diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9025 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.4282335572164673,
+  "global_step": 1500000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-06,
+      "loss": 0.8613,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-06,
+      "loss": 0.7335,
+      "step": 2000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 0.7278,
+      "step": 3000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8e-06,
+      "loss": 0.7196,
+      "step": 4000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999999999999999e-06,
+      "loss": 0.7155,
+      "step": 5000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 0.713,
+      "step": 6000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4e-05,
+      "loss": 0.7049,
+      "step": 7000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6e-05,
+      "loss": 0.6121,
+      "step": 8000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 0.5451,
+      "step": 9000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 0.4989,
+      "step": 10000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2e-05,
+      "loss": 0.4653,
+      "step": 11000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 0.4412,
+      "step": 12000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6e-05,
+      "loss": 0.4231,
+      "step": 13000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8e-05,
+      "loss": 0.408,
+      "step": 14000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 0.3962,
+      "step": 15000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2e-05,
+      "loss": 0.3866,
+      "step": 16000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.399999999999999e-05,
+      "loss": 0.3774,
+      "step": 17000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 0.3694,
+      "step": 18000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8e-05,
+      "loss": 0.3625,
+      "step": 19000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999999999999996e-05,
+      "loss": 0.3557,
+      "step": 20000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2e-05,
+      "loss": 0.3495,
+      "step": 21000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.4e-05,
+      "loss": 0.3446,
+      "step": 22000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.599999999999999e-05,
+      "loss": 0.3392,
+      "step": 23000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 0.3348,
+      "step": 24000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9999999999999996e-05,
+      "loss": 0.3302,
+      "step": 25000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.2e-05,
+      "loss": 0.3257,
+      "step": 26000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 0.322,
+      "step": 27000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.6e-05,
+      "loss": 0.3176,
+      "step": 28000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.7999999999999994e-05,
+      "loss": 0.3137,
+      "step": 29000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 0.3105,
+      "step": 30000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.199999999999999e-05,
+      "loss": 0.3073,
+      "step": 31000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.4e-05,
+      "loss": 0.3039,
+      "step": 32000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 0.3006,
+      "step": 33000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.799999999999999e-05,
+      "loss": 0.2979,
+      "step": 34000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7e-05,
+      "loss": 0.2951,
+      "step": 35000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 0.2925,
+      "step": 36000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.4e-05,
+      "loss": 0.2892,
+      "step": 37000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.6e-05,
+      "loss": 0.2866,
+      "step": 38000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.8e-05,
+      "loss": 0.285,
+      "step": 39000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.999999999999999e-05,
+      "loss": 0.2818,
+      "step": 40000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.199999999999999e-05,
+      "loss": 0.2799,
+      "step": 41000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.4e-05,
+      "loss": 0.2777,
+      "step": 42000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.6e-05,
+      "loss": 0.2755,
+      "step": 43000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.8e-05,
+      "loss": 0.2732,
+      "step": 44000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 0.2717,
+      "step": 45000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.199999999999999e-05,
+      "loss": 0.2695,
+      "step": 46000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.4e-05,
+      "loss": 0.2677,
+      "step": 47000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 0.2662,
+      "step": 48000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.799999999999998e-05,
+      "loss": 0.2642,
+      "step": 49000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.999999999999999e-05,
+      "loss": 0.2626,
+      "step": 50000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000102,
+      "loss": 0.261,
+      "step": 51000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000104,
+      "loss": 0.2599,
+      "step": 52000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00010599999999999999,
+      "loss": 0.2582,
+      "step": 53000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 0.256,
+      "step": 54000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00010999999999999998,
+      "loss": 0.2554,
+      "step": 55000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000112,
+      "loss": 0.2539,
+      "step": 56000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 0.2522,
+      "step": 57000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011599999999999999,
+      "loss": 0.2515,
+      "step": 58000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00011799999999999998,
+      "loss": 0.2499,
+      "step": 59000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 0.2486,
+      "step": 60000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000122,
+      "loss": 0.2477,
+      "step": 61000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00012399999999999998,
+      "loss": 0.2455,
+      "step": 62000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 0.2453,
+      "step": 63000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000128,
+      "loss": 0.2442,
+      "step": 64000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00013,
+      "loss": 0.2434,
+      "step": 65000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 0.2414,
+      "step": 66000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00013399999999999998,
+      "loss": 0.2408,
+      "step": 67000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00013599999999999997,
+      "loss": 0.2404,
+      "step": 68000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000138,
+      "loss": 0.2392,
+      "step": 69000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014,
+      "loss": 0.2381,
+      "step": 70000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014199999999999998,
+      "loss": 0.2375,
+      "step": 71000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 0.2362,
+      "step": 72000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000146,
+      "loss": 0.236,
+      "step": 73000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000148,
+      "loss": 0.2352,
+      "step": 74000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00015,
+      "loss": 0.2333,
+      "step": 75000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014999982988685054,
+      "loss": 0.2334,
+      "step": 76000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014999931954822897,
+      "loss": 0.2316,
+      "step": 77000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014999846898661572,
+      "loss": 0.2314,
+      "step": 78000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014999727820614486,
+      "loss": 0.2306,
+      "step": 79000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000149995747212604,
+      "loss": 0.2288,
+      "step": 80000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014999387601343436,
+      "loss": 0.2275,
+      "step": 81000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001499916646177307,
+      "loss": 0.2268,
+      "step": 82000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014998911303624118,
+      "loss": 0.2257,
+      "step": 83000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00014998622128136748,
+      "loss": 0.2249,
+      "step": 84000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001499829893671646,
+      "loss": 0.2246,
+      "step": 85000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00014997941730934087,
+      "loss": 0.2232,
+      "step": 86000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00014997550512525784,
+      "loss": 0.2215,
+      "step": 87000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00014997125283393016,
+      "loss": 0.2215,
+      "step": 88000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00014996666045602558,
+      "loss": 0.2202,
+      "step": 89000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00014996172801386482,
+      "loss": 0.2201,
+      "step": 90000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00014995645553142134,
+      "loss": 0.2187,
+      "step": 91000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001499508430343214,
+      "loss": 0.2185,
+      "step": 92000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001499448905498439,
+      "loss": 0.2173,
+      "step": 93000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00014993859810692007,
+      "loss": 0.2164,
+      "step": 94000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001499319657361336,
+      "loss": 0.2168,
+      "step": 95000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001499249934697203,
+      "loss": 0.2148,
+      "step": 96000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00014991768134156797,
+      "loss": 0.2145,
+      "step": 97000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00014991002938721633,
+      "loss": 0.2137,
+      "step": 98000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00014990203764385677,
+      "loss": 0.2129,
+      "step": 99000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00014989370615033212,
+      "loss": 0.2127,
+      "step": 100000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001498850349471366,
+      "loss": 0.2114,
+      "step": 101000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001498760240764155,
+      "loss": 0.2113,
+      "step": 102000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00014986667358196504,
+      "loss": 0.2097,
+      "step": 103000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00014985698350923214,
+      "loss": 0.2099,
+      "step": 104000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001498469539053142,
+      "loss": 0.2089,
+      "step": 105000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00014983658481895886,
+      "loss": 0.2083,
+      "step": 106000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00014982587630056377,
+      "loss": 0.2081,
+      "step": 107000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00014981482840217632,
+      "loss": 0.2076,
+      "step": 108000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00014980344117749355,
+      "loss": 0.2068,
+      "step": 109000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00014979171468186152,
+      "loss": 0.2064,
+      "step": 110000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00014977964897227547,
+      "loss": 0.2054,
+      "step": 111000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001497672441073793,
+      "loss": 0.2047,
+      "step": 112000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001497545001474653,
+      "loss": 0.2048,
+      "step": 113000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00014974141715447386,
+      "loss": 0.2041,
+      "step": 114000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00014972799519199332,
+      "loss": 0.203,
+      "step": 115000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00014971423432525937,
+      "loss": 0.2025,
+      "step": 116000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00014970013462115505,
+      "loss": 0.2018,
+      "step": 117000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00014968569614821022,
+      "loss": 0.2018,
+      "step": 118000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00014967091897660125,
+      "loss": 0.2011,
+      "step": 119000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00014965580317815078,
+      "loss": 0.2008,
+      "step": 120000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00014964034882632726,
+      "loss": 0.2004,
+      "step": 121000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00014962455599624463,
+      "loss": 0.1998,
+      "step": 122000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000149608424764662,
+      "loss": 0.1995,
+      "step": 123000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00014959195520998318,
+      "loss": 0.199,
+      "step": 124000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00014957514741225646,
+      "loss": 0.1978,
+      "step": 125000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00014955800145317397,
+      "loss": 0.1969,
+      "step": 126000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00014954051741607156,
+      "loss": 0.1978,
+      "step": 127000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00014952269538592825,
+      "loss": 0.1963,
+      "step": 128000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001495045354493657,
+      "loss": 0.1967,
+      "step": 129000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00014948603769464813,
+      "loss": 0.1964,
+      "step": 130000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00014946720221168156,
+      "loss": 0.1954,
+      "step": 131000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00014944802909201344,
+      "loss": 0.1955,
+      "step": 132000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001494285184288324,
+      "loss": 0.1953,
+      "step": 133000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00014940867031696759,
+      "loss": 0.1937,
+      "step": 134000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00014938848485288825,
+      "loss": 0.1945,
+      "step": 135000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00014936796213470336,
+      "loss": 0.1942,
+      "step": 136000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00014934710226216104,
+      "loss": 0.1936,
+      "step": 137000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00014932590533664808,
+      "loss": 0.1931,
+      "step": 138000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00014930437146118953,
+      "loss": 0.1925,
+      "step": 139000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001492825007404481,
+      "loss": 0.1923,
+      "step": 140000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001492602932807237,
+      "loss": 0.1918,
+      "step": 141000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00014923774918995295,
+      "loss": 0.1911,
+      "step": 142000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001492148685777086,
+      "loss": 0.1909,
+      "step": 143000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000149191651555199,
+      "loss": 0.1905,
+      "step": 144000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001491680982352676,
+      "loss": 0.1901,
+      "step": 145000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001491442087323924,
+      "loss": 0.1898,
+      "step": 146000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00014911998316268537,
+      "loss": 0.1894,
+      "step": 147000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001490954216438919,
+      "loss": 0.1896,
+      "step": 148000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00014907052429539012,
+      "loss": 0.1889,
+      "step": 149000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00014904529123819054,
+      "loss": 0.1882,
+      "step": 150000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00014901972259493535,
+      "loss": 0.1882,
+      "step": 151000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00014899381848989767,
+      "loss": 0.1881,
+      "step": 152000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00014896757904898125,
+      "loss": 0.1876,
+      "step": 153000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00014894100439971958,
+      "loss": 0.1868,
+      "step": 154000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001489140946712755,
+      "loss": 0.1868,
+      "step": 155000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00014888684999444035,
+      "loss": 0.1861,
+      "step": 156000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001488592705016335,
+      "loss": 0.1867,
+      "step": 157000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00014883135632690167,
+      "loss": 0.1855,
+      "step": 158000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00014880310760591824,
+      "loss": 0.1852,
+      "step": 159000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001487745244759826,
+      "loss": 0.1848,
+      "step": 160000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00014874560707601949,
+      "loss": 0.1846,
+      "step": 161000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001487163555465783,
+      "loss": 0.1843,
+      "step": 162000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00014868677002983252,
+      "loss": 0.1844,
+      "step": 163000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00014865685066957885,
+      "loss": 0.1837,
+      "step": 164000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00014862659761123663,
+      "loss": 0.1839,
+      "step": 165000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00014859601100184708,
+      "loss": 0.1838,
+      "step": 166000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00014856509099007263,
+      "loss": 0.183,
+      "step": 167000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00014853383772619612,
+      "loss": 0.1826,
+      "step": 168000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00014850225136212017,
+      "loss": 0.1829,
+      "step": 169000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00014847033205136638,
+      "loss": 0.1825,
+      "step": 170000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001484380799490746,
+      "loss": 0.1813,
+      "step": 171000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00014840549521200216,
+      "loss": 0.1817,
+      "step": 172000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001483725779985231,
+      "loss": 0.1813,
+      "step": 173000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00014833932846862748,
+      "loss": 0.1817,
+      "step": 174000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00014830574678392048,
+      "loss": 0.1808,
+      "step": 175000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001482718331076217,
+      "loss": 0.1805,
+      "step": 176000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001482375876045644,
+      "loss": 0.1804,
+      "step": 177000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00014820301044119453,
+      "loss": 0.1798,
+      "step": 178000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00014816810178557016,
+      "loss": 0.1793,
+      "step": 179000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001481328618073604,
+      "loss": 0.1794,
+      "step": 180000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001480972906778448,
+      "loss": 0.1792,
+      "step": 181000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00014806138856991247,
+      "loss": 0.1791,
+      "step": 182000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00014802515565806107,
+      "loss": 0.1782,
+      "step": 183000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001479885921183962,
+      "loss": 0.1787,
+      "step": 184000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00014795169812863038,
+      "loss": 0.1784,
+      "step": 185000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001479144738680823,
+      "loss": 0.1776,
+      "step": 186000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00014787691951767583,
+      "loss": 0.177,
+      "step": 187000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001478390352599392,
+      "loss": 0.1779,
+      "step": 188000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00014780082127900416,
+      "loss": 0.1776,
+      "step": 189000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00014776227776060503,
+      "loss": 0.1768,
+      "step": 190000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00014772340489207773,
+      "loss": 0.1766,
+      "step": 191000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00014768420286235908,
+      "loss": 0.1763,
+      "step": 192000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00014764467186198564,
+      "loss": 0.1766,
+      "step": 193000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001476048120830929,
+      "loss": 0.176,
+      "step": 194000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001475646237194144,
+      "loss": 0.1759,
+      "step": 195000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014752410696628065,
+      "loss": 0.1758,
+      "step": 196000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014748326202061832,
+      "loss": 0.1751,
