diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7848 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 1303,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 0.6228,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.6365,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-06,
+      "loss": 0.6543,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.5972,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 0.5361,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 0.5745,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-06,
+      "loss": 0.5942,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.4918,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-06,
+      "loss": 0.4242,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 0.2893,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 0.3571,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-06,
+      "loss": 0.3047,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 0.283,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7e-06,
+      "loss": 0.3107,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.2729,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.2882,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.5e-06,
+      "loss": 0.2396,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-06,
+      "loss": 0.292,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.5e-06,
+      "loss": 0.2642,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.2579,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0500000000000001e-05,
+      "loss": 0.2875,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.2772,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.15e-05,
+      "loss": 0.2433,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 0.2805,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25e-05,
+      "loss": 0.251,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.2667,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 0.2556,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4e-05,
+      "loss": 0.243,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.45e-05,
+      "loss": 0.2465,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.2405,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.55e-05,
+      "loss": 0.2587,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.2276,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.65e-05,
+      "loss": 0.2637,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7e-05,
+      "loss": 0.2574,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.2114,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8e-05,
+      "loss": 0.2324,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8500000000000002e-05,
+      "loss": 0.201,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9e-05,
+      "loss": 0.1838,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.95e-05,
+      "loss": 0.2421,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.2412,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999969064095054e-05,
+      "loss": 0.2747,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999876256571612e-05,
+      "loss": 0.2628,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999721578003894e-05,
+      "loss": 0.2518,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999505029348927e-05,
+      "loss": 0.1815,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999226611946536e-05,
+      "loss": 0.2665,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998886327519337e-05,
+      "loss": 0.2011,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998484178172735e-05,
+      "loss": 0.2118,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998020166394898e-05,
+      "loss": 0.2552,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999749429505675e-05,
+      "loss": 0.2316,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996906567411957e-05,
+      "loss": 0.191,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999625698709689e-05,
+      "loss": 0.1987,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995545558130624e-05,
+      "loss": 0.2211,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994772284914897e-05,
+      "loss": 0.2299,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993937172234094e-05,
+      "loss": 0.203,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993040225255205e-05,
+      "loss": 0.2178,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992081449527804e-05,
+      "loss": 0.2125,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991060850984007e-05,
+      "loss": 0.2457,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998997843593845e-05,
+      "loss": 0.2397,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998883421108822e-05,
+      "loss": 0.217,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987628183512854e-05,
+      "loss": 0.2744,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986360360674252e-05,
+      "loss": 0.1821,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998503075041667e-05,
+      "loss": 0.2382,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998363936096665e-05,
+      "loss": 0.2343,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982186200932964e-05,
+      "loss": 0.2371,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998067127930658e-05,
+      "loss": 0.2141,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997909460546059e-05,
+      "loss": 0.2029,
+      "step": 66
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977456189150164e-05,
+      "loss": 0.1962,
+      "step": 67
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975756040512477e-05,
+      "loss": 0.2177,
+      "step": 68
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973994170066654e-05,
+      "loss": 0.2207,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972170588713715e-05,
+      "loss": 0.1895,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997028530773648e-05,
+      "loss": 0.1954,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968338338799525e-05,
+      "loss": 0.2328,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966329693949098e-05,
+      "loss": 0.2108,
+      "step": 73
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964259385613053e-05,
+      "loss": 0.2124,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996212742660076e-05,
+      "loss": 0.1819,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959933830103034e-05,
+      "loss": 0.2227,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957678609692056e-05,
+      "loss": 0.196,
+      "step": 77
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995536177932128e-05,
+      "loss": 0.1704,
+      "step": 78
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952983353325358e-05,
+      "loss": 0.1788,
+      "step": 79
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950543346420042e-05,
+      "loss": 0.2139,
+      "step": 80
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994804177370209e-05,
+      "loss": 0.1984,
+      "step": 81
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945478650649192e-05,
+      "loss": 0.1946,
+      "step": 82
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942853993119852e-05,
+      "loss": 0.2256,
+      "step": 83
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99401678173533e-05,
+      "loss": 0.2323,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9937420139969397e-05,
+      "loss": 0.1898,
+      "step": 85
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934610977968514e-05,
+      "loss": 0.1957,
+      "step": 86
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931740348731447e-05,
+      "loss": 0.1924,
+      "step": 87
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928808270019297e-05,
+      "loss": 0.2029,
+      "step": 88
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992581475997337e-05,
+      "loss": 0.2148,
+      "step": 89
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992275983711505e-05,
+      "loss": 0.2034,
+      "step": 90
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919643520345698e-05,
+      "loss": 0.1851,
+      "step": 91
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916465828946533e-05,
+      "loss": 0.1973,
+      "step": 92
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913226782578506e-05,
+      "loss": 0.168,
+      "step": 93
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990992640128218e-05,
+      "loss": 0.2184,
+      "step": 94
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906564705477616e-05,
+      "loss": 0.2015,
+      "step": 95
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903141715964233e-05,
+      "loss": 0.1582,
+      "step": 96
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989965745392068e-05,
+      "loss": 0.1948,
+      "step": 97
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9896111940904724e-05,
+      "loss": 0.2361,
+      "step": 98
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892505198853097e-05,
+      "loss": 0.2212,
+      "step": 99
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988883725008136e-05,
+      "loss": 0.1744,
+      "step": 100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988510811728378e-05,
+      "loss": 0.1954,
+      "step": 101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881317823533176e-05,
+      "loss": 0.174,
+      "step": 102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877466392280773e-05,
+      "loss": 0.1631,
+      "step": 103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873553847356087e-05,
+      "loss": 0.1951,
+      "step": 104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986958021296673e-05,
+      "loss": 0.1931,
+      "step": 105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865545513698304e-05,
+      "loss": 0.1638,
+      "step": 106
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861449774514225e-05,
+      "loss": 0.2454,
+      "step": 107
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857293020755565e-05,
+      "loss": 0.1537,
+      "step": 108
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853075278140913e-05,
+      "loss": 0.2051,
+      "step": 109
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848796572766214e-05,
+      "loss": 0.1603,
+      "step": 110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9844456931104588e-05,
+      "loss": 0.1938,
+      "step": 111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984005638000618e-05,
+      "loss": 0.1907,
+      "step": 112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835594946698e-05,
+      "loss": 0.2059,
+      "step": 113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831072658783745e-05,
+      "loss": 0.2077,
+      "step": 114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826489544243623e-05,
+      "loss": 0.2125,
+      "step": 115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821845631434192e-05,
+      "loss": 0.1701,
+      "step": 116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817140949088193e-05,
+      "loss": 0.2443,
+      "step": 117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981237552631434e-05,
+      "loss": 0.2032,
+      "step": 118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980754939259716e-05,
+      "loss": 0.2058,
+      "step": 119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802662577796822e-05,
+      "loss": 0.2188,
+      "step": 120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797715112148937e-05,
+      "loss": 0.1956,
+      "step": 121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792707026264364e-05,
+      "loss": 0.1921,
+      "step": 122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978763835112904e-05,
+      "loss": 0.1755,
+      "step": 123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9782509118103773e-05,
+      "loss": 0.1808,
+      "step": 124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977731935892406e-05,
+      "loss": 0.1833,
+      "step": 125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772069105699878e-05,
+      "loss": 0.203,
+      "step": 126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766758390915494e-05,
+      "loss": 0.1536,
+      "step": 127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976138724742926e-05,
+      "loss": 0.1759,
+      "step": 128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755955708473416e-05,
+      "loss": 0.1912,
+      "step": 129
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750463807653873e-05,
+      "loss": 0.156,
+      "step": 130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744911578950023e-05,
+      "loss": 0.2351,
+      "step": 131
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739299056714497e-05,
+      "loss": 0.1415,
+      "step": 132
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733626275673e-05,
+      "loss": 0.1274,
+      "step": 133
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727893270924043e-05,
+      "loss": 0.1941,
+      "step": 134
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972210007793877e-05,
+      "loss": 0.1843,
+      "step": 135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716246732560715e-05,
+      "loss": 0.1786,
+      "step": 136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9710333271005586e-05,
+      "loss": 0.2245,
+      "step": 137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704359729861035e-05,
+      "loss": 0.1808,
+      "step": 138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698326146086446e-05,
+      "loss": 0.1969,
+      "step": 139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692232557012693e-05,
+      "loss": 0.1827,
+      "step": 140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686079000341914e-05,
+      "loss": 0.1504,
+      "step": 141
