647058824e-05,
+      "loss": 1.6056,
+      "step": 416000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.262605042016806e-05,
+      "loss": 1.6055,
+      "step": 416500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.231092436974789e-05,
+      "loss": 1.6019,
+      "step": 417000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.1995798319327724e-05,
+      "loss": 1.603,
+      "step": 417500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.168067226890756e-05,
+      "loss": 1.6007,
+      "step": 418000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.136554621848739e-05,
+      "loss": 1.6038,
+      "step": 418500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.105042016806722e-05,
+      "loss": 1.6017,
+      "step": 419000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.0735294117647054e-05,
+      "loss": 1.6051,
+      "step": 419500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.042016806722689e-05,
+      "loss": 1.5992,
+      "step": 420000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.0105042016806716e-05,
+      "loss": 1.6022,
+      "step": 420500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.978991596638655e-05,
+      "loss": 1.6038,
+      "step": 421000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.947478991596638e-05,
+      "loss": 1.6019,
+      "step": 421500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.9159663865546216e-05,
+      "loss": 1.6006,
+      "step": 422000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.884453781512605e-05,
+      "loss": 1.6046,
+      "step": 422500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.852941176470588e-05,
+      "loss": 1.6049,
+      "step": 423000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.8214285714285716e-05,
+      "loss": 1.6004,
+      "step": 423500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.789915966386554e-05,
+      "loss": 1.6016,
+      "step": 424000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.758403361344537e-05,
+      "loss": 1.6024,
+      "step": 424500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.72689075630252e-05,
+      "loss": 1.5994,
+      "step": 425000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.695378151260504e-05,
+      "loss": 1.5989,
+      "step": 425500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.663865546218487e-05,
+      "loss": 1.599,
+      "step": 426000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.63235294117647e-05,
+      "loss": 1.5968,
+      "step": 426500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.600840336134454e-05,
+      "loss": 1.5968,
+      "step": 427000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.5693277310924364e-05,
+      "loss": 1.5981,
+      "step": 427500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.5378151260504195e-05,
+      "loss": 1.5961,
+      "step": 428000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.5063025210084026e-05,
+      "loss": 1.5967,
+      "step": 428500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4747899159663864e-05,
+      "loss": 1.5963,
+      "step": 429000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4432773109243695e-05,
+      "loss": 1.5937,
+      "step": 429500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4117647058823526e-05,
+      "loss": 1.5963,
+      "step": 430000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.380252100840336e-05,
+      "loss": 1.5961,
+      "step": 430500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.3487394957983194e-05,
+      "loss": 1.5955,
+      "step": 431000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.317226890756302e-05,
+      "loss": 1.5905,
+      "step": 431500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.285714285714285e-05,
+      "loss": 1.5956,
+      "step": 432000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.254201680672269e-05,
+      "loss": 1.5938,
+      "step": 432500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.222689075630252e-05,
+      "loss": 1.5939,
+      "step": 433000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.191176470588235e-05,
+      "loss": 1.5919,
+      "step": 433500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.159663865546218e-05,
+      "loss": 1.5905,
+      "step": 434000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.128151260504202e-05,
+      "loss": 1.589,
+      "step": 434500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.096638655462184e-05,
+      "loss": 1.592,
+      "step": 435000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.065126050420167e-05,
+      "loss": 1.594,
+      "step": 435500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.0336134453781504e-05,
+      "loss": 1.5908,
+      "step": 436000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.002100840336134e-05,
+      "loss": 1.5876,
+      "step": 436500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.970588235294117e-05,
+      "loss": 1.5899,
+      "step": 437000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9390756302521004e-05,
+      "loss": 1.5912,
+      "step": 437500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.907563025210084e-05,
+      "loss": 1.5885,
+      "step": 438000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.876050420168067e-05,
+      "loss": 1.5914,
+      "step": 438500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.84453781512605e-05,
+      "loss": 1.5905,
+      "step": 439000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.813025210084033e-05,
+      "loss": 1.5885,
+      "step": 439500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.7815126050420166e-05,
+      "loss": 1.5861,
+      "step": 440000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.75e-05,
+      "loss": 1.5877,
+      "step": 440500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.718487394957983e-05,
+      "loss": 1.5846,
+      "step": 441000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.686974789915966e-05,
+      "loss": 1.5875,
+      "step": 441500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.655462184873949e-05,
+      "loss": 1.5854,
+      "step": 442000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.623949579831933e-05,
+      "loss": 1.5824,
+      "step": 442500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.592436974789916e-05,
+      "loss": 1.5847,
+      "step": 443000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.560924369747899e-05,
+      "loss": 1.5848,
+      "step": 443500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.529411764705882e-05,
+      "loss": 1.5862,
+      "step": 444000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.497899159663865e-05,
+      "loss": 1.583,