+      "step": 197000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001474420890809492,
+      "loss": 0.1757,
+      "step": 198000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001474005883473892,
+      "loss": 0.1747,
+      "step": 199000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001473587600216475,
+      "loss": 0.1744,
+      "step": 200000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014731660430702552,
+      "loss": 0.1742,
+      "step": 201000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001472741214084158,
+      "loss": 0.1746,
+      "step": 202000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014723131153230123,
+      "loss": 0.1744,
+      "step": 203000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014718817488675387,
+      "loss": 0.1737,
+      "step": 204000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014714471168143407,
+      "loss": 0.1735,
+      "step": 205000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001471009221275893,
+      "loss": 0.1732,
+      "step": 206000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014705680643805323,
+      "loss": 0.1724,
+      "step": 207000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001470123648272447,
+      "loss": 0.1729,
+      "step": 208000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014696759751116662,
+      "loss": 0.1733,
+      "step": 209000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014692250470740503,
+      "loss": 0.1733,
+      "step": 210000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014687708663512782,
+      "loss": 0.1724,
+      "step": 211000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001468313435150839,
+      "loss": 0.1726,
+      "step": 212000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014678527556960207,
+      "loss": 0.1713,
+      "step": 213000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014673888302258983,
+      "loss": 0.1715,
+      "step": 214000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014669216609953232,
+      "loss": 0.1708,
+      "step": 215000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014664512502749141,
+      "loss": 0.1716,
+      "step": 216000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014659776003510437,
+      "loss": 0.1714,
+      "step": 217000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014655007135258287,
+      "loss": 0.1708,
+      "step": 218000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001465020592117118,
+      "loss": 0.1703,
+      "step": 219000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001464537238458482,
+      "loss": 0.171,
+      "step": 220000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014640506548992013,
+      "loss": 0.1703,
+      "step": 221000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014635608438042546,
+      "loss": 0.1697,
+      "step": 222000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001463067807554308,
+      "loss": 0.1703,
+      "step": 223000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014625715485457025,
+      "loss": 0.1699,
+      "step": 224000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001462072069190444,
+      "loss": 0.1698,
+      "step": 225000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014615693719161898,
+      "loss": 0.1694,
+      "step": 226000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014610634591662365,
+      "loss": 0.1695,
+      "step": 227000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014605543333995113,
+      "loss": 0.1692,
+      "step": 228000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014600419970905561,
+      "loss": 0.1687,
+      "step": 229000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001459526452729518,
+      "loss": 0.1689,
+      "step": 230000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001459007702822136,
+      "loss": 0.1684,
+      "step": 231000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014584857498897304,
+      "loss": 0.1686,
+      "step": 232000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014579605964691876,
+      "loss": 0.1681,
+      "step": 233000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014574322451129507,
+      "loss": 0.168,
+      "step": 234000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014569006983890064,
+      "loss": 0.1679,
+      "step": 235000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014563659588808707,
+      "loss": 0.1678,
+      "step": 236000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001455828029187579,
+      "loss": 0.1672,
+      "step": 237000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014552869119236717,
+      "loss": 0.1671,
+      "step": 238000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014547426097191819,
+      "loss": 0.1674,
+      "step": 239000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014541951252196225,
+      "loss": 0.1666,
+      "step": 240000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014536444610859742,
+      "loss": 0.1655,
+      "step": 241000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014530906199946718,
+      "loss": 0.1669,
+      "step": 242000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014525336046375905,
+      "loss": 0.1663,
+      "step": 243000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014519734177220346,
+      "loss": 0.1663,
+      "step": 244000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014514100619707229,
+      "loss": 0.1662,
+      "step": 245000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014508435401217759,
+      "loss": 0.1661,
+      "step": 246000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014502738549287028,
+      "loss": 0.1654,
+      "step": 247000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014497010091603875,
+      "loss": 0.1657,
+      "step": 248000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014491250056010758,
+      "loss": 0.1653,
+      "step": 249000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001448545847050361,
+      "loss": 0.1651,
+      "step": 250000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014479635363231718,
+      "loss": 0.1652,
+      "step": 251000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001447378076249757,
+      "loss": 0.1652,
+      "step": 252000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014467894696756724,
+      "loss": 0.164,
+      "step": 253000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001446197719461767,
+      "loss": 0.1642,
+      "step": 254000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014456028284841693,
+      "loss": 0.1647,
+      "step": 255000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014450047996342728,
+      "loss": 0.1644,
+      "step": 256000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014444036358187223,
+      "loss": 0.1641,
+      "step": 257000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014437993399594003,
+      "loss": 0.164,
+      "step": 258000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014431919149934114,
+      "loss": 0.1638,
+      "step": 259000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014425813638730688,
+      "loss": 0.1635,
+      "step": 260000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014419676895658807,
+      "loss": 0.1641,
+      "step": 261000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014413508950545352,
+      "loss": 0.1629,
+      "step": 262000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014407309833368846,
+      "loss": 0.1631,
+      "step": 263000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001440107957425933,
+      "loss": 0.163,
+      "step": 264000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014394818203498203,
+      "loss": 0.1628,
+      "step": 265000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014388525751518085,
+      "loss": 0.1627,
+      "step": 266000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001438220224890265,
+      "loss": 0.1627,
+      "step": 267000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000143758477263865,
+      "loss": 0.1618,
+      "step": 268000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00014369462214855,
+      "loss": 0.1623,
+      "step": 269000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00014363045745344137,
+      "loss": 0.1616,
+      "step": 270000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001435659834904036,
+      "loss": 0.1618,
+      "step": 271000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001435012005728044,
+      "loss": 0.1615,
+      "step": 272000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001434361090155131,
+      "loss": 0.1606,
+      "step": 273000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001433707091348991,
+      "loss": 0.1609,
+      "step": 274000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00014330500124883042,
+      "loss": 0.1612,
+      "step": 275000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00014323898567667202,
+      "loss": 0.1607,
+      "step": 276000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00014317266273928444,
+      "loss": 0.1606,
+      "step": 277000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00014310603275902203,
+      "loss": 0.1608,
+      "step": 278000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00014303909605973154,
+      "loss": 0.1607,
+      "step": 279000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00014297185296675046,
+      "loss": 0.1604,
+      "step": 280000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00014290430380690554,
+      "loss": 0.1598,
+      "step": 281000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00014283644890851103,
+      "loss": 0.1598,
+      "step": 282000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00014276828860136726,
+      "loss": 0.1595,
+      "step": 283000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001426998232167589,
+      "loss": 0.16,
+      "step": 284000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00014263105308745343,
+      "loss": 0.1589,
+      "step": 285000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00014256197854769956,
+      "loss": 0.1595,
+      "step": 286000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00014249259993322547,
+      "loss": 0.1588,
+      "step": 287000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001424229175812373,
+      "loss": 0.1593,
+      "step": 288000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00014235293183041746,
+      "loss": 0.1586,
+      "step": 289000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00014228264302092302,
+      "loss": 0.1589,
+      "step": 290000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00014221205149438394,
+      "loss": 0.1589,
+      "step": 291000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00014214115759390164,
+      "loss": 0.1583,
+      "step": 292000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00014206996166404704,
+      "loss": 0.158,
+      "step": 293000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00014199846405085913,
+      "loss": 0.158,
+      "step": 294000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001419266651018432,
+      "loss": 0.1575,
+      "step": 295000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001418545651659691,
+      "loss": 0.158,
+      "step": 296000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00014178216459366958,
+      "loss": 0.1584,
+      "step": 297000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00014170946373683865,
+      "loss": 0.1572,
+      "step": 298000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00014163646294882977,
+      "loss": 0.1576,
+      "step": 299000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00014156316258445421,
+      "loss": 0.157,
+      "step": 300000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001414895629999793,
+      "loss": 0.1569,
+      "step": 301000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00014141566455312663,
+      "loss": 0.1566,
+      "step": 302000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00014134146760307043,
+      "loss": 0.1564,
+      "step": 303000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00014126697251043576,
+      "loss": 0.1574,
+      "step": 304000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001411921796372968,
+      "loss": 0.1559,
+      "step": 305000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001411170893471749,
+      "loss": 0.1561,
+      "step": 306000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001410417020050372,
+      "loss": 0.1561,
+      "step": 307000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001409660179772944,
+      "loss": 0.1563,
+      "step": 308000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001408900376317994,
+      "loss": 0.156,
+      "step": 309000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001408137613378452,
+      "loss": 0.156,
+      "step": 310000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001407371894661632,
+      "loss": 0.1553,
+      "step": 311000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00014066032238892152,
+      "loss": 0.1556,
+      "step": 312000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00014058316047972302,
+      "loss": 0.1552,
+      "step": 313000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00014050570411360355,
+      "loss": 0.1555,
+      "step": 314000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00014042795366703018,
+      "loss": 0.1555,
+      "step": 315000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00014034990951789923,
+      "loss": 0.1557,
+      "step": 316000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00014027157204553464,
+      "loss": 0.1544,
+      "step": 317000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00014019294163068597,
+      "loss": 0.1548,
+      "step": 318000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00014011401865552655,
+      "loss": 0.1542,
+      "step": 319000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00014003480350365167,
+      "loss": 0.154,
+      "step": 320000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001399552965600768,
+      "loss": 0.1548,
+      "step": 321000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00013987549821123545,
+      "loss": 0.1544,
+      "step": 322000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00013979540884497766,
+      "loss": 0.1541,
+      "step": 323000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001397150288505678,
+      "loss": 0.1538,
+      "step": 324000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00013963435861868287,
+      "loss": 0.1541,
+      "step": 325000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001395533985414105,
+      "loss": 0.1542,
+      "step": 326000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00013947214901224706,
+      "loss": 0.1542,
+      "step": 327000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001393906104260958,
+      "loss": 0.1535,
+      "step": 328000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001393087831792649,
+      "loss": 0.1535,
+      "step": 329000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00013922666766946545,
+      "loss": 0.1536,
+      "step": 330000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001391442642958098,
+      "loss": 0.1526,
+      "step": 331000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001390615734588092,
+      "loss": 0.1528,
+      "step": 332000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001389785955603722,
+      "loss": 0.1527,
+      "step": 333000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00013889533100380262,
+      "loss": 0.1521,
+      "step": 334000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00013881178019379745,
+      "loss": 0.153,
+      "step": 335000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000138727943536445,
+      "loss": 0.1526,
+      "step": 336000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00013864382143922294,
+      "loss": 0.1526,
+      "step": 337000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00013855941431099628,
+      "loss": 0.1527,
+      "step": 338000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00013847472256201535,
+      "loss": 0.1527,
+      "step": 339000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00013838974660391388,
+      "loss": 0.1516,
+      "step": 340000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00013830448684970696,
+      "loss": 0.1527,
+      "step": 341000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000138218943713789,
+      "loss": 0.1522,
+      "step": 342000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001381331176119318,
+      "loss": 0.1519,
+      "step": 343000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00013804700896128246,
+      "loss": 0.1517,
+      "step": 344000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00013796061818036138,
+      "loss": 0.1512,
+      "step": 345000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001378739456890602,
+      "loss": 0.1513,
+      "step": 346000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001377869919086398,
+      "loss": 0.151,
+      "step": 347000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001376997572617282,
+      "loss": 0.1515,
+      "step": 348000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00013761224217231859,
+      "loss": 0.1513,
+      "step": 349000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00013752444706576714,
+      "loss": 0.1511,
+      "step": 350000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001374363723687911,
+      "loss": 0.1508,
+      "step": 351000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00013734801850946656,
+      "loss": 0.1505,
+      "step": 352000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001372593859172265,
+      "loss": 0.1508,
+      "step": 353000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00013717047502285855,
+      "loss": 0.1506,
+      "step": 354000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00013708128625850312,
+      "loss": 0.15,
+      "step": 355000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00013699182005765112,
+      "loss": 0.1502,
+      "step": 356000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00013690207685514185,
+      "loss": 0.1503,
+      "step": 357000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000136812057087161,
+      "loss": 0.15,
+      "step": 358000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00013672176119123848,
+      "loss": 0.1499,
+      "step": 359000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001366311896062463,
+      "loss": 0.1499,
+      "step": 360000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00013654034277239634,
+      "loss": 0.1502,
+      "step": 361000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00013644922113123837,
+      "loss": 0.1492,
+      "step": 362000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001363578251256578,
+      "loss": 0.1489,
+      "step": 363000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00013626615519987353,
+      "loss": 0.1495,
+      "step": 364000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00013617421179943585,
+      "loss": 0.149,
+      "step": 365000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00013608199537122425,
+      "loss": 0.1492,
+      "step": 366000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001359895063634452,
+      "loss": 0.1493,
+      "step": 367000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00013589674522563003,
+      "loss": 0.1489,
+      "step": 368000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001358037124086327,
+      "loss": 0.1488,
+      "step": 369000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001357104083646277,
+      "loss": 0.1492,
+      "step": 370000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00013561683354710766,
+      "loss": 0.1487,
+      "step": 371000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00013552298841088144,
+      "loss": 0.1481,
+      "step": 372000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001354288734120716,
+      "loss": 0.1479,
+      "step": 373000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013533448900811243,
+      "loss": 0.1482,
+      "step": 374000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013523983565774753,
+      "loss": 0.1482,
+      "step": 375000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013514491382102782,
+      "loss": 0.148,
+      "step": 376000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013504972395930905,
+      "loss": 0.1481,
+      "step": 377000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013495426653524972,
+      "loss": 0.148,
+      "step": 378000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013485854201280878,
+      "loss": 0.1486,
+      "step": 379000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013476255085724333,
+      "loss": 0.1474,
+      "step": 380000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013466629353510651,
+      "loss": 0.148,
+      "step": 381000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013456977051424506,
+      "loss": 0.1479,
+      "step": 382000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013447298226379708,
+      "loss": 0.1477,
+      "step": 383000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013437592925418985,
+      "loss": 0.1467,
+      "step": 384000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001342786119571375,
+      "loss": 0.1472,
+      "step": 385000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001341810308456386,