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967986551414728e-05,
+      "loss": 0.2186,
+      "step": 142
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967359213687275e-05,
+      "loss": 0.1589,
+      "step": 143
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966725890733285e-05,
+      "loss": 0.1551,
+      "step": 144
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660865864712413e-05,
+      "loss": 0.1783,
+      "step": 145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654413048566353e-05,
+      "loss": 0.2355,
+      "step": 146
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647900498819412e-05,
+      "loss": 0.1827,
+      "step": 147
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641328255765916e-05,
+      "loss": 0.2344,
+      "step": 148
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634696360069517e-05,
+      "loss": 0.1632,
+      "step": 149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9628004852762952e-05,
+      "loss": 0.1532,
+      "step": 150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96212537752478e-05,
+      "loss": 0.1341,
+      "step": 151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614443169294186e-05,
+      "loss": 0.1475,
+      "step": 152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960757307704057e-05,
+      "loss": 0.1717,
+      "step": 153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600643540993453e-05,
+      "loss": 0.2061,
+      "step": 154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959365460402713e-05,
+      "loss": 0.1491,
+      "step": 155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586606309383415e-05,
+      "loss": 0.1643,
+      "step": 156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579498700671386e-05,
+      "loss": 0.1744,
+      "step": 157
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572331821867108e-05,
+      "loss": 0.1926,
+      "step": 158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565105717313352e-05,
+      "loss": 0.1363,
+      "step": 159
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557820431719333e-05,
+      "loss": 0.2054,
+      "step": 160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955047601016044e-05,
+      "loss": 0.1936,
+      "step": 161
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954307249807793e-05,
+      "loss": 0.1923,
+      "step": 162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9535609941278676e-05,
+      "loss": 0.1788,
+      "step": 163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952808838593487e-05,
+      "loss": 0.1501,
+      "step": 164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520507878583728e-05,
+      "loss": 0.1892,
+      "step": 165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512868466127232e-05,
+      "loss": 0.1671,
+      "step": 166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95051701958318e-05,
+      "loss": 0.1625,
+      "step": 167
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497413115328028e-05,
+      "loss": 0.1973,
+      "step": 168
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489597272610377e-05,
+      "loss": 0.1787,
+      "step": 169
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481722716036883e-05,
+      "loss": 0.2179,
+      "step": 170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473789494328844e-05,
+      "loss": 0.168,
+      "step": 171
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465797656570546e-05,
+      "loss": 0.1288,
+      "step": 172
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457747252208936e-05,
+      "loss": 0.1256,
+      "step": 173
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944963833105332e-05,
+      "loss": 0.1658,
+      "step": 174
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944147094327506e-05,
+      "loss": 0.171,
+      "step": 175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9433245139407266e-05,
+      "loss": 0.1676,
+      "step": 176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942496097034447e-05,
+      "loss": 0.1388,
+      "step": 177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416618487342333e-05,
+      "loss": 0.2108,
+      "step": 178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94082177420173e-05,
+      "loss": 0.173,
+      "step": 179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399758786346305e-05,
+      "loss": 0.1769,
+      "step": 180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391241672666438e-05,
+      "loss": 0.1971,
+      "step": 181
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382666453674625e-05,
+      "loss": 0.1708,
+      "step": 182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374033182427297e-05,
+      "loss": 0.1697,
+      "step": 183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936534191234006e-05,
+      "loss": 0.1718,
+      "step": 184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356592697187383e-05,
+      "loss": 0.1803,
+      "step": 185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347785591102244e-05,
+      "loss": 0.1874,
+      "step": 186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338920648575798e-05,
+      "loss": 0.1672,
+      "step": 187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329997924457046e-05,
+      "loss": 0.183,
+      "step": 188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.93210174739525e-05,
+      "loss": 0.1642,
+      "step": 189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311979352625837e-05,
+      "loss": 0.1808,
+      "step": 190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302883616397546e-05,
+      "loss": 0.1992,
+      "step": 191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929373032154459e-05,
+      "loss": 0.1766,
+      "step": 192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928451952470007e-05,
+      "loss": 0.184,
+      "step": 193
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927525128285284e-05,
+      "loss": 0.1745,
+      "step": 194
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9265925653347198e-05,
+      "loss": 0.197,
+      "step": 195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256542693882505e-05,
+      "loss": 0.2009,
+      "step": 196
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247102462512823e-05,
+      "loss": 0.1737,
+      "step": 197
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237605017646573e-05,
+      "loss": 0.1817,
+      "step": 198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922805041804617e-05,
+      "loss": 0.1366,
+      "step": 199
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218438722827644e-05,
+      "loss": 0.1855,
+      "step": 200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208769991460298e-05,
+      "loss": 0.179,
+      "step": 201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919904428376632e-05,
+      "loss": 0.1719,
+      "step": 202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189261659920428e-05,
+      "loss": 0.1447,
+      "step": 203
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179422180449478e-05,
+      "loss": 0.1615,
+      "step": 204
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916952590623212e-05,
+      "loss": 0.1361,
+      "step": 205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159572898498387e-05,
+      "loss": 0.2001,
+      "step": 206
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149563218829345e-05,
+      "loss": 0.1883,
+      "step": 207
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9139496929156685e-05,
+      "loss": 0.1318,
+      "step": 208
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129374091762375e-05,
+      "loss": 0.1536,
+      "step": 209
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911919476927823e-05,
+      "loss": 0.1636,
+      "step": 210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910895902468557e-05,
+      "loss": 0.1565,
+      "step": 211
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098666921314793e-05,
+      "loss": 0.2114,
+      "step": 212
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908831852284501e-05,
+      "loss": 0.1618,
+      "step": 213
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907791389330363e-05,
+      "loss": 0.1819,
+      "step": 214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9067453097065986e-05,
+      "loss": 0.204,
+      "step": 215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905693619885491e-05,
+      "loss": 0.2068,
+      "step": 216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904636326374036e-05,
+      "loss": 0.1831,
+      "step": 217
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035734357139e-05,
+      "loss": 0.1766,
+      "step": 218
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025049544813793e-05,
+      "loss": 0.2153,
+      "step": 219
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014308892873612e-05,
+      "loss": 0.1774,
+      "step": 220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900351246777281e-05,
+      "loss": 0.2006,
+      "step": 221
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899266033631083e-05,
+      "loss": 0.1797,
+      "step": 222
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8981752565631767e-05,
+      "loss": 0.1843,
+      "step": 223
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970789223223978e-05,
+      "loss": 0.2014,
+      "step": 224
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895977037691964e-05,
+      "loss": 0.183,
+      "step": 225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948696094894354e-05,
+      "loss": 0.1566,
+      "step": 226
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937566445666707e-05,
+      "loss": 0.2221,
+      "step": 227
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892638149809785e-05,
+      "loss": 0.1717,
+      "step": 228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8915141321391083e-05,
+      "loss": 0.1841,
+      "step": 229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903845985091406e-05,
+      "loss": 0.1818,
+      "step": 230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889249555908512e-05,
+      "loss": 0.1988,
+      "step": 231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881090113599353e-05,
+      "loss": 0.1559,
+      "step": 232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8869629719201668e-05,
+      "loss": 0.163,
+      "step": 233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88581144467996e-05,
+      "loss": 0.1536,
+      "step": 234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846544367640218e-05,
+      "loss": 0.1746,
+      "step": 235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834919553309704e-05,
+      "loss": 0.1735,
+      "step": 236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882324007573288e-05,
+      "loss": 0.2136,
+      "step": 237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881150600717279e-05,
+      "loss": 0.1572,
+      "step": 238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799717420230242e-05,
+      "loss": 0.1514,
+      "step": 239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787874387843358e-05,
+      "loss": 0.1613,
+      "step": 240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775976983287117e-05,
+      "loss": 0.2228,
+      "step": 241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876402528017292e-05,
+      "loss": 0.1721,
+      "step": 242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752019352448114e-05,
+      "loss": 0.1595,
+      "step": 243
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873995927439555e-05,
+      "loss": 0.2287,
+      "step": 244
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727845120633108e-05,
+      "loss": 0.1858,
+      "step": 245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715676966113256e-05,
+      "loss": 0.1971,
+      "step": 246
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8703454886122568e-05,
+      "loss": 0.1451,
+      "step": 247
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869117895628126e-05,
+      "loss": 0.172,
+      "step": 248
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678849252542733e-05,
+      "loss": 0.147,
+      "step": 249
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86664658511931e-05,
+      "loss": 0.1631,
+      "step": 250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8654028828850705e-05,
+      "loss": 0.1617,
+      "step": 251
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641538262465656e-05,
+      "loss": 0.158,
+      "step": 252
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862899422931934e-05,
+      "loss": 0.1745,
+      "step": 253
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8616396807023975e-05,
+      "loss": 0.1735,
+      "step": 254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860374607352208e-05,
+      "loss": 0.1605,
+      "step": 255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859104210708604e-05,
+      "loss": 0.2098,
+      "step": 256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578284986317592e-05,
+      "loss": 0.1746,
+      "step": 257
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565474790147352e-05,
+      "loss": 0.1743,
+      "step": 258
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855261159783432e-05,
+      "loss": 0.1649,
+      "step": 259
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539695488965396e-05,
+      "loss": 0.2058,
+      "step": 260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8526726543454883e-05,
+      "loss": 0.1575,
+      "step": 261
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8513704841543997e-05,
+      "loss": 0.1764,
+      "step": 262