+      "step": 444500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.466386554621849e-05,
+      "loss": 1.5854,
+      "step": 445000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.4348739495798313e-05,
+      "loss": 1.584,
+      "step": 445500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.403361344537815e-05,
+      "loss": 1.5825,
+      "step": 446000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.371848739495798e-05,
+      "loss": 1.5825,
+      "step": 446500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.340336134453781e-05,
+      "loss": 1.58,
+      "step": 447000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.3088235294117644e-05,
+      "loss": 1.5808,
+      "step": 447500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.2773109243697475e-05,
+      "loss": 1.5785,
+      "step": 448000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.2457983193277306e-05,
+      "loss": 1.5825,
+      "step": 448500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.214285714285714e-05,
+      "loss": 1.5783,
+      "step": 449000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.182773109243697e-05,
+      "loss": 1.5762,
+      "step": 449500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.1512605042016806e-05,
+      "loss": 1.5771,
+      "step": 450000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.119747899159664e-05,
+      "loss": 1.5822,
+      "step": 450500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.088235294117647e-05,
+      "loss": 1.5834,
+      "step": 451000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.05672268907563e-05,
+      "loss": 1.5763,
+      "step": 451500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.0252100840336133e-05,
+      "loss": 1.5771,
+      "step": 452000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9936974789915964e-05,
+      "loss": 1.5771,
+      "step": 452500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9621848739495795e-05,
+      "loss": 1.5819,
+      "step": 453000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9306722689075626e-05,
+      "loss": 1.5786,
+      "step": 453500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.899159663865546e-05,
+      "loss": 1.579,
+      "step": 454000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.8676470588235288e-05,
+      "loss": 1.5774,
+      "step": 454500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.8361344537815123e-05,
+      "loss": 1.5733,
+      "step": 455000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.8046218487394957e-05,
+      "loss": 1.5743,
+      "step": 455500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7731092436974788e-05,
+      "loss": 1.5715,
+      "step": 456000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.741596638655462e-05,
+      "loss": 1.5767,
+      "step": 456500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.710084033613445e-05,
+      "loss": 1.5753,
+      "step": 457000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.6785714285714284e-05,
+      "loss": 1.5726,
+      "step": 457500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.647058823529412e-05,
+      "loss": 1.5729,
+      "step": 458000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.6155462184873946e-05,
+      "loss": 1.5735,
+      "step": 458500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.584033613445378e-05,
+      "loss": 1.5719,
+      "step": 459000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.552521008403361e-05,
+      "loss": 1.5673,
+      "step": 459500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.5210084033613446e-05,
+      "loss": 1.5746,
+      "step": 460000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.4894957983193274e-05,
+      "loss": 1.5715,
+      "step": 460500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.4579831932773108e-05,
+      "loss": 1.5698,
+      "step": 461000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.426470588235294e-05,
+      "loss": 1.569,
+      "step": 461500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.394957983193277e-05,
+      "loss": 1.5693,
+      "step": 462000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.36344537815126e-05,
+      "loss": 1.5718,
+      "step": 462500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.3319327731092435e-05,
+      "loss": 1.5704,
+      "step": 463000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.300420168067227e-05,
+      "loss": 1.566,
+      "step": 463500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.2689075630252097e-05,
+      "loss": 1.5702,
+      "step": 464000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.2373949579831932e-05,
+      "loss": 1.572,
+      "step": 464500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.2058823529411763e-05,
+      "loss": 1.5689,
+      "step": 465000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.1743697478991597e-05,
+      "loss": 1.5692,
+      "step": 465500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.1428571428571425e-05,
+      "loss": 1.5679,
+      "step": 466000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.111344537815126e-05,
+      "loss": 1.5645,
+      "step": 466500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.079831932773109e-05,
+      "loss": 1.5667,
+      "step": 467000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.048319327731092e-05,
+      "loss": 1.5659,
+      "step": 467500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.0168067226890752e-05,
+      "loss": 1.5628,
+      "step": 468000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9852941176470586e-05,
+      "loss": 1.5622,
+      "step": 468500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.953781512605042e-05,
+      "loss": 1.5674,
+      "step": 469000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.922268907563025e-05,
+      "loss": 1.5645,
+      "step": 469500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8907563025210083e-05,
+      "loss": 1.5647,
+      "step": 470000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.8592436974789914e-05,
+      "loss": 1.5641,
+      "step": 470500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.8277310924369745e-05,
+      "loss": 1.5656,
+      "step": 471000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.796218487394958e-05,
+      "loss": 1.5635,
+      "step": 471500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.764705882352941e-05,
+      "loss": 1.5612,
+      "step": 472000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7331932773109245e-05,
+      "loss": 1.5619,
+      "step": 472500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7016806722689076e-05,
+      "loss": 1.5643,
+      "step": 473000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.6701680672268907e-05,