+      "loss": 0.1468,
+      "step": 386000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013408318639397405,
+      "loss": 0.1466,
+      "step": 387000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013398507907770463,
+      "loss": 0.147,
+      "step": 388000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013388670937366875,
+      "loss": 0.1462,
+      "step": 389000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013378807775998012,
+      "loss": 0.1464,
+      "step": 390000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013368918471602545,
+      "loss": 0.1468,
+      "step": 391000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013359003072246212,
+      "loss": 0.1471,
+      "step": 392000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00013349061626121578,
+      "loss": 0.147,
+      "step": 393000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000133390941815478,
+      "loss": 0.1463,
+      "step": 394000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001332910078697041,
+      "loss": 0.1464,
+      "step": 395000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001331908149096106,
+      "loss": 0.1464,
+      "step": 396000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00013309036342217288,
+      "loss": 0.1463,
+      "step": 397000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00013298965389562295,
+      "loss": 0.146,
+      "step": 398000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013288868681944692,
+      "loss": 0.1461,
+      "step": 399000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013278746268438274,
+      "loss": 0.1457,
+      "step": 400000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001326859819824177,
+      "loss": 0.1461,
+      "step": 401000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013258424520678618,
+      "loss": 0.1456,
+      "step": 402000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001324822528519671,
+      "loss": 0.1456,
+      "step": 403000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013238000541368163,
+      "loss": 0.1451,
+      "step": 404000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013227750338889077,
+      "loss": 0.1452,
+      "step": 405000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001321747472757928,
+      "loss": 0.1453,
+      "step": 406000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001320717375738211,
+      "loss": 0.1452,
+      "step": 407000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001319684747836415,
+      "loss": 0.1453,
+      "step": 408000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013186495940714993,
+      "loss": 0.145,
+      "step": 409000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013176119194747003,
+      "loss": 0.1445,
+      "step": 410000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013165717290895067,
+      "loss": 0.1455,
+      "step": 411000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013155290279716342,
+      "loss": 0.1451,
+      "step": 412000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001314483821189002,
+      "loss": 0.1446,
+      "step": 413000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001313436113821708,
+      "loss": 0.1446,
+      "step": 414000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013123859109620033,
+      "loss": 0.144,
+      "step": 415000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00013113332177142682,
+      "loss": 0.1441,
+      "step": 416000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001310278039194988,
+      "loss": 0.1444,
+      "step": 417000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013092203805327264,
+      "loss": 0.1443,
+      "step": 418000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013081602468681018,
+      "loss": 0.1443,
+      "step": 419000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013070976433537623,
+      "loss": 0.1449,
+      "step": 420000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013060325751543597,
+      "loss": 0.1441,
+      "step": 421000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013049650474465264,
+      "loss": 0.1438,
+      "step": 422000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00013038950654188476,
+      "loss": 0.1432,
+      "step": 423000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001302822634271838,
+      "loss": 0.144,
+      "step": 424000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001301747759217916,
+      "loss": 0.1436,
+      "step": 425000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001300670445481378,
+      "loss": 0.1438,
+      "step": 426000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00012995906982983733,
+      "loss": 0.1432,
+      "step": 427000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00012985085229168797,
+      "loss": 0.1437,
+      "step": 428000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00012974239245966754,
+      "loss": 0.1436,
+      "step": 429000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001296336908609316,
+      "loss": 0.143,
+      "step": 430000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012952474802381075,
+      "loss": 0.143,
+      "step": 431000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012941556447780813,
+      "loss": 0.1428,
+      "step": 432000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012930614075359675,
+      "loss": 0.1424,
+      "step": 433000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00012919647738301706,
+      "loss": 0.1431,
+      "step": 434000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001290865748990742,
+      "loss": 0.1432,
+      "step": 435000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012897643383593557,
+      "loss": 0.1428,
+      "step": 436000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012886605472892812,
+      "loss": 0.1424,
+      "step": 437000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012875543811453576,
+      "loss": 0.1426,
+      "step": 438000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012864458453039677,
+      "loss": 0.1432,
+      "step": 439000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012853349451530128,
+      "loss": 0.1421,
+      "step": 440000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00012842216860918846,
+      "loss": 0.1423,
+      "step": 441000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00012831060735314404,
+      "loss": 0.1422,
+      "step": 442000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001281988112893976,
+      "loss": 0.1418,
+      "step": 443000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001280867809613201,
+      "loss": 0.1425,
+      "step": 444000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000127974516913421,
+      "loss": 0.142,
+      "step": 445000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00012786201969134568,
+      "loss": 0.1417,
+      "step": 446000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00012774928984187297,
+      "loss": 0.1422,
+      "step": 447000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00012763632791291227,
+      "loss": 0.1414,
+      "step": 448000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00012752313445350098,
+      "loss": 0.1413,
+      "step": 449000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001274097100138019,
+      "loss": 0.1416,
+      "step": 450000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00012729605514510031,
+      "loss": 0.1417,
+      "step": 451000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001271821703998017,
+      "loss": 0.141,
+      "step": 452000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00012706805633142863,
+      "loss": 0.1414,
+      "step": 453000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00012695371349461836,
+      "loss": 0.1417,
+      "step": 454000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00012683914244512008,
+      "loss": 0.1413,
+      "step": 455000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00012672434373979207,
+      "loss": 0.1407,
+      "step": 456000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00012660931793659921,
+      "loss": 0.1407,
+      "step": 457000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00012649406559461012,
+      "loss": 0.1403,
+      "step": 458000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00012637858727399448,
+      "loss": 0.141,
+      "step": 459000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00012626288353602028,
+      "loss": 0.1401,
+      "step": 460000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001261469549430512,
+      "loss": 0.1406,
+      "step": 461000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00012603080205854372,
+      "loss": 0.1406,
+      "step": 462000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001259144254470445,
+      "loss": 0.1399,
+      "step": 463000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00012579782567418763,
+      "loss": 0.1401,
+      "step": 464000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001256810033066918,
+      "loss": 0.1403,
+      "step": 465000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001255639589123576,
+      "loss": 0.1398,
+      "step": 466000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00012544669306006475,
+      "loss": 0.1396,
+      "step": 467000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001253292063197693,
+      "loss": 0.1396,
+      "step": 468000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00012521149926250107,
+      "loss": 0.1395,
+      "step": 469000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001250935724603605,
+      "loss": 0.1399,
+      "step": 470000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00012497542648651615,
+      "loss": 0.1393,
+      "step": 471000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00012485706191520186,
+      "loss": 0.1394,
+      "step": 472000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012473847932171394,
+      "loss": 0.14,
+      "step": 473000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012461967928240828,
+      "loss": 0.1396,
+      "step": 474000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012450066237469778,
+      "loss": 0.1394,
+      "step": 475000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012438142917704925,
+      "loss": 0.1396,
+      "step": 476000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001242619802689809,
+      "loss": 0.1393,
+      "step": 477000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012414231623105923,
+      "loss": 0.1398,
+      "step": 478000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012402243764489648,
+      "loss": 0.1389,
+      "step": 479000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001239023450931476,
+      "loss": 0.1388,
+      "step": 480000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012378203915950747,
+      "loss": 0.1391,
+      "step": 481000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001236615204287082,
+      "loss": 0.1388,
+      "step": 482000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012354078948651604,
+      "loss": 0.1388,
+      "step": 483000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012341984691972876,
+      "loss": 0.1389,
+      "step": 484000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012329869331617266,
+      "loss": 0.1394,
+      "step": 485000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012317732926469976,
+      "loss": 0.1383,
+      "step": 486000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012305575535518498,
+      "loss": 0.1383,
+      "step": 487000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012293397217852313,
+      "loss": 0.1385,
+      "step": 488000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001228119803266263,
+      "loss": 0.1385,
+      "step": 489000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012268978039242062,
+      "loss": 0.1382,
+      "step": 490000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00012256737296984378,
+      "loss": 0.1383,
+      "step": 491000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012244475865384177,
+      "loss": 0.1383,
+      "step": 492000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012232193804036625,
+      "loss": 0.1376,
+      "step": 493000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012219891172637155,
+      "loss": 0.1383,
+      "step": 494000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012207568030981174,
+      "loss": 0.1377,
+      "step": 495000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012195224438963788,
+      "loss": 0.1378,
+      "step": 496000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00012182860456579486,
+      "loss": 0.138,
+      "step": 497000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001217047614392187,
+      "loss": 0.1375,
+      "step": 498000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001215807156118335,
+      "loss": 0.1378,
+      "step": 499000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001214564676865486,
+      "loss": 0.1375,
+      "step": 500000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012133201826725558,
+      "loss": 0.1371,
+      "step": 501000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001212073679588254,
+      "loss": 0.138,
+      "step": 502000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00012108251736710539,
+      "loss": 0.1371,
+      "step": 503000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012095746709891632,
+      "loss": 0.1375,
+      "step": 504000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012083221776204947,
+      "loss": 0.1365,
+      "step": 505000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012070676996526366,
+      "loss": 0.137,
+      "step": 506000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001205811243182823,
+      "loss": 0.1374,
+      "step": 507000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012045528143179054,
+      "loss": 0.1363,
+      "step": 508000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00012032924191743198,
+      "loss": 0.1371,
+      "step": 509000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012020300638780604,
+      "loss": 0.1369,
+      "step": 510000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00012007657545646484,
+      "loss": 0.1374,
+      "step": 511000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00011994994973791022,
+      "loss": 0.1366,
+      "step": 512000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00011982312984759068,
+      "loss": 0.1364,
+      "step": 513000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001196961164018986,
+      "loss": 0.1368,
+      "step": 514000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00011956891001816701,
+      "loss": 0.1361,
+      "step": 515000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011944151131466675,
+      "loss": 0.1366,
+      "step": 516000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011931392091060336,
+      "loss": 0.1363,
+      "step": 517000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011918613942611415,
+      "loss": 0.1359,
+      "step": 518000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011905816748226513,
+      "loss": 0.1363,
+      "step": 519000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011893000570104802,
+      "loss": 0.136,
+      "step": 520000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00011880165470537728,
+      "loss": 0.1357,
+      "step": 521000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011867311511908693,
+      "loss": 0.137,
+      "step": 522000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011854438756692763,
+      "loss": 0.1359,
+      "step": 523000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011841547267456366,
+      "loss": 0.1358,
+      "step": 524000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011828637106856989,
+      "loss": 0.136,
+      "step": 525000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011815708337642854,
+      "loss": 0.1365,
+      "step": 526000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011802761022652641,
+      "loss": 0.1358,
+      "step": 527000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00011789795224815164,
+      "loss": 0.1352,
+      "step": 528000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011776811007149076,
+      "loss": 0.1353,
+      "step": 529000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011763808432762546,
+      "loss": 0.1359,
+      "step": 530000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011750787564852973,
+      "loss": 0.1353,
+      "step": 531000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011737748466706673,
+      "loss": 0.1353,
+      "step": 532000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011724691201698554,
+      "loss": 0.1352,
+      "step": 533000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00011711615833291833,
+      "loss": 0.1351,
+      "step": 534000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011698522425037712,
+      "loss": 0.1349,
+      "step": 535000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011685411040575073,
+      "loss": 0.135,
+      "step": 536000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011672281743630175,
+      "loss": 0.1345,
+      "step": 537000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011659134598016327,
+      "loss": 0.1353,
+      "step": 538000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011645969667633602,
+      "loss": 0.1351,
+      "step": 539000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00011632787016468506,
+      "loss": 0.1347,
+      "step": 540000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011619586708593674,
+      "loss": 0.1348,
+      "step": 541000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011606368808167565,
+      "loss": 0.1347,
+      "step": 542000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011593133379434138,
+      "loss": 0.1347,
+      "step": 543000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011579880486722555,
+      "loss": 0.1336,
+      "step": 544000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011566610194446844,
+      "loss": 0.1339,
+      "step": 545000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00011553322567105619,
+      "loss": 0.1336,
+      "step": 546000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011540017669281736,
+      "loss": 0.1342,
+      "step": 547000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011526695565642001,
+      "loss": 0.1335,
+      "step": 548000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011513356320936841,
+      "loss": 0.134,
+      "step": 549000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000115,
+      "loss": 0.1338,
+      "step": 550000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011486626667748209,
+      "loss": 0.1336,
+      "step": 551000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00011473236389180894,
+      "loss": 0.1331,
+      "step": 552000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011459829229379834,
+      "loss": 0.1337,
+      "step": 553000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011446405253508869,
+      "loss": 0.1336,
+      "step": 554000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011432964526813558,
+      "loss": 0.1344,
+      "step": 555000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011419507114620884,
+      "loss": 0.1338,
+      "step": 556000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001140603308233893,
+      "loss": 0.1333,
+      "step": 557000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011392542495456556,
+      "loss": 0.1337,
+      "step": 558000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00011379035419543074,
+      "loss": 0.1333,
+      "step": 559000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011365511920247955,
+      "loss": 0.1326,
+      "step": 560000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011351972063300484,
+      "loss": 0.1333,
+      "step": 561000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011338415914509456,
+      "loss": 0.1335,
+      "step": 562000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011324843539762842,
+      "loss": 0.1327,
+      "step": 563000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011311255005027487,
+      "loss": 0.1329,
+      "step": 564000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00011297650376348771,
+      "loss": 0.1333,
+      "step": 565000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011284029719850299,
+      "loss": 0.1331,
+      "step": 566000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011270393101733585,
+      "loss": 0.1326,
+      "step": 567000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011256740588277714,
+      "loss": 0.1332,
+      "step": 568000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011243072245839033,
+      "loss": 0.1325,
+      "step": 569000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011229388140850814,
+      "loss": 0.1326,
+      "step": 570000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00011215688339822954,
+      "loss": 0.1319,
+      "step": 571000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011201972909341632,
+      "loss": 0.1328,
+      "step": 572000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011188241916068993,
+      "loss": 0.1331,
+      "step": 573000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011174495426742821,
+      "loss": 0.1326,
+      "step": 574000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011160733508176224,