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500630463800364e-05,
+      "loss": 0.1185,
+      "step": 263
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8487503491117523e-05,
+      "loss": 0.2048,
+      "step": 264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847432400471443e-05,
+      "loss": 0.1627,
+      "step": 265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461092086134953e-05,
+      "loss": 0.1704,
+      "step": 266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447807817247368e-05,
+      "loss": 0.1682,
+      "step": 267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8434471280243854e-05,
+      "loss": 0.1464,
+      "step": 268
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8421082557639973e-05,
+      "loss": 0.1623,
+      "step": 269
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407641732274174e-05,
+      "loss": 0.1785,
+      "step": 270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8394148887307286e-05,
+      "loss": 0.1913,
+      "step": 271
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8380604106221973e-05,
+      "loss": 0.1482,
+      "step": 272
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367007472822253e-05,
+      "loss": 0.1428,
+      "step": 273
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353359071232954e-05,
+      "loss": 0.14,
+      "step": 274
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833965898589921e-05,
+      "loss": 0.1552,
+      "step": 275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8325907301585924e-05,
+      "loss": 0.1795,
+      "step": 276
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8312104103377266e-05,
+      "loss": 0.1943,
+      "step": 277
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8298249476676107e-05,
+      "loss": 0.2051,
+      "step": 278
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828434350720354e-05,
+      "loss": 0.1447,
+      "step": 279
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827038628099831e-05,
+      "loss": 0.1593,
+      "step": 280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8256377884416307e-05,
+      "loss": 0.2025,
+      "step": 281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824231840413001e-05,
+      "loss": 0.1594,
+      "step": 282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822820792712797e-05,
+      "loss": 0.2052,
+      "step": 283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8214046540714268e-05,
+      "loss": 0.1827,
+      "step": 284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8199834332507955e-05,
+      "loss": 0.1561,
+      "step": 285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8185571390442542e-05,
+      "loss": 0.186,
+      "step": 286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171257802765436e-05,
+      "loss": 0.1912,
+      "step": 287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8156893658037386e-05,
+      "loss": 0.1793,
+      "step": 288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8142479045131956e-05,
+      "loss": 0.1619,
+      "step": 289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812801405323497e-05,
+      "loss": 0.1592,
+      "step": 290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811349877184395e-05,
+      "loss": 0.1606,
+      "step": 291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809893329076757e-05,
+      "loss": 0.1774,
+      "step": 292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808431770012509e-05,
+      "loss": 0.1843,
+      "step": 293
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069652090345825e-05,
+      "loss": 0.1843,
+      "step": 294
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054936552168548e-05,
+      "loss": 0.2021,
+      "step": 295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040171176640967e-05,
+      "loss": 0.1893,
+      "step": 296
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8025356055119122e-05,
+      "loss": 0.1358,
+      "step": 297
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801049127926686e-05,
+      "loss": 0.1947,
+      "step": 298
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7995576941055236e-05,
+      "loss": 0.1525,
+      "step": 299
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7980613132761956e-05,
+      "loss": 0.1676,
+      "step": 300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7965599946970814e-05,
+      "loss": 0.1277,
+      "step": 301
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795053747657111e-05,
+      "loss": 0.152,
+      "step": 302
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793542581475707e-05,
+      "loss": 0.1774,
+      "step": 303
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7920265055027285e-05,
+      "loss": 0.1532,
+      "step": 304
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790505529118412e-05,
+      "loss": 0.1925,
+      "step": 305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788979661733313e-05,
+      "loss": 0.1935,
+      "step": 306
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7874489127882496e-05,
+      "loss": 0.1649,
+      "step": 307
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7859132917542427e-05,
+      "loss": 0.1666,
+      "step": 308
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7843728081324575e-05,
+      "loss": 0.1657,
+      "step": 309
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7828274714541445e-05,
+      "loss": 0.1985,
+      "step": 310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781277291280582e-05,
+      "loss": 0.1566,
+      "step": 311
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7797222772030155e-05,
+      "loss": 0.1545,
+      "step": 312
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7781624388425974e-05,
+      "loss": 0.1806,
+      "step": 313
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7765977858503314e-05,
+      "loss": 0.1703,
+      "step": 314
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750283279070077e-05,
+      "loss": 0.1627,
+      "step": 315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773454074723147e-05,
+      "loss": 0.188,
+      "step": 316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771875036038939e-05,
+      "loss": 0.1873,
+      "step": 317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770291221624181e-05,
+      "loss": 0.1747,
+      "step": 318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.76870264127822e-05,
+      "loss": 0.1897,
+      "step": 319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7671093048298905e-05,
+      "loss": 0.1751,
+      "step": 320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765511222137453e-05,
+      "loss": 0.1655,
+      "step": 321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763908403088534e-05,
+      "loss": 0.1514,
+      "step": 322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762300857600066e-05,
+      "loss": 0.1576,
+      "step": 323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7606885956182228e-05,
+      "loss": 0.1656,
+      "step": 324
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759071627118362e-05,
+      "loss": 0.1844,
+      "step": 325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75744996210496e-05,
+      "loss": 0.1579,
+      "step": 326
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7558236106115514e-05,
+      "loss": 0.1735,
+      "step": 327
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754192582700668e-05,
+      "loss": 0.159,
+      "step": 328
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752556888463774e-05,
+      "loss": 0.1721,
+      "step": 329
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750916538021206e-05,
+      "loss": 0.1626,
+      "step": 330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492715415221087e-05,
+      "loss": 0.1454,
+      "step": 331
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7476219091443738e-05,
+      "loss": 0.1846,
+      "step": 332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745967651094575e-05,
+      "loss": 0.17,
+      "step": 333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7443087776079068e-05,
+      "loss": 0.1989,
+      "step": 334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7426452989481187e-05,
+      "loss": 0.1668,
+      "step": 335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740977225407455e-05,
+      "loss": 0.15,
+      "step": 336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739304567306588e-05,
+      "loss": 0.186,
+      "step": 337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7376273349945568e-05,
+      "loss": 0.2104,
+      "step": 338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7359455388487008e-05,
+      "loss": 0.1669,
+      "step": 339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7342591892745978e-05,
+      "loss": 0.1796,
+      "step": 340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732568296705997e-05,
+      "loss": 0.1844,
+      "step": 341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7308728716047568e-05,
+      "loss": 0.2021,
+      "step": 342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72917292446078e-05,
+      "loss": 0.1658,
+      "step": 343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727468465791946e-05,
+      "loss": 0.178,
+      "step": 344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7257595061440502e-05,
+      "loss": 0.1884,
+      "step": 345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7240460560907345e-05,
+      "loss": 0.1221,
+      "step": 346
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7223281262334252e-05,
+      "loss": 0.1917,
+      "step": 347
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7206057272012647e-05,
+      "loss": 0.2018,
+      "step": 348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7188788696510477e-05,
+      "loss": 0.1758,
+      "step": 349
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7171475642671546e-05,
+      "loss": 0.1558,
+      "step": 350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7154118217614844e-05,
+      "loss": 0.1481,
+      "step": 351
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7136716528733916e-05,
+      "loss": 0.1893,
+      "step": 352
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7119270683696144e-05,
+      "loss": 0.1838,
+      "step": 353
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710178079044214e-05,
+      "loss": 0.1898,
+      "step": 354
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7084246957185036e-05,
+      "loss": 0.1707,
+      "step": 355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706666929240983e-05,
+      "loss": 0.1949,
+      "step": 356
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7049047904872715e-05,
+      "loss": 0.1887,
+      "step": 357
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703138290360041e-05,
+      "loss": 0.1765,
+      "step": 358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701367439788947e-05,
+      "loss": 0.1746,
+      "step": 359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699592249730563e-05,
+      "loss": 0.1614,
+      "step": 360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6978127311683103e-05,
+      "loss": 0.2043,
+      "step": 361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960288951123933e-05,
+      "loss": 0.1299,
+      "step": 362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6942407525997277e-05,
+      "loss": 0.1686,
+      "step": 363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6924483146938756e-05,
+      "loss": 0.1363,
+      "step": 364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6906515924849745e-05,
+      "loss": 0.1557,
+      "step": 365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6888505970896696e-05,
+      "loss": 0.1445,
+      "step": 366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6870453396510456e-05,
+      "loss": 0.1707,
+      "step": 367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6852358313385568e-05,
+      "loss": 0.1439,
+      "step": 368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683422083347959e-05,
+      "loss": 0.1688,
+      "step": 369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681604106901239e-05,
+      "loss": 0.1727,
+      "step": 370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6797819132465462e-05,
+      "loss": 0.1892,
+      "step": 371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6779555136581228e-05,
+      "loss": 0.1981,
+      "step": 372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676124919436233e-05,
+      "loss": 0.1728,
+      "step": 373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674290141907095e-05,
+      "loss": 0.179,
+      "step": 374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6724511924228093e-05,
+      "loss": 0.1432,
+      "step": 375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6706080823612897e-05,
+      "loss": 0.1774,
+      "step": 376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668760823126191e-05,
+      "loss": 0.1309,
+      "step": 377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6669094261468405e-05,
+      "loss": 0.1511,
+      "step": 378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665053902878167e-05,
+      "loss": 0.1712,
+      "step": 379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6631942648006284e-05,
+      "loss": 0.1736,
+      "step": 380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661330523420142e-05,
+      "loss": 0.1523,
+      "step": 381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659462690268013e-05,
+      "loss": 0.1563,
+      "step": 382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6575907769008633e-05,
+      "loss": 0.1881,