+      "loss": 1.5607,
+      "step": 473500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.6386554621848738e-05,
+      "loss": 1.5642,
+      "step": 474000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.607142857142857e-05,
+      "loss": 1.5603,
+      "step": 474500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5756302521008403e-05,
+      "loss": 1.5591,
+      "step": 475000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5441176470588234e-05,
+      "loss": 1.5632,
+      "step": 475500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5126050420168067e-05,
+      "loss": 1.5606,
+      "step": 476000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4810924369747898e-05,
+      "loss": 1.5598,
+      "step": 476500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.449579831932773e-05,
+      "loss": 1.5616,
+      "step": 477000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4180672268907561e-05,
+      "loss": 1.5609,
+      "step": 477500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.3865546218487394e-05,
+      "loss": 1.56,
+      "step": 478000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3550420168067225e-05,
+      "loss": 1.5609,
+      "step": 478500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.323529411764706e-05,
+      "loss": 1.5603,
+      "step": 479000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.292016806722689e-05,
+      "loss": 1.5612,
+      "step": 479500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2605042016806723e-05,
+      "loss": 1.5655,
+      "step": 480000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2289915966386554e-05,
+      "loss": 1.5588,
+      "step": 480500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1974789915966385e-05,
+      "loss": 1.561,
+      "step": 481000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1659663865546218e-05,
+      "loss": 1.5585,
+      "step": 481500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1344537815126049e-05,
+      "loss": 1.5569,
+      "step": 482000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1029411764705881e-05,
+      "loss": 1.5576,
+      "step": 482500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0714285714285712e-05,
+      "loss": 1.5551,
+      "step": 483000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0399159663865545e-05,
+      "loss": 1.5576,
+      "step": 483500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0084033613445376e-05,
+      "loss": 1.558,
+      "step": 484000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.76890756302521e-06,
+      "loss": 1.5595,
+      "step": 484500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.453781512605041e-06,
+      "loss": 1.5606,
+      "step": 485000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.138655462184872e-06,
+      "loss": 1.5629,
+      "step": 485500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.823529411764705e-06,
+      "loss": 1.5605,
+      "step": 486000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.508403361344538e-06,
+      "loss": 1.5582,
+      "step": 486500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.193277310924369e-06,
+      "loss": 1.5558,
+      "step": 487000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.878151260504201e-06,
+      "loss": 1.5562,
+      "step": 487500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.563025210084033e-06,
+      "loss": 1.559,
+      "step": 488000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.247899159663865e-06,
+      "loss": 1.5608,
+      "step": 488500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.932773109243697e-06,
+      "loss": 1.5595,
+      "step": 489000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.61764705882353e-06,
+      "loss": 1.5569,
+      "step": 489500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.3025210084033615e-06,
+      "loss": 1.5545,
+      "step": 490000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.9873949579831925e-06,
+      "loss": 1.557,
+      "step": 490500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.672268907563024e-06,
+      "loss": 1.5559,
+      "step": 491000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.357142857142856e-06,
+      "loss": 1.5577,
+      "step": 491500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.042016806722688e-06,
+      "loss": 1.5555,
+      "step": 492000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.726890756302521e-06,
+      "loss": 1.5554,
+      "step": 492500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.4117647058823526e-06,
+      "loss": 1.5552,
+      "step": 493000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.096638655462184e-06,
+      "loss": 1.5573,
+      "step": 493500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.7815126050420167e-06,
+      "loss": 1.5555,
+      "step": 494000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.4663865546218485e-06,
+      "loss": 1.5551,
+      "step": 494500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.1512605042016808e-06,
+      "loss": 1.5499,
+      "step": 495000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.836134453781512e-06,
+      "loss": 1.5534,
+      "step": 495500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.521008403361344e-06,
+      "loss": 1.5509,
+      "step": 496000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.2058823529411763e-06,
+      "loss": 1.5551,
+      "step": 496500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8907563025210083e-06,
+      "loss": 1.5567,
+      "step": 497000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.5756302521008404e-06,
+      "loss": 1.5546,
+      "step": 497500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.260504201680672e-06,
+      "loss": 1.5567,
+      "step": 498000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.453781512605042e-07,
+      "loss": 1.5546,
+      "step": 498500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.30252100840336e-07,
+      "loss": 1.5575,
+      "step": 499000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.15126050420168e-07,
+      "loss": 1.5552,
+      "step": 499500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0,
+      "loss": 1.552,
+      "step": 500000
+    },
+    {
+      "epoch": 1.35,
+      "step": 500000,
+      "total_flos": 8.422691657052488e+18,
+      "train_loss": 1.8320032868652343,
+      "train_runtime": 99252.1755,
+      "train_samples_per_second": 1289.644,
+      "train_steps_per_second": 5.038
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 500000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 10000,
+  "total_flos": 8.422691657052488e+18,
+  "train_batch_size": 256,
+  "trial_name": null,
+  "trial_params": null
+}