+      "loss": 0.1328,
+      "step": 575000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011146956227257293,
+      "loss": 0.1324,
+      "step": 576000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011133163650948788,
+      "loss": 0.1323,
+      "step": 577000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011119355846287812,
+      "loss": 0.1327,
+      "step": 578000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011105532880385487,
+      "loss": 0.1323,
+      "step": 579000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011091694820426618,
+      "loss": 0.1316,
+      "step": 580000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011077841733669374,
+      "loss": 0.1318,
+      "step": 581000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011063973687444962,
+      "loss": 0.1322,
+      "step": 582000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011050090749157299,
+      "loss": 0.1322,
+      "step": 583000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011036192986282679,
+      "loss": 0.1319,
+      "step": 584000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011022280466369448,
+      "loss": 0.1317,
+      "step": 585000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011008353257037687,
+      "loss": 0.1318,
+      "step": 586000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010994411425978862,
+      "loss": 0.1313,
+      "step": 587000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010980455040955506,
+      "loss": 0.1314,
+      "step": 588000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010966484169800902,
+      "loss": 0.1311,
+      "step": 589000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010952498880418724,
+      "loss": 0.1313,
+      "step": 590000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010938499240782739,
+      "loss": 0.1307,
+      "step": 591000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010924485318936452,
+      "loss": 0.1317,
+      "step": 592000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001091045718299279,
+      "loss": 0.1312,
+      "step": 593000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010896414901133761,
+      "loss": 0.1313,
+      "step": 594000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010882358541610135,
+      "loss": 0.1309,
+      "step": 595000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010868288172741095,
+      "loss": 0.1316,
+      "step": 596000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010854203862913927,
+      "loss": 0.131,
+      "step": 597000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010840105680583659,
+      "loss": 0.131,
+      "step": 598000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001082599369427276,
+      "loss": 0.1303,
+      "step": 599000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010811867972570786,
+      "loss": 0.1312,
+      "step": 600000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001079772858413405,
+      "loss": 0.1309,
+      "step": 601000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001078357559768529,
+      "loss": 0.1308,
+      "step": 602000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010769409082013337,
+      "loss": 0.1307,
+      "step": 603000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010755229105972785,
+      "loss": 0.1306,
+      "step": 604000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001074103573848364,
+      "loss": 0.1304,
+      "step": 605000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010726829048531,
+      "loss": 0.1306,
+      "step": 606000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001071260910516472,
+      "loss": 0.1302,
+      "step": 607000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010698375977499065,
+      "loss": 0.1309,
+      "step": 608000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001068412973471238,
+      "loss": 0.1297,
+      "step": 609000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010669870446046762,
+      "loss": 0.1297,
+      "step": 610000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010655598180807708,
+      "loss": 0.1303,
+      "step": 611000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001064131300836379,
+      "loss": 0.1302,
+      "step": 612000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010627014998146313,
+      "loss": 0.1304,
+      "step": 613000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010612704219648977,
+      "loss": 0.1302,
+      "step": 614000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010598380742427543,
+      "loss": 0.1299,
+      "step": 615000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010584044636099496,
+      "loss": 0.1297,
+      "step": 616000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001056969597034369,
+      "loss": 0.1302,
+      "step": 617000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001055533481490004,
+      "loss": 0.13,
+      "step": 618000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010540961239569151,
+      "loss": 0.1296,
+      "step": 619000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010526575314212003,
+      "loss": 0.1288,
+      "step": 620000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010512177108749594,
+      "loss": 0.1287,
+      "step": 621000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010497766693162617,
+      "loss": 0.1297,
+      "step": 622000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000104833441374911,
+      "loss": 0.129,
+      "step": 623000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010468909511834088,
+      "loss": 0.1291,
+      "step": 624000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010454462886349281,
+      "loss": 0.1294,
+      "step": 625000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010440004331252707,
+      "loss": 0.129,
+      "step": 626000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010425533916818376,
+      "loss": 0.1293,
+      "step": 627000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010411051713377937,
+      "loss": 0.1287,
+      "step": 628000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010396557791320346,
+      "loss": 0.1287,
+      "step": 629000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000103820522210915,
+      "loss": 0.1289,
+      "step": 630000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001036753507319393,
+      "loss": 0.1288,
+      "step": 631000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010353006418186419,
+      "loss": 0.1287,
+      "step": 632000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010338466326683697,
+      "loss": 0.1287,
+      "step": 633000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010323914869356067,
+      "loss": 0.1291,
+      "step": 634000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010309352116929079,
+      "loss": 0.1285,
+      "step": 635000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010294778140183182,
+      "loss": 0.1288,
+      "step": 636000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001028019300995338,
+      "loss": 0.1286,
+      "step": 637000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010265596797128882,
+      "loss": 0.1284,
+      "step": 638000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010250989572652766,
+      "loss": 0.1283,
+      "step": 639000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010236371407521637,
+      "loss": 0.1284,
+      "step": 640000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010221742372785262,
+      "loss": 0.1278,
+      "step": 641000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010207102539546251,
+      "loss": 0.1279,
+      "step": 642000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001019245197895969,
+      "loss": 0.1279,
+      "step": 643000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001017779076223281,
+      "loss": 0.1279,
+      "step": 644000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010163118960624632,
+      "loss": 0.1284,
+      "step": 645000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010148436645445622,
+      "loss": 0.1275,
+      "step": 646000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010133743888057353,
+      "loss": 0.128,
+      "step": 647000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010119040759872142,
+      "loss": 0.1274,
+      "step": 648000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010104327332352717,
+      "loss": 0.1275,
+      "step": 649000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010089603677011861,
+      "loss": 0.1275,
+      "step": 650000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010074869865412074,
+      "loss": 0.1272,
+      "step": 651000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00010060125969165216,
+      "loss": 0.1278,
+      "step": 652000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001004537205993216,
+      "loss": 0.1271,
+      "step": 653000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001003060820942245,
+      "loss": 0.1277,
+      "step": 654000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00010015834489393943,
+      "loss": 0.128,
+      "step": 655000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00010001050971652473,
+      "loss": 0.1269,
+      "step": 656000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.986257728051483e-05,
+      "loss": 0.1271,
+      "step": 657000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.971454830491704e-05,
+      "loss": 0.1271,
+      "step": 658000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.956642350920775e-05,
+      "loss": 0.1273,
+      "step": 659000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.94182036133291e-05,
+      "loss": 0.1271,
+      "step": 660000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.926988933768553e-05,
+      "loss": 0.1271,
+      "step": 661000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.912148140314006e-05,
+      "loss": 0.127,
+      "step": 662000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.89729805310111e-05,
+      "loss": 0.1265,
+      "step": 663000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.882438744306858e-05,
+      "loss": 0.1262,
+      "step": 664000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.867570286153083e-05,
+      "loss": 0.1263,
+      "step": 665000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.852692750906071e-05,
+      "loss": 0.1268,
+      "step": 666000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.837806210876233e-05,
+      "loss": 0.1265,
+      "step": 667000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.822910738417753e-05,
+      "loss": 0.1262,
+      "step": 668000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.808006405928215e-05,
+      "loss": 0.1271,
+      "step": 669000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.79309328584828e-05,
+      "loss": 0.1265,
+      "step": 670000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.778171450661311e-05,
+      "loss": 0.1269,
+      "step": 671000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.763240972893037e-05,
+      "loss": 0.1265,
+      "step": 672000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.748301925111184e-05,
+      "loss": 0.126,
+      "step": 673000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.733354379925142e-05,
+      "loss": 0.1263,
+      "step": 674000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.718398409985593e-05,
+      "loss": 0.1263,
+      "step": 675000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.703434087984172e-05,
+      "loss": 0.1264,
+      "step": 676000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.688461486653107e-05,
+      "loss": 0.1263,
+      "step": 677000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.673480678764858e-05,
+      "loss": 0.1266,
+      "step": 678000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.658491737131792e-05,
+      "loss": 0.1261,
+      "step": 679000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.643494734605786e-05,
+      "loss": 0.1255,
+      "step": 680000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.628489744077911e-05,
+      "loss": 0.1256,
+      "step": 681000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.613476838478058e-05,
+      "loss": 0.1255,
+      "step": 682000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.59845609077459e-05,
+      "loss": 0.1257,
+      "step": 683000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.583427573973982e-05,
+      "loss": 0.1261,
+      "step": 684000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.568391361120473e-05,
+      "loss": 0.1262,
+      "step": 685000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.553347525295703e-05,
+      "loss": 0.1255,
+      "step": 686000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.538296139618371e-05,
+      "loss": 0.125,
+      "step": 687000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.523237277243868e-05,
+      "loss": 0.1252,
+      "step": 688000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.508171011363912e-05,
+      "loss": 0.1251,
+      "step": 689000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.493097415206228e-05,
+      "loss": 0.1257,
+      "step": 690000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.478016562034148e-05,
+      "loss": 0.1249,
+      "step": 691000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.462928525146289e-05,
+      "loss": 0.1256,
+      "step": 692000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.447833377876176e-05,
+      "loss": 0.125,
+      "step": 693000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.432731193591898e-05,
+      "loss": 0.1252,
+      "step": 694000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.417622045695745e-05,
+      "loss": 0.1251,
+      "step": 695000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.402506007623848e-05,
+      "loss": 0.1248,
+      "step": 696000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.387383152845835e-05,
+      "loss": 0.1248,
+      "step": 697000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.372253554864461e-05,
+      "loss": 0.1249,
+      "step": 698000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.357117287215258e-05,
+      "loss": 0.1248,
+      "step": 699000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.341974423466169e-05,
+      "loss": 0.1252,
+      "step": 700000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.326825037217204e-05,
+      "loss": 0.1249,
+      "step": 701000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.311669202100073e-05,
+      "loss": 0.1247,
+      "step": 702000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.296506991777826e-05,
+      "loss": 0.125,
+      "step": 703000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.2813384799445e-05,
+      "loss": 0.1241,
+      "step": 704000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.26616374032477e-05,
+      "loss": 0.1244,
+      "step": 705000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.250982846673562e-05,
+      "loss": 0.1242,
+      "step": 706000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.235795872775724e-05,
+      "loss": 0.1242,
+      "step": 707000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.220602892445661e-05,
+      "loss": 0.1243,
+      "step": 708000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.205403979526962e-05,
+      "loss": 0.1267,
+      "step": 709000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.19019920789205e-05,
+      "loss": 0.1238,
+      "step": 710000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.174988651441833e-05,
+      "loss": 0.1246,
+      "step": 711000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.159772384105328e-05,
+      "loss": 0.1243,
+      "step": 712000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.144550479839314e-05,
+      "loss": 0.124,
+      "step": 713000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.129323012627956e-05,
+      "loss": 0.1238,
+      "step": 714000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.114090056482476e-05,
+      "loss": 0.1237,
+      "step": 715000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.098851685440756e-05,
+      "loss": 0.1232,
+      "step": 716000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.08360797356701e-05,
+      "loss": 0.1239,
+      "step": 717000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.068358994951399e-05,
+      "loss": 0.1242,
+      "step": 718000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.053104823709692e-05,
+      "loss": 0.1235,
+      "step": 719000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.037845533982892e-05,
+      "loss": 0.1242,
+      "step": 720000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.022581199936881e-05,
+      "loss": 0.1244,
+      "step": 721000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.007311895762054e-05,
+      "loss": 0.1235,
+      "step": 722000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.992037695672967e-05,
+      "loss": 0.1239,
+      "step": 723000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.976758673907974e-05,
+      "loss": 0.1233,
+      "step": 724000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.961474904728856e-05,
+      "loss": 0.1229,
+      "step": 725000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.946186462420478e-05,
+      "loss": 0.1237,
+      "step": 726000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.93089342129041e-05,
+      "loss": 0.1228,
+      "step": 727000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.91559585566858e-05,
+      "loss": 0.123,
+      "step": 728000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.900293839906903e-05,
+      "loss": 0.1232,
+      "step": 729000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.884987448378922e-05,
+      "loss": 0.1225,
+      "step": 730000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.869676755479458e-05,
+      "loss": 0.1231,
+      "step": 731000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.85436183562422e-05,
+      "loss": 0.1223,
+      "step": 732000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.839042763249482e-05,
+      "loss": 0.1227,
+      "step": 733000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.823719612811688e-05,
+      "loss": 0.1227,
+      "step": 734000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.808392458787103e-05,
+      "loss": 0.1231,
+      "step": 735000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.793061375671455e-05,
+      "loss": 0.1226,
+      "step": 736000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.777726437979571e-05,
+      "loss": 0.1232,
+      "step": 737000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.762387720245008e-05,
+      "loss": 0.1231,
+      "step": 738000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.747045297019693e-05,
+      "loss": 0.1224,
+      "step": 739000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.731699242873574e-05,
+      "loss": 0.122,
+      "step": 740000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.716349632394235e-05,
+      "loss": 0.1229,
+      "step": 741000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.700996540186551e-05,
+      "loss": 0.1224,
+      "step": 742000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.685640040872317e-05,
+      "loss": 0.1225,
+      "step": 743000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.67028020908989e-05,
+      "loss": 0.1227,
+      "step": 744000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.654917119493824e-05,
+      "loss": 0.1224,
+      "step": 745000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.639550846754506e-05,
+      "loss": 0.1231,
+      "step": 746000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.624181465557794e-05,
+      "loss": 0.122,
+      "step": 747000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.608809050604653e-05,
+      "loss": 0.1221,
+      "step": 748000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.5934336766108e-05,
+      "loss": 0.122,
+      "step": 749000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.578055418306327e-05,
+      "loss": 0.1225,
+      "step": 750000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.562674350435344e-05,
+      "loss": 0.1218,
+      "step": 751000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.547290547755621e-05,
+      "loss": 0.1218,
+      "step": 752000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.531904085038221e-05,
+      "loss": 0.1215,
+      "step": 753000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.51651503706713e-05,
+      "loss": 0.1215,
+      "step": 754000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.501123478638909e-05,
+      "loss": 0.1221,
+      "step": 755000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.485729484562307e-05,
+      "loss": 0.122,
+      "step": 756000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.470333129657925e-05,
+      "loss": 0.1212,
+      "step": 757000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.454934488757825e-05,
+      "loss": 0.1215,
+      "step": 758000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.439533636705194e-05,
+      "loss": 0.1219,
+      "step": 759000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.424130648353954e-05,
+      "loss": 0.1215,
+      "step": 760000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.408725598568418e-05,
+      "loss": 0.1208,
+      "step": 761000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.393318562222916e-05,
+      "loss": 0.1216,
+      "step": 762000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.377909614201437e-05,