+      "step": 383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6557147949005597e-05,
+      "loss": 0.139,
+      "step": 384
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6538347558741424e-05,
+      "loss": 0.1664,
+      "step": 385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651950671453753e-05,
+      "loss": 0.1446,
+      "step": 386
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650062553296563e-05,
+      "loss": 0.1598,
+      "step": 387
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6481704130847013e-05,
+      "loss": 0.1639,
+      "step": 388
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646274262525181e-05,
+      "loss": 0.1521,
+      "step": 389
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6443741133498297e-05,
+      "loss": 0.1823,
+      "step": 390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642469977315214e-05,
+      "loss": 0.143,
+      "step": 391
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6405618662025678e-05,
+      "loss": 0.1797,
+      "step": 392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6386497918177208e-05,
+      "loss": 0.1504,
+      "step": 393
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6367337659910223e-05,
+      "loss": 0.1612,
+      "step": 394
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348138005772717e-05,
+      "loss": 0.1428,
+      "step": 395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6328899074556415e-05,
+      "loss": 0.1461,
+      "step": 396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6309620985296075e-05,
+      "loss": 0.1414,
+      "step": 397
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6290303857268724e-05,
+      "loss": 0.1598,
+      "step": 398
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6270947809992924e-05,
+      "loss": 0.1652,
+      "step": 399
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625155296322805e-05,
+      "loss": 0.1763,
+      "step": 400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6232119436973525e-05,
+      "loss": 0.1559,
+      "step": 401
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6212647351468088e-05,
+      "loss": 0.158,
+      "step": 402
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6193136827189067e-05,
+      "loss": 0.1583,
+      "step": 403
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6173587984851596e-05,
+      "loss": 0.1823,
+      "step": 404
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6154000945407914e-05,
+      "loss": 0.165,
+      "step": 405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6134375830046566e-05,
+      "loss": 0.1606,
+      "step": 406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6114712760191697e-05,
+      "loss": 0.1712,
+      "step": 407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6095011857502282e-05,
+      "loss": 0.2072,
+      "step": 408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607527324387137e-05,
+      "loss": 0.1614,
+      "step": 409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6055497041425333e-05,
+      "loss": 0.1514,
+      "step": 410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603568337252312e-05,
+      "loss": 0.1722,
+      "step": 411
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6015832359755483e-05,
+      "loss": 0.1444,
+      "step": 412
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5995944125944235e-05,
+      "loss": 0.137,
+      "step": 413
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597601879414147e-05,
+      "loss": 0.1322,
+      "step": 414
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5956056487628832e-05,
+      "loss": 0.1699,
+      "step": 415
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5936057329916718e-05,
+      "loss": 0.1318,
+      "step": 416
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5916021444743535e-05,
+      "loss": 0.1461,
+      "step": 417
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5895948956074937e-05,
+      "loss": 0.1544,
+      "step": 418
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587583998810304e-05,
+      "loss": 0.2183,
+      "step": 419
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5855694665245665e-05,
+      "loss": 0.1257,
+      "step": 420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5835513112145583e-05,
+      "loss": 0.1715,
+      "step": 421
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5815295453669702e-05,
+      "loss": 0.1283,
+      "step": 422
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579504181490834e-05,
+      "loss": 0.1766,
+      "step": 423
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5774752321174428e-05,
+      "loss": 0.1359,
+      "step": 424
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5754427098002735e-05,
+      "loss": 0.1571,
+      "step": 425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573406627114909e-05,
+      "loss": 0.16,
+      "step": 426
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571366996658962e-05,
+      "loss": 0.199,
+      "step": 427
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693238310519952e-05,
+      "loss": 0.1494,
+      "step": 428
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5672771429354436e-05,
+      "loss": 0.1779,
+      "step": 429
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5652269449725375e-05,
+      "loss": 0.1879,
+      "step": 430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5631732498482222e-05,
+      "loss": 0.1899,
+      "step": 431
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561116070269082e-05,
+      "loss": 0.1733,
+      "step": 432
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5590554189632585e-05,
+      "loss": 0.132,
+      "step": 433
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5569913086803744e-05,
+      "loss": 0.1451,
+      "step": 434
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5549237521914537e-05,
+      "loss": 0.1439,
+      "step": 435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552852762288843e-05,
+      "loss": 0.1458,
+      "step": 436
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5507783517861308e-05,
+      "loss": 0.1711,
+      "step": 437
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548700533518071e-05,
+      "loss": 0.1427,
+      "step": 438
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5466193203405017e-05,
+      "loss": 0.1492,
+      "step": 439
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5445347251302642e-05,
+      "loss": 0.1391,
+      "step": 440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5424467607851274e-05,
+      "loss": 0.1373,
+      "step": 441
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540355440223704e-05,
+      "loss": 0.1694,
+      "step": 442
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5382607763853736e-05,
+      "loss": 0.1696,
+      "step": 443
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5361627822301997e-05,
+      "loss": 0.1806,
+      "step": 444
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534061470738852e-05,
+      "loss": 0.1771,
+      "step": 445
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5319568549125253e-05,
+      "loss": 0.1438,
+      "step": 446
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5298489477728587e-05,
+      "loss": 0.1574,
+      "step": 447
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527737762361855e-05,
+      "loss": 0.1293,
+      "step": 448
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5256233117418e-05,
+      "loss": 0.1812,
+      "step": 449
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.523505608995183e-05,
+      "loss": 0.1493,
+      "step": 450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5213846672246139e-05,
+      "loss": 0.1402,
+      "step": 451
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5192604995527433e-05,
+      "loss": 0.1629,
+      "step": 452
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5171331191221814e-05,
+      "loss": 0.1923,
+      "step": 453
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5150025390954153e-05,
+      "loss": 0.1578,
+      "step": 454
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5128687726547297e-05,
+      "loss": 0.1849,
+      "step": 455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5107318330021237e-05,
+      "loss": 0.1554,
+      "step": 456
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5085917333592299e-05,
+      "loss": 0.1608,
+      "step": 457
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064484869672318e-05,
+      "loss": 0.1281,
+      "step": 458
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043021070867825e-05,
+      "loss": 0.2036,
+      "step": 459
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5021526069979232e-05,
+      "loss": 0.1752,
+      "step": 460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.1619,
+      "step": 461
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4978442994115821e-05,
+      "loss": 0.158,
+      "step": 462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4956855185703787e-05,
+      "loss": 0.1711,
+      "step": 463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4935236708331582e-05,
+      "loss": 0.174,
+      "step": 464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913587695756632e-05,
+      "loss": 0.1359,
+      "step": 465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48919082819253e-05,
+      "loss": 0.1531,
+      "step": 466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4870198600972043e-05,
+      "loss": 0.1546,
+      "step": 467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4848458787218583e-05,
+      "loss": 0.1509,
+      "step": 468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4826688975173085e-05,
+      "loss": 0.1888,
+      "step": 469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480488929952932e-05,
+      "loss": 0.1388,
+      "step": 470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4783059895165818e-05,
+      "loss": 0.1703,
+      "step": 471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4761200897145063e-05,
+      "loss": 0.175,
+      "step": 472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4739312440712627e-05,
+      "loss": 0.1566,
+      "step": 473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4717394661296353e-05,
+      "loss": 0.1714,
+      "step": 474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4695447694505512e-05,
+      "loss": 0.2009,
+      "step": 475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.467347167612995e-05,
+      "loss": 0.1402,
+      "step": 476
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4651466742139284e-05,
+      "loss": 0.1637,
+      "step": 477
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4629433028682014e-05,
+      "loss": 0.1657,
+      "step": 478
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4607370672084717e-05,
+      "loss": 0.1577,
+      "step": 479
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458527980885118e-05,
+      "loss": 0.1331,
+      "step": 480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456316057566158e-05,
+      "loss": 0.1654,
+      "step": 481
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454101310937161e-05,
+      "loss": 0.1318,
+      "step": 482
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4518837547011657e-05,
+      "loss": 0.1051,
+      "step": 483
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4496634025785938e-05,
+      "loss": 0.1705,
+      "step": 484
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4474402683071659e-05,
+      "loss": 0.1568,
+      "step": 485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452143656418157e-05,
+      "loss": 0.1725,
+      "step": 486
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4429857083546054e-05,
+      "loss": 0.1793,
+      "step": 487
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4407543102346423e-05,
+      "loss": 0.1368,
+      "step": 488
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4385201850879895e-05,
+      "loss": 0.199,
+      "step": 489
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4362833467375839e-05,
+      "loss": 0.1596,
+      "step": 490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4340438090231492e-05,
+      "loss": 0.1677,
+      "step": 491
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4318015858011108e-05,
+      "loss": 0.1423,
+      "step": 492
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429556690944509e-05,
+      "loss": 0.1628,
+      "step": 493
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427309138342915e-05,
+      "loss": 0.1359,
+      "step": 494
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4250589419023435e-05,
+      "loss": 0.1379,
+      "step": 495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4228061155451671e-05,
+      "loss": 0.1494,
+      "step": 496
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4205506732100303e-05,
+      "loss": 0.188,
+      "step": 497
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4182926288517628e-05,
+      "loss": 0.1556,
+      "step": 498
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4160319964412943e-05,
+      "loss": 0.1462,
+      "step": 499
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137687899655662e-05,
+      "loss": 0.1661,
+      "step": 500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4115030234274461e-05,
+      "loss": 0.1377,
+      "step": 501
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4092347108456425e-05,
+      "loss": 0.1414,
+      "step": 502
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4069638662546158e-05,
+      "loss": 0.1763,
+      "step": 503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404690503704492e-05,