+      "loss": 0.1213,
+      "step": 763000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.362498829397247e-05,
+      "loss": 0.1215,
+      "step": 764000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.347086282712556e-05,
+      "loss": 0.1212,
+      "step": 765000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.331672049058131e-05,
+      "loss": 0.1211,
+      "step": 766000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.316256203352942e-05,
+      "loss": 0.1215,
+      "step": 767000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.300838820523784e-05,
+      "loss": 0.1211,
+      "step": 768000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.285419975504935e-05,
+      "loss": 0.1214,
+      "step": 769000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.269999743237772e-05,
+      "loss": 0.1215,
+      "step": 770000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.254578198670421e-05,
+      "loss": 0.1211,
+      "step": 771000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.239155416757376e-05,
+      "loss": 0.1208,
+      "step": 772000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.223731472459157e-05,
+      "loss": 0.1209,
+      "step": 773000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.208306440741926e-05,
+      "loss": 0.1205,
+      "step": 774000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.192880396577134e-05,
+      "loss": 0.1207,
+      "step": 775000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.177453414941149e-05,
+      "loss": 0.1203,
+      "step": 776000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.162025570814896e-05,
+      "loss": 0.1204,
+      "step": 777000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.146596939183499e-05,
+      "loss": 0.1204,
+      "step": 778000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.131167595035897e-05,
+      "loss": 0.12,
+      "step": 779000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.115737613364511e-05,
+      "loss": 0.1205,
+      "step": 780000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.100307069164836e-05,
+      "loss": 0.1198,
+      "step": 781000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.08487603743512e-05,
+      "loss": 0.1204,
+      "step": 782000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.069444593175975e-05,
+      "loss": 0.1201,
+      "step": 783000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.054012811390017e-05,
+      "loss": 0.1203,
+      "step": 784000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.038580767081503e-05,
+      "loss": 0.1198,
+      "step": 785000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.023148535255965e-05,
+      "loss": 0.1202,
+      "step": 786000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.007716190919852e-05,
+      "loss": 0.12,
+      "step": 787000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.99228380908015e-05,
+      "loss": 0.1201,
+      "step": 788000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.976851464744033e-05,
+      "loss": 0.1197,
+      "step": 789000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.961419232918497e-05,
+      "loss": 0.1202,
+      "step": 790000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.945987188609984e-05,
+      "loss": 0.1197,
+      "step": 791000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.930555406824026e-05,
+      "loss": 0.1198,
+      "step": 792000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.915123962564879e-05,
+      "loss": 0.1197,
+      "step": 793000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.899692930835165e-05,
+      "loss": 0.1199,
+      "step": 794000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.884262386635489e-05,
+      "loss": 0.1202,
+      "step": 795000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.8688324049641e-05,
+      "loss": 0.1192,
+      "step": 796000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.853403060816501e-05,
+      "loss": 0.1194,
+      "step": 797000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.837974429185103e-05,
+      "loss": 0.1194,
+      "step": 798000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.822546585058854e-05,
+      "loss": 0.1194,
+      "step": 799000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.807119603422866e-05,
+      "loss": 0.1193,
+      "step": 800000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.791693559258072e-05,
+      "loss": 0.1193,
+      "step": 801000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.77626852754084e-05,
+      "loss": 0.1188,
+      "step": 802000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.760844583242621e-05,
+      "loss": 0.1189,
+      "step": 803000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.74542180132958e-05,
+      "loss": 0.1198,
+      "step": 804000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.730000256762226e-05,
+      "loss": 0.1191,
+      "step": 805000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.714580024495065e-05,
+      "loss": 0.1194,
+      "step": 806000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.699161179476217e-05,
+      "loss": 0.1187,
+      "step": 807000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.683743796647058e-05,
+      "loss": 0.1184,
+      "step": 808000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.668327950941869e-05,
+      "loss": 0.1195,
+      "step": 809000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.652913717287443e-05,
+      "loss": 0.119,
+      "step": 810000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.637501170602755e-05,
+      "loss": 0.1185,
+      "step": 811000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.622090385798566e-05,
+      "loss": 0.119,
+      "step": 812000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.606681437777081e-05,
+      "loss": 0.1183,
+      "step": 813000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.59127440143158e-05,
+      "loss": 0.1183,
+      "step": 814000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.575869351646046e-05,
+      "loss": 0.1185,
+      "step": 815000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.560466363294806e-05,
+      "loss": 0.1184,
+      "step": 816000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.545065511242175e-05,
+      "loss": 0.1189,
+      "step": 817000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.529666870342077e-05,
+      "loss": 0.1185,
+      "step": 818000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.514270515437691e-05,
+      "loss": 0.1186,
+      "step": 819000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.498876521361092e-05,
+      "loss": 0.1179,
+      "step": 820000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.483484962932867e-05,
+      "loss": 0.1185,
+      "step": 821000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.468095914961777e-05,
+      "loss": 0.1187,
+      "step": 822000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.452709452244376e-05,
+      "loss": 0.1182,
+      "step": 823000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.437325649564655e-05,
+      "loss": 0.1179,
+      "step": 824000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.421944581693674e-05,
+      "loss": 0.1181,
+      "step": 825000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.406566323389196e-05,
+      "loss": 0.1178,
+      "step": 826000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.391190949395345e-05,
+      "loss": 0.1179,
+      "step": 827000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.375818534442207e-05,
+      "loss": 0.118,
+      "step": 828000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.360449153245494e-05,
+      "loss": 0.1172,
+      "step": 829000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.345082880506175e-05,
+      "loss": 0.1179,
+      "step": 830000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.329719790910108e-05,
+      "loss": 0.1176,
+      "step": 831000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.314359959127683e-05,
+      "loss": 0.1172,
+      "step": 832000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.29900345981345e-05,
+      "loss": 0.1177,
+      "step": 833000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.283650367605764e-05,
+      "loss": 0.1175,
+      "step": 834000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.268300757126426e-05,
+      "loss": 0.1176,
+      "step": 835000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.252954702980306e-05,
+      "loss": 0.1174,
+      "step": 836000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.23761227975499e-05,
+      "loss": 0.1173,
+      "step": 837000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.222273562020429e-05,
+      "loss": 0.1179,
+      "step": 838000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.206938624328543e-05,
+      "loss": 0.1172,
+      "step": 839000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.191607541212897e-05,
+      "loss": 0.1169,
+      "step": 840000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.176280387188314e-05,
+      "loss": 0.117,
+      "step": 841000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.160957236750516e-05,
+      "loss": 0.1166,
+      "step": 842000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.145638164375779e-05,
+      "loss": 0.1169,
+      "step": 843000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.130323244520544e-05,
+      "loss": 0.1169,
+      "step": 844000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.115012551621076e-05,
+      "loss": 0.1171,
+      "step": 845000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.099706160093098e-05,
+      "loss": 0.1174,
+      "step": 846000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.08440414433142e-05,
+      "loss": 0.1173,
+      "step": 847000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.06910657870959e-05,
+      "loss": 0.117,
+      "step": 848000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.053813537579523e-05,
+      "loss": 0.1163,
+      "step": 849000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.038525095271142e-05,
+      "loss": 0.117,
+      "step": 850000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.023241326092026e-05,
+      "loss": 0.117,
+      "step": 851000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.00796230432703e-05,
+      "loss": 0.1167,
+      "step": 852000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.992688104237945e-05,
+      "loss": 0.1162,
+      "step": 853000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.977418800063119e-05,
+      "loss": 0.1167,
+      "step": 854000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.962154466017105e-05,
+      "loss": 0.116,
+      "step": 855000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.946895176290307e-05,
+      "loss": 0.1165,
+      "step": 856000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.931641005048601e-05,
+      "loss": 0.1162,
+      "step": 857000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.91639202643299e-05,
+      "loss": 0.116,
+      "step": 858000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.901148314559244e-05,
+      "loss": 0.1159,
+      "step": 859000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.885909943517524e-05,
+      "loss": 0.1163,
+      "step": 860000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.870676987372044e-05,
+      "loss": 0.1162,
+      "step": 861000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.855449520160687e-05,
+      "loss": 0.1161,
+      "step": 862000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.84022761589467e-05,
+      "loss": 0.1159,
+      "step": 863000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.825011348558167e-05,
+      "loss": 0.1157,
+      "step": 864000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.80980079210795e-05,
+      "loss": 0.1162,
+      "step": 865000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.794596020473038e-05,
+      "loss": 0.1161,
+      "step": 866000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.779397107554339e-05,
+      "loss": 0.116,
+      "step": 867000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.764204127224273e-05,
+      "loss": 0.1162,
+      "step": 868000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.749017153326438e-05,
+      "loss": 0.1154,
+      "step": 869000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.733836259675233e-05,
+      "loss": 0.1156,
+      "step": 870000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.718661520055499e-05,
+      "loss": 0.1159,
+      "step": 871000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.703493008222175e-05,
+      "loss": 0.1154,
+      "step": 872000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.688330797899925e-05,
+      "loss": 0.1154,
+      "step": 873000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.673174962782796e-05,
+      "loss": 0.1155,
+      "step": 874000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.658025576533833e-05,
+      "loss": 0.1159,
+      "step": 875000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.642882712784742e-05,
+      "loss": 0.1159,
+      "step": 876000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.627746445135539e-05,
+      "loss": 0.1153,
+      "step": 877000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.612616847154166e-05,
+      "loss": 0.1151,
+      "step": 878000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.597493992376152e-05,
+      "loss": 0.1153,
+      "step": 879000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.582377954304257e-05,
+      "loss": 0.115,
+      "step": 880000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.5672688064081e-05,
+      "loss": 0.1151,
+      "step": 881000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.552166622123824e-05,
+      "loss": 0.1148,
+      "step": 882000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.537071474853711e-05,
+      "loss": 0.1148,
+      "step": 883000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.52198343796585e-05,
+      "loss": 0.1145,
+      "step": 884000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.506902584793773e-05,
+      "loss": 0.1153,
+      "step": 885000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.491828988636088e-05,
+      "loss": 0.1146,
+      "step": 886000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.476762722756134e-05,
+      "loss": 0.1146,
+      "step": 887000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.461703860381628e-05,
+      "loss": 0.1149,
+      "step": 888000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.446652474704295e-05,
+      "loss": 0.1145,
+      "step": 889000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.431608638879529e-05,
+      "loss": 0.1144,
+      "step": 890000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.41657242602602e-05,
+      "loss": 0.114,
+      "step": 891000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.40154390922541e-05,
+      "loss": 0.1142,
+      "step": 892000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.386523161521942e-05,
+      "loss": 0.1138,
+      "step": 893000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.371510255922088e-05,
+      "loss": 0.114,
+      "step": 894000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.356505265394214e-05,
+      "loss": 0.1146,
+      "step": 895000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.34150826286821e-05,
+      "loss": 0.1143,
+      "step": 896000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.326519321235139e-05,
+      "loss": 0.1144,
+      "step": 897000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.311538513346895e-05,
+      "loss": 0.1141,
+      "step": 898000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.296565912015829e-05,
+      "loss": 0.1152,
+      "step": 899000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.281601590014407e-05,
+      "loss": 0.1141,
+      "step": 900000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.266645620074858e-05,
+      "loss": 0.1141,
+      "step": 901000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.251698074888813e-05,
+      "loss": 0.1141,
+      "step": 902000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.236759027106965e-05,
+      "loss": 0.1141,
+      "step": 903000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.221828549338689e-05,
+      "loss": 0.1139,
+      "step": 904000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.206906714151719e-05,
+      "loss": 0.1141,
+      "step": 905000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.191993594071785e-05,
+      "loss": 0.1134,
+      "step": 906000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.17708926158225e-05,
+      "loss": 0.1135,
+      "step": 907000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.162193789123765e-05,
+      "loss": 0.114,
+      "step": 908000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.147307249093929e-05,
+      "loss": 0.1137,
+      "step": 909000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.132429713846917e-05,
+      "loss": 0.1133,
+      "step": 910000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.11756125569314e-05,
+      "loss": 0.1135,
+      "step": 911000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.102701946898891e-05,
+      "loss": 0.1135,
+      "step": 912000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.087851859685991e-05,
+      "loss": 0.1134,
+      "step": 913000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.073011066231448e-05,
+      "loss": 0.1136,
+      "step": 914000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.058179638667089e-05,
+      "loss": 0.1131,
+      "step": 915000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.043357649079223e-05,
+      "loss": 0.1133,
+      "step": 916000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.0285451695082965e-05,
+      "loss": 0.1135,
+      "step": 917000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.0137422719485145e-05,
+      "loss": 0.1128,
+      "step": 918000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.9989490283475275e-05,
+      "loss": 0.1129,
+      "step": 919000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.984165510606058e-05,
+      "loss": 0.113,
+      "step": 920000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.969391790577551e-05,
+      "loss": 0.1132,
+      "step": 921000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.9546279400678404e-05,
+      "loss": 0.1128,
+      "step": 922000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.939874030834782e-05,
+      "loss": 0.1129,
+      "step": 923000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.925130134587924e-05,
+      "loss": 0.1127,
+      "step": 924000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.91039632298814e-05,
+      "loss": 0.1128,
+      "step": 925000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.895672667647284e-05,
+      "loss": 0.1128,
+      "step": 926000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.880959240127858e-05,
+      "loss": 0.1127,
+      "step": 927000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.8662561119426485e-05,
+      "loss": 0.1123,
+      "step": 928000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.8515633545543756e-05,
+      "loss": 0.1123,
+      "step": 929000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.8368810393753684e-05,
+      "loss": 0.1125,
+      "step": 930000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.8222092377671884e-05,
+      "loss": 0.1128,
+      "step": 931000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.80754802104031e-05,
+      "loss": 0.1123,
+      "step": 932000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.7928974604537494e-05,
+      "loss": 0.1122,
+      "step": 933000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.778257627214737e-05,
+      "loss": 0.1125,
+      "step": 934000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.763628592478362e-05,
+      "loss": 0.1128,
+      "step": 935000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.749010427347233e-05,
+      "loss": 0.1127,
+      "step": 936000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.734403202871118e-05,
+      "loss": 0.1124,
+      "step": 937000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.7198069900466204e-05,
+      "loss": 0.1119,
+      "step": 938000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.7052218598168154e-05,
+      "loss": 0.1119,
+      "step": 939000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.690647883070921e-05,
+      "loss": 0.113,
+      "step": 940000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.676085130643933e-05,
+      "loss": 0.1116,
+      "step": 941000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.661533673316303e-05,
+      "loss": 0.1114,
+      "step": 942000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.6469935818135804e-05,
+      "loss": 0.1118,
+      "step": 943000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.632464926806071e-05,
+      "loss": 0.1121,
+      "step": 944000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.617947778908498e-05,
+      "loss": 0.1113,
+      "step": 945000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.603442208679657e-05,
+      "loss": 0.1116,
+      "step": 946000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.5889482866220613e-05,
+      "loss": 0.1118,
+      "step": 947000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.574466083181624e-05,
+      "loss": 0.1117,
+      "step": 948000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.559995668747294e-05,
+      "loss": 0.1119,
+      "step": 949000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.545537113650719e-05,
+      "loss": 0.1115,
+      "step": 950000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.5310904881659116e-05,
+      "loss": 0.1119,
+      "step": 951000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.516655862508897e-05,
+      "loss": 0.1117,
+      "step": 952000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.5022333068373835e-05,
+      "loss": 0.1111,
+      "step": 953000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.487822891250406e-05,
+      "loss": 0.1112,
+      "step": 954000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.473424685787998e-05,
+      "loss": 0.1115,
+      "step": 955000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.4590387604308476e-05,
+      "loss": 0.1115,
+      "step": 956000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.4446651850999604e-05,
+      "loss": 0.1113,
+      "step": 957000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.4303040296563075e-05,
+      "loss": 0.111,
+      "step": 958000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.4159553639005044e-05,
+      "loss": 0.1108,
+      "step": 959000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.401619257572453e-05,
+      "loss": 0.1112,
+      "step": 960000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.3872957803510226e-05,
+      "loss": 0.111,
+      "step": 961000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.3729850018536874e-05,
+      "loss": 0.1107,
+      "step": 962000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.358686991636209e-05,
+      "loss": 0.1111,
+      "step": 963000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.344401819192292e-05,
+      "loss": 0.1108,
+      "step": 964000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.330129553953237e-05,
+      "loss": 0.1107,
+      "step": 965000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.315870265287618e-05,
+      "loss": 0.1104,
+      "step": 966000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.3016240225009354e-05,
+      "loss": 0.1101,
+      "step": 967000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.287390894835279e-05,
+      "loss": 0.11,
+      "step": 968000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.2731709514689995e-05,
+      "loss": 0.1108,
+      "step": 969000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.2589642615163604e-05,
+      "loss": 0.1108,
+      "step": 970000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.244770894027215e-05,
+      "loss": 0.1101,
+      "step": 971000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.2305909179866635e-05,
+      "loss": 0.1106,
+      "step": 972000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.2164244023147105e-05,
+      "loss": 0.1101,
+      "step": 973000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.202271415865952e-05,
+      "loss": 0.1104,
+      "step": 974000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.188132027429215e-05,
+      "loss": 0.1103,
+      "step": 975000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.1740063057272385e-05,
+      "loss": 0.1102,
+      "step": 976000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.1598943194163414e-05,
+      "loss": 0.1098,
+      "step": 977000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.145796137086076e-05,
+      "loss": 0.1105,
+      "step": 978000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.131711827258904e-05,
+      "loss": 0.1104,
+      "step": 979000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.117641458389866e-05,
+      "loss": 0.1098,
+      "step": 980000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.103585098866237e-05,
+      "loss": 0.1095,
+      "step": 981000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.089542817007211e-05,
+      "loss": 0.1101,
+      "step": 982000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.075514681063549e-05,
+      "loss": 0.1096,
+      "step": 983000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.061500759217261e-05,
+      "loss": 0.1102,
+      "step": 984000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.047501119581276e-05,
+      "loss": 0.11,
+      "step": 985000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.033515830199098e-05,
+      "loss": 0.1096,
+      "step": 986000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.01954495904449e-05,
+      "loss": 0.11,
+      "step": 987000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.005588574021138e-05,
+      "loss": 0.1093,
+      "step": 988000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.9916467429623105e-05,
+      "loss": 0.1105,
+      "step": 989000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.97771953363055e-05,
+      "loss": 0.1088,
+      "step": 990000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.963807013717323e-05,
+      "loss": 0.1093,
+      "step": 991000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.949909250842702e-05,
+      "loss": 0.1095,
+      "step": 992000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.936026312555037e-05,
+      "loss": 0.1092,
+      "step": 993000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.922158266330625e-05,
+      "loss": 0.1095,
+      "step": 994000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.908305179573382e-05,
+      "loss": 0.1094,
+      "step": 995000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.8944671196145136e-05,
+      "loss": 0.109,
+      "step": 996000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.880644153712186e-05,
+      "loss": 0.1089,
+      "step": 997000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.8668363490512126e-05,
+      "loss": 0.1091,
+      "step": 998000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.853043772742709e-05,
+      "loss": 0.1088,
+      "step": 999000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.839266491823776e-05,
+      "loss": 0.109,
+      "step": 1000000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.825504573257176e-05,
+      "loss": 0.1088,
+      "step": 1001000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.811758083931005e-05,
+      "loss": 0.1087,
+      "step": 1002000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.7980270906583677e-05,
+      "loss": 0.1086,
+      "step": 1003000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.784311660177046e-05,
+      "loss": 0.1089,
+      "step": 1004000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.770611859149185e-05,
+      "loss": 0.1091,
+      "step": 1005000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.756927754160969e-05,
+      "loss": 0.109,
+      "step": 1006000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.7432594117222835e-05,
+      "loss": 0.1082,
+      "step": 1007000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.729606898266411e-05,
+      "loss": 0.1083,
+      "step": 1008000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.715970280149699e-05,
+      "loss": 0.1085,
+      "step": 1009000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.702349623651229e-05,
+      "loss": 0.1085,
+      "step": 1010000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.688744994972514e-05,
+      "loss": 0.1086,
+      "step": 1011000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.675156460237158e-05,
+      "loss": 0.1084,
+      "step": 1012000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.6615840854905425e-05,
+      "loss": 0.1082,
+      "step": 1013000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.6480279366995116e-05,
+      "loss": 0.1078,
+      "step": 1014000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.634488079752041e-05,
+      "loss": 0.1081,
+      "step": 1015000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.620964580456925e-05,
+      "loss": 0.108,
+      "step": 1016000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.607457504543447e-05,
+      "loss": 0.1079,
+      "step": 1017000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.5939669176610684e-05,
+      "loss": 0.108,
+      "step": 1018000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.5804928853791155e-05,
+      "loss": 0.1079,
+      "step": 1019000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.567035473186444e-05,
+      "loss": 0.1083,
+      "step": 1020000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.553594746491134e-05,
+      "loss": 0.1072,
+      "step": 1021000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.5401707706201646e-05,
+      "loss": 0.1077,
+      "step": 1022000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.5267636108191036e-05,
+      "loss": 0.1079,
+      "step": 1023000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.5133733322517886e-05,
+      "loss": 0.108,
+      "step": 1024000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.500000000000001e-05,
+      "loss": 0.1078,
+      "step": 1025000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4866436790631564e-05,
+      "loss": 0.1078,
+      "step": 1026000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4733044343579996e-05,
+      "loss": 0.1078,
+      "step": 1027000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4599823307182626e-05,
+      "loss": 0.1077,
+      "step": 1028000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4466774328943796e-05,
+      "loss": 0.1077,
+      "step": 1029000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.433389805553156e-05,
+      "loss": 0.1075,
+      "step": 1030000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.420119513277446e-05,
+      "loss": 0.1074,
+      "step": 1031000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.406866620565862e-05,
+      "loss": 0.1074,
+      "step": 1032000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.393631191832436e-05,
+      "loss": 0.1075,
+      "step": 1033000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.380413291406325e-05,
+      "loss": 0.1074,
+      "step": 1034000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.3672129835314955e-05,
+      "loss": 0.1069,
+      "step": 1035000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.3540303323663976e-05,
+      "loss": 0.1074,
+      "step": 1036000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.340865401983671e-05,
+      "loss": 0.1069,
+      "step": 1037000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.327718256369826e-05,
+      "loss": 0.1065,
+      "step": 1038000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.314588959424926e-05,
+      "loss": 0.1069,
+      "step": 1039000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.3014775749622876e-05,
+      "loss": 0.1072,
+      "step": 1040000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.2883841667081675e-05,
+      "loss": 0.107,
+      "step": 1041000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.275308798301446e-05,
+      "loss": 0.1069,
+      "step": 1042000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.2622515332933254e-05,
+      "loss": 0.1069,
+      "step": 1043000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.2492124351470214e-05,
+      "loss": 0.1068,
+      "step": 1044000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.2361915672374536e-05,
+      "loss": 0.1064,
+      "step": 1045000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.223188992850925e-05,
+      "loss": 0.1067,
+      "step": 1046000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.210204775184834e-05,
+      "loss": 0.1066,
+      "step": 1047000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1972389773473585e-05,
+      "loss": 0.1066,
+      "step": 1048000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1842916623571476e-05,
+      "loss": 0.1061,
+      "step": 1049000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.171362893143013e-05,
+      "loss": 0.1061,
+      "step": 1050000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1584527325436315e-05,
+      "loss": 0.1063,
+      "step": 1051000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1455612433072357e-05,
+      "loss": 0.1065,
+      "step": 1052000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1326884880913074e-05,
+      "loss": 0.1064,
+      "step": 1053000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.119834529462273e-05,
+      "loss": 0.1061,
+      "step": 1054000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.106999429895195e-05,
+      "loss": 0.1062,
+      "step": 1055000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.0941832517734885e-05,
+      "loss": 0.1057,
+      "step": 1056000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.0813860573885845e-05,
+      "loss": 0.106,
+      "step": 1057000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.068607908939662e-05,
+      "loss": 0.1065,
+      "step": 1058000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.0558488685333235e-05,
+      "loss": 0.1059,
+      "step": 1059000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.0431089981832965e-05,
+      "loss": 0.1059,
+      "step": 1060000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.030388359810139e-05,
+      "loss": 0.1061,
+      "step": 1061000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.0176870152409324e-05,
+      "loss": 0.1063,
+      "step": 1062000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.005005026208979e-05,
+      "loss": 0.1063,
+      "step": 1063000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.992342454353516e-05,
+      "loss": 0.1058,
+      "step": 1064000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.979699361219395e-05,
+      "loss": 0.1057,
+      "step": 1065000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.967075808256801e-05,
+      "loss": 0.1057,
+      "step": 1066000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.954471856820947e-05,
+      "loss": 0.1057,
+      "step": 1067000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.941887568171766e-05,
+      "loss": 0.1057,
+      "step": 1068000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.9293230034736354e-05,
+      "loss": 0.1052,
+      "step": 1069000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.916778223795057e-05,
+      "loss": 0.1055,
+      "step": 1070000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.904253290108369e-05,
+      "loss": 0.1053,
+      "step": 1071000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.89174826328946e-05,
+      "loss": 0.1057,
+      "step": 1072000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.879263204117456e-05,
+      "loss": 0.1054,
+      "step": 1073000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.86679817327444e-05,
+      "loss": 0.1053,
+      "step": 1074000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.854353231345141e-05,
+      "loss": 0.1054,
+      "step": 1075000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.84192843881665e-05,
+      "loss": 0.1051,
+      "step": 1076000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.8295238560781317e-05,
+      "loss": 0.1049,
+      "step": 1077000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.817139543420515e-05,
+      "loss": 0.1049,
+      "step": 1078000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.804775561036213e-05,
+      "loss": 0.105,
+      "step": 1079000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.792431969018824e-05,
+      "loss": 0.105,
+      "step": 1080000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.780108827362846e-05,
+      "loss": 0.1047,
+      "step": 1081000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.767806195963376e-05,
+      "loss": 0.1046,
+      "step": 1082000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.755524134615825e-05,
+      "loss": 0.1048,
+      "step": 1083000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.743262703015622e-05,
+      "loss": 0.1047,
+      "step": 1084000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.731021960757937e-05,
+      "loss": 0.1053,
+      "step": 1085000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7188019673373706e-05,
+      "loss": 0.1047,
+      "step": 1086000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.706602782147684e-05,
+      "loss": 0.1056,
+      "step": 1087000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.694424464481503e-05,
+      "loss": 0.1045,
+      "step": 1088000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.682267073530023e-05,
+      "loss": 0.1045,
+      "step": 1089000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.670130668382735e-05,
+      "loss": 0.105,
+      "step": 1090000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.6580153080271256e-05,
+      "loss": 0.1038,
+      "step": 1091000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.645921051348396e-05,
+      "loss": 0.1046,
+      "step": 1092000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.6338479571291814e-05,
+      "loss": 0.1044,
+      "step": 1093000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.621796084049252e-05,
+      "loss": 0.1041,
+      "step": 1094000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.6097654906852405e-05,
+      "loss": 0.1043,
+      "step": 1095000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.5977562355103535e-05,
+      "loss": 0.1037,
+      "step": 1096000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.5857683768940766e-05,
+      "loss": 0.1048,
+      "step": 1097000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.573801973101913e-05,
+      "loss": 0.1042,
+      "step": 1098000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.561857082295077e-05,
+      "loss": 0.1042,
+      "step": 1099000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.5499337625302255e-05,
+      "loss": 0.1045,
+      "step": 1100000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.5380320717591716e-05,
+      "loss": 0.1038,
+      "step": 1101000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.526152067828607e-05,
+      "loss": 0.1037,
+      "step": 1102000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.514293808479813e-05,
+      "loss": 0.1041,
+      "step": 1103000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.5024573513483864e-05,
+      "loss": 0.1047,
+      "step": 1104000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.490642753963951e-05,
+      "loss": 0.1039,
+      "step": 1105000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.4788500737498943e-05,
+      "loss": 0.104,
+      "step": 1106000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.467079368023068e-05,
+      "loss": 0.1037,
+      "step": 1107000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.4553306939935267e-05,
+      "loss": 0.1039,
+      "step": 1108000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.443604108764242e-05,
+      "loss": 0.1037,
+      "step": 1109000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.431899669330819e-05,
+      "loss": 0.1037,
+      "step": 1110000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.420217432581236e-05,
+      "loss": 0.1036,
+      "step": 1111000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.40855745529555e-05,
+      "loss": 0.1036,
+      "step": 1112000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.396919794145629e-05,
+      "loss": 0.1036,
+      "step": 1113000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.385304505694882e-05,
+      "loss": 0.1033,
+      "step": 1114000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.373711646397972e-05,
+      "loss": 0.1036,
+      "step": 1115000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.362141272600552e-05,
+      "loss": 0.1033,
+      "step": 1116000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.3505934405389874e-05,
+      "loss": 0.1029,
+      "step": 1117000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.339068206340077e-05,
+      "loss": 0.1032,
+      "step": 1118000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.327565626020793e-05,
+      "loss": 0.1031,
+      "step": 1119000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.3160857554879946e-05,
+      "loss": 0.1029,
+      "step": 1120000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.3046286505381636e-05,
+      "loss": 0.1033,
+      "step": 1121000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.293194366857137e-05,
+      "loss": 0.1026,
+      "step": 1122000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.281782960019829e-05,
+      "loss": 0.1033,
+      "step": 1123000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.270394485489967e-05,
+      "loss": 0.1028,
+      "step": 1124000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.259028998619814e-05,
+      "loss": 0.1024,
+      "step": 1125000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.2476865546499015e-05,
+      "loss": 0.1029,
+      "step": 1126000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.236367208708775e-05,
+      "loss": 0.1031,
+      "step": 1127000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.2250710158127045e-05,
+      "loss": 0.1031,
+      "step": 1128000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.2137980308654335e-05,
+      "loss": 0.1025,
+      "step": 1129000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.202548308657902e-05,
+      "loss": 0.1031,
+      "step": 1130000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.191321903867988e-05,
+      "loss": 0.1023,
+      "step": 1131000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.180118871060238e-05,
+      "loss": 0.102,
+      "step": 1132000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.168939264685598e-05,
+      "loss": 0.1028,
+      "step": 1133000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.157783139081155e-05,
+      "loss": 0.1024,