+      "loss": 0.1616,
+      "step": 504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402414637260977e-05,
+      "loss": 0.1792,
+      "step": 505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4001362810052684e-05,
+      "loss": 0.1776,
+      "step": 506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3978554490339687e-05,
+      "loss": 0.1617,
+      "step": 507
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3955721554589979e-05,
+      "loss": 0.1276,
+      "step": 508
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3932864144075064e-05,
+      "loss": 0.1528,
+      "step": 509
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3909982400217885e-05,
+      "loss": 0.1324,
+      "step": 510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388707646459193e-05,
+      "loss": 0.1289,
+      "step": 511
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3864146478920366e-05,
+      "loss": 0.1123,
+      "step": 512
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3841192585075164e-05,
+      "loss": 0.1663,
+      "step": 513
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3818214925076226e-05,
+      "loss": 0.1562,
+      "step": 514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3795213641090483e-05,
+      "loss": 0.1518,
+      "step": 515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3772188875431054e-05,
+      "loss": 0.1689,
+      "step": 516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3749140770556322e-05,
+      "loss": 0.1257,
+      "step": 517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726069469069083e-05,
+      "loss": 0.1672,
+      "step": 518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3702975113715661e-05,
+      "loss": 0.1568,
+      "step": 519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.367985784738501e-05,
+      "loss": 0.1649,
+      "step": 520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365671781310784e-05,
+      "loss": 0.1476,
+      "step": 521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3633555154055733e-05,
+      "loss": 0.16,
+      "step": 522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361037001354025e-05,
+      "loss": 0.1575,
+      "step": 523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3587162535012051e-05,
+      "loss": 0.1755,
+      "step": 524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356393286206001e-05,
+      "loss": 0.1418,
+      "step": 525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3540681138410317e-05,
+      "loss": 0.1648,
+      "step": 526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3517407507925593e-05,
+      "loss": 0.1678,
+      "step": 527
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3494112114604001e-05,
+      "loss": 0.1302,
+      "step": 528
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3470795102578358e-05,
+      "loss": 0.1494,
+      "step": 529
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3447456616115238e-05,
+      "loss": 0.15,
+      "step": 530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342409679961408e-05,
+      "loss": 0.1729,
+      "step": 531
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3400715797606293e-05,
+      "loss": 0.1377,
+      "step": 532
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3377313754754373e-05,
+      "loss": 0.1398,
+      "step": 533
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3353890815850993e-05,
+      "loss": 0.1501,
+      "step": 534
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3330447125818115e-05,
+      "loss": 0.1294,
+      "step": 535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3306982829706093e-05,
+      "loss": 0.1425,
+      "step": 536
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3283498072692771e-05,
+      "loss": 0.1606,
+      "step": 537
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259993000082599e-05,
+      "loss": 0.1425,
+      "step": 538
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3236467757305704e-05,
+      "loss": 0.1227,
+      "step": 539
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321292248991703e-05,
+      "loss": 0.1601,
+      "step": 540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3189357343595405e-05,
+      "loss": 0.1459,
+      "step": 541
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3165772464142654e-05,
+      "loss": 0.1351,
+      "step": 542
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3142167997482693e-05,
+      "loss": 0.1617,
+      "step": 543
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3118544089660635e-05,
+      "loss": 0.1509,
+      "step": 544
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3094900886841867e-05,
+      "loss": 0.1467,
+      "step": 545
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3071238535311164e-05,
+      "loss": 0.1607,
+      "step": 546
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3047557181471784e-05,
+      "loss": 0.1551,
+      "step": 547
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3023856971844543e-05,
+      "loss": 0.1519,
+      "step": 548
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3000138053066931e-05,
+      "loss": 0.1434,
+      "step": 549
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2976400571892189e-05,
+      "loss": 0.1422,
+      "step": 550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2952644675188413e-05,
+      "loss": 0.1603,
+      "step": 551
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2928870509937633e-05,
+      "loss": 0.1939,
+      "step": 552
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2905078223234907e-05,
+      "loss": 0.1395,
+      "step": 553
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.288126796228743e-05,
+      "loss": 0.1453,
+      "step": 554
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2857439874413589e-05,
+      "loss": 0.1609,
+      "step": 555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2833594107042078e-05,
+      "loss": 0.1581,
+      "step": 556
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2809730807710973e-05,
+      "loss": 0.1776,
+      "step": 557
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2785850124066829e-05,
+      "loss": 0.1639,
+      "step": 558
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2761952203863759e-05,
+      "loss": 0.1727,
+      "step": 559
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738037194962516e-05,
+      "loss": 0.1725,
+      "step": 560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2714105245329591e-05,
+      "loss": 0.1625,
+      "step": 561
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2690156503036288e-05,
+      "loss": 0.1178,
+      "step": 562
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266619111625781e-05,
+      "loss": 0.1387,
+      "step": 563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264220923327234e-05,
+      "loss": 0.1333,
+      "step": 564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2618211002460135e-05,
+      "loss": 0.1608,
+      "step": 565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2594196572302586e-05,
+      "loss": 0.1705,
+      "step": 566
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.257016609138132e-05,
+      "loss": 0.1398,
+      "step": 567
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2546119708377273e-05,
+      "loss": 0.1556,
+      "step": 568
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2522057572069772e-05,
+      "loss": 0.1579,
+      "step": 569
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2497979831335604e-05,
+      "loss": 0.1472,
+      "step": 570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2473886635148109e-05,
+      "loss": 0.1385,
+      "step": 571
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2449778132576256e-05,
+      "loss": 0.1447,
+      "step": 572
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2425654472783712e-05,
+      "loss": 0.1469,
+      "step": 573
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2401515805027924e-05,
+      "loss": 0.1749,
+      "step": 574
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2377362278659202e-05,
+      "loss": 0.1406,
+      "step": 575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2353194043119783e-05,
+      "loss": 0.1743,
+      "step": 576
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2329011247942915e-05,
+      "loss": 0.1695,
+      "step": 577
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2304814042751927e-05,
+      "loss": 0.1807,
+      "step": 578
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2280602577259312e-05,
+      "loss": 0.1356,
+      "step": 579
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2256377001265785e-05,
+      "loss": 0.1706,
+      "step": 580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2232137464659368e-05,
+      "loss": 0.1539,
+      "step": 581
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2207884117414465e-05,
+      "loss": 0.1608,
+      "step": 582
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2183617109590925e-05,
+      "loss": 0.1359,
+      "step": 583
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215933659133311e-05,
+      "loss": 0.142,
+      "step": 584
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213504271286899e-05,
+      "loss": 0.1407,
+      "step": 585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2110735624509184e-05,
+      "loss": 0.175,
+      "step": 586
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2086415476646047e-05,
+      "loss": 0.1622,
+      "step": 587
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2062082419752737e-05,
+      "loss": 0.1487,
+      "step": 588
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2037736604382279e-05,
+      "loss": 0.1431,
+      "step": 589
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201337818116664e-05,
+      "loss": 0.1998,
+      "step": 590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1989007300815793e-05,
+      "loss": 0.1755,
+      "step": 591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1964624114116784e-05,
+      "loss": 0.1401,
+      "step": 592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1940228771932808e-05,
+      "loss": 0.1581,
+      "step": 593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1915821425202258e-05,
+      "loss": 0.1665,
+      "step": 594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1891402224937805e-05,
+      "loss": 0.1521,
+      "step": 595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1866971322225462e-05,
+      "loss": 0.149,
+      "step": 596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1842528868223645e-05,
+      "loss": 0.1561,
+      "step": 597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1818075014162243e-05,
+      "loss": 0.1748,
+      "step": 598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1793609911341672e-05,
+      "loss": 0.1284,
+      "step": 599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176913371113196e-05,
+      "loss": 0.16,
+      "step": 600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1744646564971777e-05,
+      "loss": 0.15,
+      "step": 601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1720148624367533e-05,
+      "loss": 0.1507,
+      "step": 602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1695640040892424e-05,
+      "loss": 0.1599,
+      "step": 603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1671120966185486e-05,
+      "loss": 0.1293,
+      "step": 604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1646591551950677e-05,
+      "loss": 0.1552,
+      "step": 605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1622051949955915e-05,
+      "loss": 0.1371,
+      "step": 606
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159750231203217e-05,
+      "loss": 0.1745,
+      "step": 607
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1572942790072486e-05,
+      "loss": 0.1589,
+      "step": 608
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154837353603107e-05,
+      "loss": 0.1422,
+      "step": 609
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1523794701922351e-05,
+      "loss": 0.1644,
+      "step": 610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149920643982002e-05,
+      "loss": 0.1649,
+      "step": 611
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1474608901856103e-05,
+      "loss": 0.1237,
+      "step": 612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.145000224022002e-05,
+      "loss": 0.1696,
+      "step": 613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1425386607157642e-05,
+      "loss": 0.1442,
+      "step": 614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1400762154970346e-05,
+      "loss": 0.1315,
+      "step": 615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1376129036014073e-05,
+      "loss": 0.1257,
+      "step": 616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1351487402698388e-05,
+      "loss": 0.1306,
+      "step": 617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1326837407485537e-05,
+      "loss": 0.1968,
+      "step": 618
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1302179202889505e-05,
+      "loss": 0.1537,
+      "step": 619
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1277512941475059e-05,
+      "loss": 0.1595,
+      "step": 620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1252838775856829e-05,
+      "loss": 0.1656,
+      "step": 621
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1228156858698344e-05,
+      "loss": 0.1618,
+      "step": 622
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1203467342711087e-05,
+      "loss": 0.1572,
+      "step": 623
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1178770380653567e-05,
+      "loss": 0.1497,
+      "step": 624
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1154066125330358e-05,
+      "loss": 0.1502,
+      "step": 625
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1129354729591158e-05,
+      "loss": 0.1497,
+      "step": 626
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1104636346329838e-05,
+      "loss": 0.1498,
+      "step": 627
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107991112848352e-05,
+      "loss": 0.1829,
+      "step": 628
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055179229031598e-05,
+      "loss": 0.1269,
+      "step": 629
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1030440800994806e-05,
+      "loss": 0.177,
+      "step": 630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.100569599743428e-05,
+      "loss": 0.1429,
+      "step": 631
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.09809449714506e-05,
+      "loss": 0.1397,
+      "step": 632
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0956187876182837e-05,
+      "loss": 0.1304,
+      "step": 633
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0931424864807624e-05,
+      "loss": 0.1517,
+      "step": 634
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0906656090538196e-05,
+      "loss": 0.1763,
+      "step": 635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0881881706623443e-05,
+      "loss": 0.1377,
+      "step": 636
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0857101866346953e-05,
+      "loss": 0.1616,
+      "step": 637
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0832316723026092e-05,
+      "loss": 0.1833,
+      "step": 638
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0807526430011028e-05,
+      "loss": 0.1581,
+      "step": 639
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0782731140683786e-05,
+      "loss": 0.1369,
+      "step": 640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0757931008457308e-05,
+      "loss": 0.1349,
+      "step": 641
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0733126186774504e-05,
+      "loss": 0.1469,
+      "step": 642
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0708316829107295e-05,
+      "loss": 0.1632,
+      "step": 643
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0683503088955663e-05,
+      "loss": 0.1455,
+      "step": 644
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0658685119846713e-05,
+      "loss": 0.178,
+      "step": 645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0633863075333713e-05,
+      "loss": 0.1727,
+      "step": 646
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0609037108995143e-05,
+      "loss": 0.1544,
+      "step": 647
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058420737443375e-05,
+      "loss": 0.1403,
+      "step": 648
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0559374025275597e-05,
+      "loss": 0.1364,
+      "step": 649
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0534537215169106e-05,
+      "loss": 0.1556,
+      "step": 650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0509697097784122e-05,
+      "loss": 0.1201,
+      "step": 651
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0484853826810942e-05,
+      "loss": 0.1265,
+      "step": 652
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0460007555959384e-05,
+      "loss": 0.1558,
+      "step": 653
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043515843895782e-05,
+      "loss": 0.1304,
+      "step": 654
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0410306629552231e-05,
+      "loss": 0.1529,
+      "step": 655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0385452281505269e-05,
+      "loss": 0.149,
+      "step": 656
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0360595548595274e-05,
+      "loss": 0.14,
+      "step": 657
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0335736584615357e-05,
+      "loss": 0.1418,
+      "step": 658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0310875543372425e-05,
+      "loss": 0.1474,
+      "step": 659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0286012578686244e-05,
+      "loss": 0.2043,
+      "step": 660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0261147844388472e-05,
+      "loss": 0.1749,
+      "step": 661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023628149432172e-05,
+      "loss": 0.1605,
+      "step": 662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0211413682338603e-05,
+      "loss": 0.1263,
+      "step": 663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0186544562300766e-05,
+      "loss": 0.129,
+      "step": 664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0161674288077959e-05,
+      "loss": 0.1573,
+      "step": 665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0136803013547071e-05,
+      "loss": 0.158,
+      "step": 666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.011193089259118e-05,
+      "loss": 0.1378,
+      "step": 667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0087058079098595e-05,
+      "loss": 0.1514,
+      "step": 668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062184726961918e-05,
+      "loss": 0.1923,
+      "step": 669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0037310990077083e-05,
+      "loss": 0.1526,
+      "step": 670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0012437022342399e-05,
+      "loss": 0.1386,
+      "step": 671
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987562977657605e-06,
+      "loss": 0.1846,
+      "step": 672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.962689009922918e-06,
+      "loss": 0.1565,
+      "step": 673
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937815273038082e-06,
+      "loss": 0.1765,
+      "step": 674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912941920901408e-06,
+      "loss": 0.1717,
+      "step": 675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.888069107408824e-06,
+      "loss": 0.1339,
+      "step": 676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86319698645293e-06,
+      "loss": 0.1597,
+      "step": 677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.838325711922045e-06,
+      "loss": 0.1493,
+      "step": 678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.813455437699238e-06,
+      "loss": 0.1155,
+      "step": 679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.788586317661404e-06,
+      "loss": 0.1222,
+      "step": 680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.763718505678282e-06,
+      "loss": 0.162,
+      "step": 681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.738852155611531e-06,
+      "loss": 0.1505,
+      "step": 682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.71398742131376e-06,
+      "loss": 0.1562,
+      "step": 683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.689124456627577e-06,
+      "loss": 0.1824,
+      "step": 684
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.664263415384644e-06,
+      "loss": 0.1526,
+      "step": 685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.639404451404729e-06,
+      "loss": 0.1288,
+      "step": 686
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.614547718494736e-06,
+      "loss": 0.1289,
+      "step": 687
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.589693370447769e-06,
+      "loss": 0.149,
+      "step": 688
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.564841561042187e-06,
+      "loss": 0.1466,
+      "step": 689
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53999244404062e-06,
+      "loss": 0.1608,
+      "step": 690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.515146173189058e-06,
+      "loss": 0.1835,
+      "step": 691
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.490302902215882e-06,
+      "loss": 0.1229,
+      "step": 692
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.465462784830895e-06,
+      "loss": 0.1499,
+      "step": 693
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.440625974724408e-06,
+      "loss": 0.1361,
+      "step": 694
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.415792625566253e-06,
+      "loss": 0.1902,
+      "step": 695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.390962891004858e-06,
+      "loss": 0.1227,
+      "step": 696
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.366136924666289e-06,
+      "loss": 0.1444,
+      "step": 697
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34131488015329e-06,
+      "loss": 0.1432,
+      "step": 698
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31649691104434e-06,
+      "loss": 0.1489,
+      "step": 699
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.291683170892712e-06,
+      "loss": 0.1693,
+      "step": 700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.266873813225498e-06,
+      "loss": 0.17,
+      "step": 701
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.242068991542694e-06,
+      "loss": 0.14,
+      "step": 702
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.217268859316219e-06,
+      "loss": 0.1736,
+      "step": 703
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.192473569988975e-06,
+      "loss": 0.1615,
+      "step": 704
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.167683276973906e-06,
+      "loss": 0.1528,
+      "step": 705
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.142898133653049e-06,
+      "loss": 0.1623,
+      "step": 706
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11811829337656e-06,
+      "loss": 0.1485,
+      "step": 707
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.093343909461807e-06,
+      "loss": 0.1675,
+      "step": 708
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.068575135192377e-06,
+      "loss": 0.1639,
+      "step": 709
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.043812123817165e-06,
+      "loss": 0.1341,
+      "step": 710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.019055028549403e-06,
+      "loss": 0.1243,
+      "step": 711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.994304002565723e-06,
+      "loss": 0.1823,
+      "step": 712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.969559199005196e-06,
+      "loss": 0.1382,
+      "step": 713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.944820770968407e-06,
+      "loss": 0.151,
+      "step": 714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.920088871516482e-06,
+      "loss": 0.1578,
+      "step": 715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.895363653670162e-06,
+      "loss": 0.1764,
+      "step": 716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.870645270408849e-06,
+      "loss": 0.1588,
+      "step": 717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.845933874669645e-06,
+      "loss": 0.1785,
+      "step": 718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.821229619346433e-06,
+      "loss": 0.1792,
+      "step": 719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.796532657288916e-06,
+      "loss": 0.1313,
+      "step": 720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771843141301659e-06,
+      "loss": 0.1253,
+      "step": 721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.747161224143175e-06,
+      "loss": 0.1489,
+      "step": 722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.722487058524945e-06,
+      "loss": 0.1428,
+      "step": 723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.697820797110499e-06,
+      "loss": 0.1637,
+      "step": 724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.673162592514466e-06,
+      "loss": 0.162,
+      "step": 725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.648512597301613e-06,
+      "loss": 0.1533,
+      "step": 726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62387096398593e-06,
+      "loss": 0.0983,
+      "step": 727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.599237845029657e-06,
+      "loss": 0.1464,
+      "step": 728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.574613392842361e-06,
+      "loss": 0.1644,
+      "step": 729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.549997759779981e-06,
+      "loss": 0.1322,
+      "step": 730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.525391098143902e-06,
+      "loss": 0.1486,
+      "step": 731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.500793560179984e-06,
+      "loss": 0.1199,
+      "step": 732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.47620529807765e-06,
+      "loss": 0.157,
+      "step": 733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.451626463968932e-06,
+      "loss": 0.1303,
+      "step": 734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.427057209927519e-06,
+      "loss": 0.1253,
+      "step": 735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.402497687967837e-06,
+      "loss": 0.121,
+      "step": 736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.377948050044087e-06,
+      "loss": 0.1411,
+      "step": 737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.353408448049327e-06,
+      "loss": 0.1296,
+      "step": 738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.328879033814516e-06,
+      "loss": 0.1353,
+      "step": 739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.304359959107579e-06,
+      "loss": 0.1236,
+      "step": 740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.279851375632469e-06,
+      "loss": 0.1898,
+      "step": 741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.255353435028228e-06,
+      "loss": 0.1373,
+      "step": 742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.230866288868045e-06,
+      "loss": 0.1348,
+      "step": 743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.206390088658326e-06,
+      "loss": 0.1464,