+      "step": 1134000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1466505484698735e-05,
+      "loss": 0.1028,
+      "step": 1135000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.135541546960322e-05,
+      "loss": 0.1022,
+      "step": 1136000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1244561885464244e-05,
+      "loss": 0.1023,
+      "step": 1137000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.113394527107189e-05,
+      "loss": 0.1026,
+      "step": 1138000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.102356616406441e-05,
+      "loss": 0.1024,
+      "step": 1139000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0913425100925795e-05,
+      "loss": 0.1024,
+      "step": 1140000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.080352261698297e-05,
+      "loss": 0.1022,
+      "step": 1141000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.069385924640326e-05,
+      "loss": 0.1022,
+      "step": 1142000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0584435522191896e-05,
+      "loss": 0.1018,
+      "step": 1143000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0475251976189245e-05,
+      "loss": 0.1019,
+      "step": 1144000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.036630913906839e-05,
+      "loss": 0.1021,
+      "step": 1145000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.025760754033246e-05,
+      "loss": 0.1019,
+      "step": 1146000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0149147708312025e-05,
+      "loss": 0.1019,
+      "step": 1147000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.004093017016266e-05,
+      "loss": 0.1027,
+      "step": 1148000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.993295545186223e-05,
+      "loss": 0.1011,
+      "step": 1149000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9825224078208425e-05,
+      "loss": 0.102,
+      "step": 1150000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9717736572816195e-05,
+      "loss": 0.1018,
+      "step": 1151000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.961049345811523e-05,
+      "loss": 0.1014,
+      "step": 1152000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.950349525534735e-05,
+      "loss": 0.1013,
+      "step": 1153000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9396742484564013e-05,
+      "loss": 0.1013,
+      "step": 1154000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.929023566462377e-05,
+      "loss": 0.1013,
+      "step": 1155000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9183975313189824e-05,
+      "loss": 0.1017,
+      "step": 1156000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9077961946727368e-05,
+      "loss": 0.1017,
+      "step": 1157000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.8972196080501208e-05,
+      "loss": 0.1018,
+      "step": 1158000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8866678228573168e-05,
+      "loss": 0.1015,
+      "step": 1159000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.876140890379968e-05,
+      "loss": 0.1007,
+      "step": 1160000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.865638861782922e-05,
+      "loss": 0.1008,
+      "step": 1161000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8551617881099812e-05,
+      "loss": 0.1013,
+      "step": 1162000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.844709720283657e-05,
+      "loss": 0.101,
+      "step": 1163000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8342827091049336e-05,
+      "loss": 0.1011,
+      "step": 1164000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.8238808052529954e-05,
+      "loss": 0.1015,
+      "step": 1165000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.8135040592850052e-05,
+      "loss": 0.1015,
+      "step": 1166000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.803152521635851e-05,
+      "loss": 0.1008,
+      "step": 1167000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.7928262426178886e-05,
+      "loss": 0.1007,
+      "step": 1168000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.7825252724207188e-05,
+      "loss": 0.1007,
+      "step": 1169000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.7722496611109243e-05,
+      "loss": 0.101,
+      "step": 1170000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.7619994586318354e-05,
+      "loss": 0.1006,
+      "step": 1171000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.7517747148032903e-05,
+      "loss": 0.1008,
+      "step": 1172000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.7415754793213826e-05,
+      "loss": 0.101,
+      "step": 1173000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.731401801758229e-05,
+      "loss": 0.1011,
+      "step": 1174000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.721253731561727e-05,
+      "loss": 0.1008,
+      "step": 1175000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.7111313180553077e-05,
+      "loss": 0.1011,
+      "step": 1176000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.701034610437706e-05,
+      "loss": 0.1006,
+      "step": 1177000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6909636577827112e-05,
+      "loss": 0.1006,
+      "step": 1178000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6809185090389406e-05,
+      "loss": 0.1009,
+      "step": 1179000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6708992130295876e-05,
+      "loss": 0.0998,
+      "step": 1180000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6609058184521975e-05,
+      "loss": 0.1002,
+      "step": 1181000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6509383738784218e-05,
+      "loss": 0.0997,
+      "step": 1182000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6409969277537865e-05,
+      "loss": 0.1004,
+      "step": 1183000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6310815283974518e-05,
+      "loss": 0.0999,
+      "step": 1184000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6211922240019883e-05,
+      "loss": 0.1002,
+      "step": 1185000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.611329062633126e-05,
+      "loss": 0.1003,
+      "step": 1186000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6014920922295384e-05,
+      "loss": 0.0999,
+      "step": 1187000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.591681360602595e-05,
+      "loss": 0.1,
+      "step": 1188000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.5818969154361387e-05,
+      "loss": 0.0998,
+      "step": 1189000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.57213880428625e-05,
+      "loss": 0.1006,
+      "step": 1190000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.562407074581014e-05,
+      "loss": 0.1001,
+      "step": 1191000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.552701773620292e-05,
+      "loss": 0.0998,
+      "step": 1192000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.543022948575495e-05,
+      "loss": 0.0997,
+      "step": 1193000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.533370646489347e-05,
+      "loss": 0.0992,
+      "step": 1194000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.523744914275664e-05,
+      "loss": 0.0996,
+      "step": 1195000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.5141457987191217e-05,
+      "loss": 0.0999,
+      "step": 1196000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.504573346475026e-05,
+      "loss": 0.0996,
+      "step": 1197000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.495027604069094e-05,
+      "loss": 0.0997,
+      "step": 1198000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.485508617897216e-05,
+      "loss": 0.0994,
+      "step": 1199000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.476016434225246e-05,
+      "loss": 0.1001,
+      "step": 1200000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.466551099188757e-05,
+      "loss": 0.0992,
+      "step": 1201000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.4571126587928378e-05,
+      "loss": 0.0996,
+      "step": 1202000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.447701158911855e-05,
+      "loss": 0.0998,
+      "step": 1203000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.438316645289233e-05,
+      "loss": 0.0995,
+      "step": 1204000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.4289591635372306e-05,
+      "loss": 0.0995,
+      "step": 1205000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.4196287591367296e-05,
+      "loss": 0.0993,
+      "step": 1206000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.4103254774369966e-05,
+      "loss": 0.099,
+      "step": 1207000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4010493636554805e-05,
+      "loss": 0.099,
+      "step": 1208000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3918004628775736e-05,
+      "loss": 0.0992,
+      "step": 1209000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3825788200564126e-05,
+      "loss": 0.099,
+      "step": 1210000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3733844800126465e-05,
+      "loss": 0.099,
+      "step": 1211000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.364217487434221e-05,
+      "loss": 0.0994,
+      "step": 1212000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.355077886876161e-05,
+      "loss": 0.0996,
+      "step": 1213000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3459657227603646e-05,
+      "loss": 0.0991,
+      "step": 1214000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3368810393753687e-05,
+      "loss": 0.0994,
+      "step": 1215000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3278238808761482e-05,
+      "loss": 0.0991,
+      "step": 1216000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3187942912838988e-05,
+      "loss": 0.0987,
+      "step": 1217000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.309792314485815e-05,
+      "loss": 0.0991,
+      "step": 1218000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3008179942348898e-05,
+      "loss": 0.0985,
+      "step": 1219000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2918713741496864e-05,
+      "loss": 0.0989,
+      "step": 1220000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.282952497714145e-05,
+      "loss": 0.0993,
+      "step": 1221000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2740614082773525e-05,
+      "loss": 0.099,
+      "step": 1222000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2651981490533433e-05,
+      "loss": 0.099,
+      "step": 1223000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2563627631208887e-05,
+      "loss": 0.0984,
+      "step": 1224000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2475552934232845e-05,
+      "loss": 0.0987,
+      "step": 1225000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.23877578276814e-05,
+      "loss": 0.0984,
+      "step": 1226000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.230024273827179e-05,
+      "loss": 0.0985,
+      "step": 1227000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.221300809136019e-05,
+      "loss": 0.0983,
+      "step": 1228000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.212605431093979e-05,
+      "loss": 0.0985,
+      "step": 1229000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2039381819638596e-05,
+      "loss": 0.0985,
+      "step": 1230000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.19529910387175e-05,
+      "loss": 0.0984,
+      "step": 1231000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1866882388068175e-05,
+      "loss": 0.0982,
+      "step": 1232000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1781056286210997e-05,
+      "loss": 0.0984,
+      "step": 1233000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.169551315029304e-05,
+      "loss": 0.0984,
+      "step": 1234000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.161025339608612e-05,
+      "loss": 0.0981,
+      "step": 1235000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1525277437984636e-05,
+      "loss": 0.0983,
+      "step": 1236000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1440585689003713e-05,
+      "loss": 0.098,
+      "step": 1237000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.135617856077704e-05,
+      "loss": 0.098,
+      "step": 1238000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.1272056463554978e-05,
+      "loss": 0.0983,
+      "step": 1239000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.1188219806202544e-05,
+      "loss": 0.0974,
+      "step": 1240000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.1104668996197374e-05,
+      "loss": 0.0975,
+      "step": 1241000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.1021404439627775e-05,
+      "loss": 0.0976,
+      "step": 1242000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.093842654119081e-05,
+      "loss": 0.0979,
+      "step": 1243000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0855735704190202e-05,
+      "loss": 0.0974,
+      "step": 1244000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0773332330534513e-05,
+      "loss": 0.0974,
+      "step": 1245000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0691216820735112e-05,
+      "loss": 0.0973,
+      "step": 1246000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.060938957390418e-05,
+      "loss": 0.0974,
+      "step": 1247000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.052785098775293e-05,
+      "loss": 0.0976,
+      "step": 1248000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.044660145858949e-05,
+      "loss": 0.0971,
+      "step": 1249000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0365641381317115e-05,
+      "loss": 0.0977,
+      "step": 1250000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.028497114943219e-05,
+      "loss": 0.0971,
+      "step": 1251000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.020459115502234e-05,
+      "loss": 0.0972,
+      "step": 1252000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0124501788764537e-05,
+      "loss": 0.0976,
+      "step": 1253000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0044703439923217e-05,
+      "loss": 0.0972,
+      "step": 1254000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9965196496348313e-05,
+      "loss": 0.0972,
+      "step": 1255000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9885981344473453e-05,
+      "loss": 0.0967,
+      "step": 1256000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9807058369314016e-05,
+      "loss": 0.0972,
+      "step": 1257000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.972842795446534e-05,
+      "loss": 0.0972,
+      "step": 1258000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9650090482100758e-05,
+      "loss": 0.0968,
+      "step": 1259000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9572046332969825e-05,
+      "loss": 0.0969,
+      "step": 1260000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9494295886396442e-05,
+      "loss": 0.0972,
+      "step": 1261000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9416839520276984e-05,
+      "loss": 0.0966,
+      "step": 1262000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.933967761107847e-05,
+      "loss": 0.0969,
+      "step": 1263000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.926281053383679e-05,
+      "loss": 0.0971,
+      "step": 1264000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9186238662154803e-05,
+      "loss": 0.0971,
+      "step": 1265000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9109962368200602e-05,
+      "loss": 0.0971,
+      "step": 1266000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.903398202270558e-05,
+      "loss": 0.0968,
+      "step": 1267000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.895829799496281e-05,
+      "loss": 0.0972,
+      "step": 1268000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.888291065282509e-05,
+      "loss": 0.0968,
+      "step": 1269000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8807820362703212e-05,
+      "loss": 0.097,
+      "step": 1270000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8733027489564216e-05,
+      "loss": 0.0963,
+      "step": 1271000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8658532396929565e-05,
+      "loss": 0.0963,
+      "step": 1272000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8584335446873366e-05,
+      "loss": 0.0967,
+      "step": 1273000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8510437000020687e-05,
+      "loss": 0.0966,
+      "step": 1274000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8436837415545772e-05,
+      "loss": 0.0965,
+      "step": 1275000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.836353705117021e-05,
+      "loss": 0.0963,
+      "step": 1276000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.829053626316135e-05,
+      "loss": 0.0965,
+      "step": 1277000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8217835406330415e-05,
+      "loss": 0.0966,
+      "step": 1278000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8145434834030907e-05,
+      "loss": 0.0965,
+      "step": 1279000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8073334898156788e-05,
+      "loss": 0.0965,
+      "step": 1280000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.800153594914084e-05,
+      "loss": 0.0961,
+      "step": 1281000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.7930038335952955e-05,
+      "loss": 0.096,
+      "step": 1282000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.7858842406098367e-05,
+      "loss": 0.096,
+      "step": 1283000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.778794850561604e-05,
+      "loss": 0.0965,
+      "step": 1284000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.771735697907699e-05,
+      "loss": 0.0962,
+      "step": 1285000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.764706816958253e-05,
+      "loss": 0.0959,
+      "step": 1286000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.75770824187627e-05,
+      "loss": 0.0964,
+      "step": 1287000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.7507400066774532e-05,
+      "loss": 0.0958,
+      "step": 1288000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.7438021452300434e-05,
+      "loss": 0.096,
+      "step": 1289000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.7368946912546556e-05,
+      "loss": 0.096,
+      "step": 1290000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.7300176783241095e-05,
+      "loss": 0.0959,
+      "step": 1291000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.7231711398632736e-05,
+      "loss": 0.096,
+      "step": 1292000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.7163551091488952e-05,
+      "loss": 0.0964,
+      "step": 1293000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.7095696193094437e-05,
+      "loss": 0.0958,
+      "step": 1294000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.70281470332495e-05,
+      "loss": 0.0959,
+      "step": 1295000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.6960903940268456e-05,
+      "loss": 0.0963,
+      "step": 1296000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.6893967240977967e-05,
+      "loss": 0.0959,
+      "step": 1297000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.6827337260715564e-05,
+      "loss": 0.0954,
+      "step": 1298000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.6761014323327962e-05,
+      "loss": 0.0957,
+      "step": 1299000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.669499875116958e-05,
+      "loss": 0.0953,
+      "step": 1300000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.6629290865100885e-05,
+      "loss": 0.0954,
+      "step": 1301000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.6563890984486884e-05,
+      "loss": 0.0951,
+      "step": 1302000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.649879942719557e-05,
+      "loss": 0.0956,
+      "step": 1303000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.643401650959639e-05,
+      "loss": 0.0952,
+      "step": 1304000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.6369542546558626e-05,
+      "loss": 0.0958,
+      "step": 1305000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.6305377851449992e-05,
+      "loss": 0.0955,
+      "step": 1306000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6241522736134993e-05,
+      "loss": 0.0952,
+      "step": 1307000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.617797751097349e-05,
+      "loss": 0.095,
+      "step": 1308000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6114742484819136e-05,
+      "loss": 0.0954,
+      "step": 1309000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6051817965017933e-05,
+      "loss": 0.0952,
+      "step": 1310000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5989204257406693e-05,
+      "loss": 0.0954,
+      "step": 1311000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5926901666311547e-05,
+      "loss": 0.0959,
+      "step": 1312000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.586491049454648e-05,
+      "loss": 0.0951,
+      "step": 1313000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5803231043411912e-05,
+      "loss": 0.095,
+      "step": 1314000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5741863612693108e-05,
+      "loss": 0.0946,
+      "step": 1315000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5680808500658874e-05,
+      "loss": 0.0946,
+      "step": 1316000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.562006600405996e-05,
+      "loss": 0.0955,
+      "step": 1317000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5559636418127746e-05,
+      "loss": 0.0954,
+      "step": 1318000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5499520036572725e-05,