+      "step": 744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.181924985837762e-06,
+      "loss": 0.1536,
+      "step": 745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.157471131776356e-06,
+      "loss": 0.1459,
+      "step": 746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13302867777454e-06,
+      "loss": 0.1177,
+      "step": 747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.108597775062198e-06,
+      "loss": 0.1281,
+      "step": 748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.084178574797745e-06,
+      "loss": 0.1549,
+      "step": 749
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.059771228067197e-06,
+      "loss": 0.1098,
+      "step": 750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.035375885883217e-06,
+      "loss": 0.1377,
+      "step": 751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.010992699184208e-06,
+      "loss": 0.136,
+      "step": 752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.986621818833364e-06,
+      "loss": 0.1519,
+      "step": 753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.962263395617724e-06,
+      "loss": 0.1469,
+      "step": 754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.937917580247265e-06,
+      "loss": 0.1231,
+      "step": 755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.913584523353958e-06,
+      "loss": 0.1518,
+      "step": 756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88926437549082e-06,
+      "loss": 0.1262,
+      "step": 757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.864957287131012e-06,
+      "loss": 0.1462,
+      "step": 758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.840663408666893e-06,
+      "loss": 0.1082,
+      "step": 759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81638289040908e-06,
+      "loss": 0.1541,
+      "step": 760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.792115882585536e-06,
+      "loss": 0.1821,
+      "step": 761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.767862535340635e-06,
+      "loss": 0.1318,
+      "step": 762
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.743622998734217e-06,
+      "loss": 0.1402,
+      "step": 763
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71939742274069e-06,
+      "loss": 0.1638,
+      "step": 764
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.695185957248074e-06,
+      "loss": 0.1342,
+      "step": 765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670988752057088e-06,
+      "loss": 0.131,
+      "step": 766
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.646805956880222e-06,
+      "loss": 0.1409,
+      "step": 767
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.622637721340801e-06,
+      "loss": 0.1612,
+      "step": 768
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.598484194972076e-06,
+      "loss": 0.1199,
+      "step": 769
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.574345527216293e-06,
+      "loss": 0.1552,
+      "step": 770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.550221867423746e-06,
+      "loss": 0.1327,
+      "step": 771
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.526113364851891e-06,
+      "loss": 0.1688,
+      "step": 772
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5020201686644e-06,
+      "loss": 0.138,
+      "step": 773
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.47794242793023e-06,
+      "loss": 0.1579,
+      "step": 774
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.453880291622726e-06,
+      "loss": 0.1556,
+      "step": 775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.429833908618682e-06,
+      "loss": 0.1631,
+      "step": 776
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.405803427697417e-06,
+      "loss": 0.1072,
+      "step": 777
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.381788997539869e-06,
+      "loss": 0.1691,
+      "step": 778
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.357790766727663e-06,
+      "loss": 0.1448,
+      "step": 779
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.333808883742192e-06,
+      "loss": 0.1266,
+      "step": 780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.309843496963715e-06,
+      "loss": 0.1395,
+      "step": 781
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.285894754670413e-06,
+      "loss": 0.1482,
+      "step": 782
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.261962805037486e-06,
+      "loss": 0.1088,
+      "step": 783
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.238047796136247e-06,
+      "loss": 0.1393,
+      "step": 784
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.214149875933173e-06,
+      "loss": 0.1528,
+      "step": 785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.190269192289028e-06,
+      "loss": 0.1342,
+      "step": 786
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.166405892957926e-06,
+      "loss": 0.1461,
+      "step": 787
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.142560125586413e-06,
+      "loss": 0.1223,
+      "step": 788
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.11873203771257e-06,
+      "loss": 0.1385,
+      "step": 789
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.094921776765095e-06,
+      "loss": 0.1687,
+      "step": 790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.071129490062372e-06,
+      "loss": 0.146,
+      "step": 791
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.047355324811591e-06,
+      "loss": 0.1597,
+      "step": 792
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.023599428107815e-06,
+      "loss": 0.1599,
+      "step": 793
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.999861946933072e-06,
+      "loss": 0.1318,
+      "step": 794
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97614302815546e-06,
+      "loss": 0.1526,
+      "step": 795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.95244281852822e-06,
+      "loss": 0.1495,
+      "step": 796
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.928761464688836e-06,
+      "loss": 0.158,
+      "step": 797
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.905099113158138e-06,
+      "loss": 0.1516,
+      "step": 798
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.881455910339369e-06,
+      "loss": 0.1445,
+      "step": 799
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.857832002517307e-06,
+      "loss": 0.1675,
+      "step": 800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.834227535857349e-06,
+      "loss": 0.1384,
+      "step": 801
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8106426564045965e-06,
+      "loss": 0.1508,
+      "step": 802
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.78707751008297e-06,
+      "loss": 0.1315,
+      "step": 803
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.763532242694298e-06,
+      "loss": 0.152,
+      "step": 804
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.740006999917406e-06,
+      "loss": 0.17,
+      "step": 805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.716501927307231e-06,
+      "loss": 0.1817,
+      "step": 806
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.69301717029391e-06,
+      "loss": 0.1542,
+      "step": 807
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.669552874181888e-06,
+      "loss": 0.1599,
+      "step": 808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.646109184149011e-06,
+      "loss": 0.1501,
+      "step": 809
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6226862452456295e-06,
+      "loss": 0.1547,
+      "step": 810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.599284202393709e-06,
+      "loss": 0.1898,
+      "step": 811
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.575903200385927e-06,
+      "loss": 0.1584,
+      "step": 812
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.552543383884766e-06,
+      "loss": 0.1512,
+      "step": 813
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.529204897421644e-06,
+      "loss": 0.153,
+      "step": 814
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.505887885396003e-06,
+      "loss": 0.1422,
+      "step": 815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.482592492074411e-06,
+      "loss": 0.1531,
+      "step": 816
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4593188615896855e-06,
+      "loss": 0.1213,
+      "step": 817
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.436067137939994e-06,
+      "loss": 0.131,
+      "step": 818
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.412837464987951e-06,
+      "loss": 0.1306,
+      "step": 819
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.389629986459756e-06,
+      "loss": 0.1619,
+      "step": 820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.36644484594427e-06,
+      "loss": 0.1307,
+      "step": 821
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.343282186892163e-06,
+      "loss": 0.1339,
+      "step": 822
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3201421526149945e-06,
+      "loss": 0.1501,
+      "step": 823
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.297024886284342e-06,
+      "loss": 0.1694,
+      "step": 824
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273930530930919e-06,
+      "loss": 0.1494,
+      "step": 825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.250859229443684e-06,
+      "loss": 0.1433,
+      "step": 826
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2278111245689495e-06,
+      "loss": 0.1362,
+      "step": 827
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204786358909516e-06,
+      "loss": 0.136,
+      "step": 828
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.181785074923778e-06,
+      "loss": 0.1478,
+      "step": 829
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1588074149248365e-06,
+      "loss": 0.1317,
+      "step": 830
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.135853521079636e-06,
+      "loss": 0.1588,
+      "step": 831
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.112923535408074e-06,
+      "loss": 0.1459,
+      "step": 832
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0900175997821165e-06,
+      "loss": 0.1692,
+      "step": 833
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.067135855924937e-06,
+      "loss": 0.1084,
+      "step": 834
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.044278445410025e-06,
+      "loss": 0.1339,
+      "step": 835
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.021445509660315e-06,
+      "loss": 0.1735,
+      "step": 836
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9986371899473205e-06,
+      "loss": 0.1054,
+      "step": 837
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.975853627390233e-06,
+      "loss": 0.1334,
+      "step": 838
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.953094962955081e-06,
+      "loss": 0.1395,
+      "step": 839
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.930361337453847e-06,
+      "loss": 0.1121,
+      "step": 840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.907652891543576e-06,
+      "loss": 0.153,
+      "step": 841
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.88496976572554e-06,
+      "loss": 0.1368,
+      "step": 842
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.862312100344345e-06,
+      "loss": 0.1368,
+      "step": 843
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.839680035587061e-06,
+      "loss": 0.1158,
+      "step": 844
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8170737114823715e-06,
+      "loss": 0.1136,
+      "step": 845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.794493267899699e-06,
+      "loss": 0.1198,
+      "step": 846
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.771938844548331e-06,
+      "loss": 0.1435,
+      "step": 847
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.749410580976569e-06,
+      "loss": 0.1306,
+      "step": 848
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7269086165708546e-06,
+      "loss": 0.1386,
+      "step": 849
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.704433090554912e-06,
+      "loss": 0.1523,
+      "step": 850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6819841419889e-06,
+      "loss": 0.1416,
+      "step": 851
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.659561909768509e-06,
+      "loss": 0.1427,
+      "step": 852
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6371665326241635e-06,
+      "loss": 0.1183,
+      "step": 853
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.614798149120108e-06,
+      "loss": 0.128,
+      "step": 854
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5924568976535775e-06,
+      "loss": 0.1332,
+      "step": 855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.570142916453944e-06,
+      "loss": 0.1309,
+      "step": 856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.547856343581848e-06,
+      "loss": 0.1473,
+      "step": 857
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.525597316928344e-06,
+      "loss": 0.1506,
+      "step": 858
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.503365974214059e-06,
+      "loss": 0.1292,
+      "step": 859
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.481162452988344e-06,
+      "loss": 0.1241,
+      "step": 860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.458986890628389e-06,
+      "loss": 0.1215,
+      "step": 861
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.436839424338426e-06,
+      "loss": 0.1758,
+      "step": 862
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.414720191148822e-06,
+      "loss": 0.1288,
+      "step": 863
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.392629327915289e-06,
+      "loss": 0.107,
+      "step": 864
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.37056697131799e-06,