+      "loss": 0.095,
+      "step": 1319000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.543971715158307e-05,
+      "loss": 0.0952,
+      "step": 1320000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5380228053823295e-05,
+      "loss": 0.0955,
+      "step": 1321000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5321053032432762e-05,
+      "loss": 0.0954,
+      "step": 1322000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5262192375024284e-05,
+      "loss": 0.0947,
+      "step": 1323000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5203646367682786e-05,
+      "loss": 0.0947,
+      "step": 1324000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5145415294963877e-05,
+      "loss": 0.0947,
+      "step": 1325000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.508749943989242e-05,
+      "loss": 0.0951,
+      "step": 1326000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5029899083961249e-05,
+      "loss": 0.0947,
+      "step": 1327000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4972614507129716e-05,
+      "loss": 0.0944,
+      "step": 1328000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4915645987822406e-05,
+      "loss": 0.0948,
+      "step": 1329000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4858993802927718e-05,
+      "loss": 0.095,
+      "step": 1330000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4802658227796541e-05,
+      "loss": 0.0941,
+      "step": 1331000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4746639536240942e-05,
+      "loss": 0.0945,
+      "step": 1332000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4690938000532829e-05,
+      "loss": 0.0948,
+      "step": 1333000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.463555389140256e-05,
+      "loss": 0.0948,
+      "step": 1334000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4580487478037748e-05,
+      "loss": 0.0949,
+      "step": 1335000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4525739028081821e-05,
+      "loss": 0.0943,
+      "step": 1336000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4471308807632825e-05,
+      "loss": 0.0942,
+      "step": 1337000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4417197081242083e-05,
+      "loss": 0.0945,
+      "step": 1338000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.436340411191291e-05,
+      "loss": 0.0939,
+      "step": 1339000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4309930161099358e-05,
+      "loss": 0.0945,
+      "step": 1340000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4256775488704904e-05,
+      "loss": 0.0941,
+      "step": 1341000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.420394035308124e-05,
+      "loss": 0.0943,
+      "step": 1342000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4151425011026964e-05,
+      "loss": 0.0942,
+      "step": 1343000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4099229717786368e-05,
+      "loss": 0.0945,
+      "step": 1344000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.40473547270482e-05,
+      "loss": 0.0944,
+      "step": 1345000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3995800290944385e-05,
+      "loss": 0.0935,
+      "step": 1346000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3944566660048863e-05,
+      "loss": 0.0942,
+      "step": 1347000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3893654083376331e-05,
+      "loss": 0.0947,
+      "step": 1348000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3843062808381025e-05,
+      "loss": 0.0941,
+      "step": 1349000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3792793080955574e-05,
+      "loss": 0.0947,
+      "step": 1350000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.374284514542973e-05,
+      "loss": 0.0941,
+      "step": 1351000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3693219244569205e-05,
+      "loss": 0.0945,
+      "step": 1352000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3643915619574529e-05,
+      "loss": 0.0939,
+      "step": 1353000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3594934510079872e-05,
+      "loss": 0.0936,
+      "step": 1354000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3546276154151786e-05,
+      "loss": 0.0937,
+      "step": 1355000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3497940788288195e-05,
+      "loss": 0.0945,
+      "step": 1356000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3449928647417122e-05,
+      "loss": 0.094,
+      "step": 1357000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3402239964895614e-05,
+      "loss": 0.0939,
+      "step": 1358000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3354874972508582e-05,
+      "loss": 0.0942,
+      "step": 1359000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3307833900467679e-05,
+      "loss": 0.0939,
+      "step": 1360000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3261116977410184e-05,
+      "loss": 0.0934,
+      "step": 1361000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3214724430397915e-05,
+      "loss": 0.094,
+      "step": 1362000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.316865648491607e-05,
+      "loss": 0.0945,
+      "step": 1363000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3122913364872178e-05,
+      "loss": 0.0941,
+      "step": 1364000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3077495292594966e-05,
+      "loss": 0.0933,
+      "step": 1365000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3032402488833359e-05,
+      "loss": 0.0937,
+      "step": 1366000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2987635172755293e-05,
+      "loss": 0.0937,
+      "step": 1367000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2943193561946762e-05,
+      "loss": 0.0936,
+      "step": 1368000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2899077872410698e-05,
+      "loss": 0.0939,
+      "step": 1369000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2855288318565913e-05,
+      "loss": 0.094,
+      "step": 1370000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.28118251132461e-05,
+      "loss": 0.0932,
+      "step": 1371000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2768688467698762e-05,
+      "loss": 0.0936,
+      "step": 1372000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2725878591584188e-05,
+      "loss": 0.0931,
+      "step": 1373000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2683395692974472e-05,
+      "loss": 0.094,
+      "step": 1374000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.264123997835247e-05,
+      "loss": 0.0937,
+      "step": 1375000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2599411652610788e-05,
+      "loss": 0.0928,
+      "step": 1376000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2557910919050803e-05,
+      "loss": 0.0934,
+      "step": 1377000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2516737979381662e-05,
+      "loss": 0.0938,
+      "step": 1378000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2475893033719332e-05,
+      "loss": 0.0936,
+      "step": 1379000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.24353762805856e-05,
+      "loss": 0.0934,
+      "step": 1380000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.239518791690709e-05,
+      "loss": 0.0931,
+      "step": 1381000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2355328138014358e-05,
+      "loss": 0.0934,
+      "step": 1382000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2315797137640906e-05,
+      "loss": 0.0934,
+      "step": 1383000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2276595107922245e-05,
+      "loss": 0.0932,
+      "step": 1384000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2237722239394975e-05,
+      "loss": 0.0934,
+      "step": 1385000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2199178720995825e-05,
+      "loss": 0.0931,
+      "step": 1386000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2160964740060795e-05,
+      "loss": 0.0935,
+      "step": 1387000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2123080482324163e-05,
+      "loss": 0.0926,
+      "step": 1388000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2085526131917685e-05,
+      "loss": 0.0933,
+      "step": 1389000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2048301871369602e-05,
+      "loss": 0.0938,
+      "step": 1390000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2011407881603791e-05,
+      "loss": 0.0933,
+      "step": 1391000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.197484434193893e-05,
+      "loss": 0.0931,
+      "step": 1392000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1938611430087536e-05,
+      "loss": 0.0931,
+      "step": 1393000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1902709322155175e-05,
+      "loss": 0.0931,
+      "step": 1394000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1867138192639601e-05,
+      "loss": 0.0932,
+      "step": 1395000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1831898214429833e-05,
+      "loss": 0.093,
+      "step": 1396000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1796989558805432e-05,
+      "loss": 0.093,
+      "step": 1397000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.176241239543558e-05,
+      "loss": 0.0929,
+      "step": 1398000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1728166892378262e-05,
+      "loss": 0.0928,
+      "step": 1399000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1694253216079506e-05,
+      "loss": 0.093,
+      "step": 1400000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1660671531372517e-05,
+      "loss": 0.0932,
+      "step": 1401000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1627422001476893e-05,
+      "loss": 0.093,
+      "step": 1402000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.159450478799784e-05,
+      "loss": 0.0931,
+      "step": 1403000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.156192005092539e-05,
+      "loss": 0.0926,
+      "step": 1404000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1529667948633603e-05,
+      "loss": 0.0924,
+      "step": 1405000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1497748637879824e-05,
+      "loss": 0.0929,
+      "step": 1406000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1466162273803876e-05,
+      "loss": 0.0927,
+      "step": 1407000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1434909009927368e-05,
+      "loss": 0.0929,
+      "step": 1408000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1403988998152904e-05,
+      "loss": 0.0927,
+      "step": 1409000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1373402388763346e-05,
+      "loss": 0.0928,
+      "step": 1410000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1343149330421124e-05,
+      "loss": 0.0926,
+      "step": 1411000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1313229970167456e-05,
+      "loss": 0.0929,
+      "step": 1412000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1283644453421678e-05,
+      "loss": 0.0928,
+      "step": 1413000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1254392923980524e-05,
+      "loss": 0.0925,
+      "step": 1414000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1225475524017396e-05,
+      "loss": 0.0921,
+      "step": 1415000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1196892394081743e-05,
+      "loss": 0.0928,
+      "step": 1416000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1168643673098307e-05,
+      "loss": 0.0926,
+      "step": 1417000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1140729498366477e-05,
+      "loss": 0.0928,
+      "step": 1418000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1113150005559644e-05,
+      "loss": 0.0926,
+      "step": 1419000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1085905328724497e-05,
+      "loss": 0.0925,
+      "step": 1420000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1058995600280405e-05,
+      "loss": 0.0927,
+      "step": 1421000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1032420951018755e-05,
+      "loss": 0.0926,
+      "step": 1422000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1006181510102326e-05,
+      "loss": 0.0925,
+      "step": 1423000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.098027740506466e-05,
+      "loss": 0.0927,
+      "step": 1424000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0954708761809438e-05,
+      "loss": 0.0923,
+      "step": 1425000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0929475704609869e-05,
+      "loss": 0.0924,
+      "step": 1426000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0904578356108106e-05,
+      "loss": 0.092,
+      "step": 1427000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0880016837314599e-05,
+      "loss": 0.0927,
+      "step": 1428000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0855791267607573e-05,
+      "loss": 0.092,
+      "step": 1429000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0831901764732388e-05,
+      "loss": 0.0922,
+      "step": 1430000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0808348444801e-05,
+      "loss": 0.0923,
+      "step": 1431000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0785131422291397e-05,
+      "loss": 0.0923,
+      "step": 1432000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0762250810047028e-05,
+      "loss": 0.0919,
+      "step": 1433000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.073970671927628e-05,
+      "loss": 0.0923,
+      "step": 1434000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0717499259551896e-05,
+      "loss": 0.092,
+      "step": 1435000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0695628538810458e-05,
+      "loss": 0.0915,
+      "step": 1436000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0674094663351906e-05,
+      "loss": 0.092,
+      "step": 1437000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0652897737838953e-05,
+      "loss": 0.0923,
+      "step": 1438000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0632037865296626e-05,
+      "loss": 0.0922,
+      "step": 1439000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0611515147111736e-05,
+      "loss": 0.0922,
+      "step": 1440000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.059132968303241e-05,
+      "loss": 0.092,
+      "step": 1441000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.057148157116759e-05,
+      "loss": 0.0916,
+      "step": 1442000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0551970907986557e-05,
+      "loss": 0.0924,
+      "step": 1443000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0532797788318453e-05,
+      "loss": 0.0918,
+      "step": 1444000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0513962305351853e-05,
+      "loss": 0.0924,
+      "step": 1445000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0495464550634267e-05,
+      "loss": 0.092,
+      "step": 1446000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0477304614071756e-05,
+      "loss": 0.0924,
+      "step": 1447000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0459482583928413e-05,
+      "loss": 0.0918,
+      "step": 1448000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.044199854682601e-05,
+      "loss": 0.0921,
+      "step": 1449000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0424852587743537e-05,
+      "loss": 0.0923,
+      "step": 1450000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.04080447900168e-05,
+      "loss": 0.0918,
+      "step": 1451000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0391575235337991e-05,
+      "loss": 0.0919,
+      "step": 1452000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0375444003755357e-05,
+      "loss": 0.092,
+      "step": 1453000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.035965117367273e-05,
+      "loss": 0.0919,
+      "step": 1454000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0344196821849202e-05,
+      "loss": 0.0923,
+      "step": 1455000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0329081023398733e-05,
+      "loss": 0.0914,
+      "step": 1456000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0314303851789775e-05,
+      "loss": 0.0918,
+      "step": 1457000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0299865378844936e-05,
+      "loss": 0.0919,
+      "step": 1458000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0285765674740627e-05,
+      "loss": 0.0919,
+      "step": 1459000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0272004808006687e-05,
+      "loss": 0.0916,
+      "step": 1460000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.025858284552612e-05,
+      "loss": 0.0919,
+      "step": 1461000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0245499852534695e-05,
+      "loss": 0.0917,
+      "step": 1462000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.023275589262068e-05,
+      "loss": 0.0916,
+      "step": 1463000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.02203510277245e-05,
+      "loss": 0.0916,
+      "step": 1464000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0208285318138464e-05,
+      "loss": 0.0922,
+      "step": 1465000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0196558822506461e-05,
+      "loss": 0.0917,
+      "step": 1466000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.018517159782365e-05,
+      "loss": 0.0915,
+      "step": 1467000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0174123699436224e-05,
+      "loss": 0.092,
+      "step": 1468000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0163415181041128e-05,
+      "loss": 0.0919,
+      "step": 1469000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0153046094685783e-05,
+      "loss": 0.0916,
+      "step": 1470000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0143016490767838e-05,
+      "loss": 0.0919,
+      "step": 1471000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0133326418034942e-05,
+      "loss": 0.0916,
+      "step": 1472000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0123975923584488e-05,
+      "loss": 0.0919,
+      "step": 1473000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0114965052863392e-05,
+      "loss": 0.0917,
+      "step": 1474000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0106293849667868e-05,
+      "loss": 0.0919,
+      "step": 1475000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0097962356143219e-05,
+      "loss": 0.0915,
+      "step": 1476000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0089970612783647e-05,
+      "loss": 0.0915,
+      "step": 1477000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0082318658432013e-05,
+      "loss": 0.0918,
+      "step": 1478000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0075006530279694e-05,
+      "loss": 0.0917,
+      "step": 1479000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.006803426386639e-05,
+      "loss": 0.0912,
+      "step": 1480000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0061401893079914e-05,
+      "loss": 0.0915,
+      "step": 1481000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0055109450156098e-05,
+      "loss": 0.0919,
+      "step": 1482000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0049156965678577e-05,
+      "loss": 0.0919,
+      "step": 1483000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0043544468578657e-05,
+      "loss": 0.0914,
+      "step": 1484000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0038271986135177e-05,
+      "loss": 0.0912,
+      "step": 1485000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0033339543974399e-05,
+      "loss": 0.0915,
+      "step": 1486000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0028747166069832e-05,
+      "loss": 0.0915,
+      "step": 1487000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0024494874742152e-05,
+      "loss": 0.0916,
+      "step": 1488000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.002058269065911e-05,
+      "loss": 0.0913,
+      "step": 1489000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0017010632835382e-05,
+      "loss": 0.0915,
+      "step": 1490000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0013778718632507e-05,
+      "loss": 0.0915,
+      "step": 1491000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0010886963758806e-05,
+      "loss": 0.0923,
+      "step": 1492000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0008335382269296e-05,
+      "loss": 0.0914,
+      "step": 1493000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0006123986565623e-05,
+      "loss": 0.0911,
+      "step": 1494000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0004252787395989e-05,
+      "loss": 0.0913,
+      "step": 1495000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.000272179385514e-05,
+      "loss": 0.0918,
+      "step": 1496000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.000153101338428e-05,
+      "loss": 0.0913,
+      "step": 1497000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0000680451771032e-05,
+      "loss": 0.0915,
+      "step": 1498000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0000170113149459e-05,
+      "loss": 0.0918,
+      "step": 1499000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1e-05,
+      "loss": 0.0912,
+      "step": 1500000
+    },
+    {
+      "epoch": 2.43,
+      "step": 1500000,
+      "total_flos": 1.0515027621862995e+23,
+      "train_loss": 0.1377392346496582,
+      "train_runtime": 935051.3554,
+      "train_samples_per_second": 410.673,
+      "train_steps_per_second": 1.604
+    }
+  ],
+  "max_steps": 1500000,
+  "num_train_epochs": 3,
+  "total_flos": 1.0515027621862995e+23,
+  "trial_name": null,
+  "trial_params": null
+}