+      "loss": 0.1331,
+      "step": 865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3485332578607205e-06,
+      "loss": 0.1093,
+      "step": 866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.326528323870049e-06,
+      "loss": 0.1636,
+      "step": 867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.304552305494493e-06,
+      "loss": 0.1128,
+      "step": 868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2826053387036495e-06,
+      "loss": 0.1729,
+      "step": 869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.260687559287376e-06,
+      "loss": 0.1187,
+      "step": 870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.238799102854941e-06,
+      "loss": 0.158,
+      "step": 871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2169401048341815e-06,
+      "loss": 0.1252,
+      "step": 872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1951107004706825e-06,
+      "loss": 0.1401,
+      "step": 873
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.173311024826916e-06,
+      "loss": 0.1047,
+      "step": 874
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.15154121278142e-06,
+      "loss": 0.1689,
+      "step": 875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.129801399027962e-06,
+      "loss": 0.151,
+      "step": 876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.108091718074706e-06,
+      "loss": 0.1289,
+      "step": 877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08641230424337e-06,
+      "loss": 0.117,
+      "step": 878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064763291668425e-06,
+      "loss": 0.1656,
+      "step": 879
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.043144814296214e-06,
+      "loss": 0.1221,
+      "step": 880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.021557005884183e-06,
+      "loss": 0.1306,
+      "step": 881
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.1522,
+      "step": 882
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9784739300207675e-06,
+      "loss": 0.0971,
+      "step": 883
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.956978929132177e-06,
+      "loss": 0.1313,
+      "step": 884
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935515130327686e-06,
+      "loss": 0.1356,
+      "step": 885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.914082666407705e-06,
+      "loss": 0.1385,
+      "step": 886
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.892681669978762e-06,
+      "loss": 0.1486,
+      "step": 887
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.871312273452706e-06,
+      "loss": 0.1226,
+      "step": 888
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.849974609045849e-06,
+      "loss": 0.1167,
+      "step": 889
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.828668808778194e-06,
+      "loss": 0.1118,
+      "step": 890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.807395004472569e-06,
+      "loss": 0.1064,
+      "step": 891
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.786153327753865e-06,
+      "loss": 0.1519,
+      "step": 892
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764943910048175e-06,
+      "loss": 0.1515,
+      "step": 893
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.743766882582006e-06,
+      "loss": 0.1672,
+      "step": 894
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.722622376381455e-06,
+      "loss": 0.1354,
+      "step": 895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.701510522271416e-06,
+      "loss": 0.1295,
+      "step": 896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.680431450874748e-06,
+      "loss": 0.1541,
+      "step": 897
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659385292611479e-06,
+      "loss": 0.1615,
+      "step": 898
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.638372177698007e-06,
+      "loss": 0.1323,
+      "step": 899
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.617392236146267e-06,
+      "loss": 0.1592,
+      "step": 900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.59644559776296e-06,
+      "loss": 0.1747,
+      "step": 901
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.575532392148729e-06,
+      "loss": 0.1382,
+      "step": 902
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.554652748697361e-06,
+      "loss": 0.1252,
+      "step": 903
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.533806796594989e-06,
+      "loss": 0.1432,
+      "step": 904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5129946648192926e-06,
+      "loss": 0.1538,
+      "step": 905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4922164821386926e-06,
+      "loss": 0.1437,
+      "step": 906
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.471472377111574e-06,
+      "loss": 0.1414,
+      "step": 907
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.450762478085465e-06,
+      "loss": 0.1168,
+      "step": 908
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.43008691319626e-06,
+      "loss": 0.1405,
+      "step": 909
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.409445810367421e-06,
+      "loss": 0.1372,
+      "step": 910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.388839297309182e-06,
+      "loss": 0.134,
+      "step": 911
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368267501517778e-06,
+      "loss": 0.1462,
+      "step": 912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.347730550274628e-06,
+      "loss": 0.1419,
+      "step": 913
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.327228570645566e-06,
+      "loss": 0.124,
+      "step": 914
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306761689480049e-06,
+      "loss": 0.1397,
+      "step": 915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.286330033410384e-06,
+      "loss": 0.1179,
+      "step": 916
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.26593372885091e-06,
+      "loss": 0.1735,
+      "step": 917
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2455729019972715e-06,
+      "loss": 0.1429,
+      "step": 918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2252476788255735e-06,
+      "loss": 0.1198,
+      "step": 919
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.204958185091662e-06,
+      "loss": 0.133,
+      "step": 920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.184704546330302e-06,
+      "loss": 0.1344,
+      "step": 921
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.164486887854424e-06,
+      "loss": 0.1655,
+      "step": 922
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.144305334754335e-06,
+      "loss": 0.1162,
+      "step": 923
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.124160011896965e-06,
+      "loss": 0.1405,
+      "step": 924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104051043925068e-06,
+      "loss": 0.1249,
+      "step": 925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083978555256466e-06,
+      "loss": 0.1201,
+      "step": 926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.063942670083288e-06,
+      "loss": 0.1511,
+      "step": 927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.043943512371171e-06,
+      "loss": 0.0911,
+      "step": 928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.02398120585853e-06,
+      "loss": 0.1402,
+      "step": 929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.004055874055769e-06,
+      "loss": 0.1402,
+      "step": 930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984167640244518e-06,
+      "loss": 0.1791,
+      "step": 931
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964316627476883e-06,
+      "loss": 0.1511,
+      "step": 932
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9445029585746695e-06,
+      "loss": 0.1492,
+      "step": 933
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924726756128632e-06,
+      "loss": 0.1592,
+      "step": 934
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90498814249772e-06,
+      "loss": 0.1299,
+      "step": 935
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.885287239808306e-06,
+      "loss": 0.1415,
+      "step": 936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8656241699534396e-06,
+      "loss": 0.1324,
+      "step": 937
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.845999054592092e-06,
+      "loss": 0.1171,
+      "step": 938
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.826412015148403e-06,
+      "loss": 0.1458,
+      "step": 939
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8068631728109364e-06,
+      "loss": 0.1119,
+      "step": 940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7873526485319135e-06,
+      "loss": 0.1619,
+      "step": 941
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7678805630264802e-06,
+      "loss": 0.1647,
+      "step": 942
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7484470367719493e-06,
+      "loss": 0.1176,
+      "step": 943
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7290521900070774e-06,
+      "loss": 0.1519,
+      "step": 944
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.709696142731277e-06,
+      "loss": 0.1189,
+      "step": 945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6903790147039286e-06,
+      "loss": 0.1123,
+      "step": 946
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.671100925443586e-06,
+      "loss": 0.1374,
+      "step": 947
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6518619942272883e-06,
+      "loss": 0.1244,
+      "step": 948
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6326623400897797e-06,
+      "loss": 0.1542,
+      "step": 949
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.613502081822794e-06,
+      "loss": 0.1219,
+      "step": 950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.594381337974322e-06,
+      "loss": 0.146,
+      "step": 951
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.575300226847863e-06,
+      "loss": 0.1529,
+      "step": 952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5562588665017063e-06,
+      "loss": 0.1328,
+      "step": 953
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5372573747481907e-06,
+      "loss": 0.1257,
+      "step": 954
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5182958691529945e-06,
+      "loss": 0.1434,
+      "step": 955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4993744670343723e-06,
+      "loss": 0.1478,
+      "step": 956
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4804932854624716e-06,
+      "loss": 0.1598,
+      "step": 957
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4616524412585797e-06,
+      "loss": 0.1108,
+      "step": 958
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4428520509944065e-06,
+      "loss": 0.1122,
+      "step": 959
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4240922309913715e-06,
+      "loss": 0.1046,
+      "step": 960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.405373097319875e-06,
+      "loss": 0.1217,
+      "step": 961
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3866947657985827e-06,
+      "loss": 0.1311,
+      "step": 962
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3680573519937187e-06,
+      "loss": 0.151,
+      "step": 963
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3494609712183323e-06,
+      "loss": 0.1498,
+      "step": 964
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3309057385315966e-06,
+      "loss": 0.1071,
+      "step": 965
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.312391768738095e-06,
+      "loss": 0.1489,
+      "step": 966
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2939191763871047e-06,
+      "loss": 0.1398,
+      "step": 967
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2754880757719065e-06,
+      "loss": 0.148,
+      "step": 968
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2570985809290525e-06,
+      "loss": 0.1647,
+      "step": 969
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2387508056376726e-06,
+      "loss": 0.1411,
+      "step": 970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2204448634187734e-06,
+      "loss": 0.1137,
+      "step": 971
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.202180867534541e-06,
+      "loss": 0.1285,
+      "step": 972
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.183958930987612e-06,
+      "loss": 0.122,
+      "step": 973
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.165779166520415e-06,
+      "loss": 0.113,
+      "step": 974
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1476416866144343e-06,
+      "loss": 0.1422,
+      "step": 975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1295466034895482e-06,
+      "loss": 0.1307,
+      "step": 976
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1114940291033092e-06,
+      "loss": 0.1409,
+      "step": 977
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.093484075150257e-06,
+      "loss": 0.1454,
+      "step": 978
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0755168530612444e-06,
+      "loss": 0.1587,
+      "step": 979
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.057592474002723e-06,
+      "loss": 0.1316,
+      "step": 980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0397110488760707e-06,
+      "loss": 0.1523,
+      "step": 981
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.021872688316896e-06,
+      "loss": 0.1177,
+      "step": 982
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0040775026943757e-06,
+      "loss": 0.1644,
+      "step": 983
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.98632560211053e-06,
+      "loss": 0.1648,
+      "step": 984
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.968617096399592e-06,
+      "loss": 0.1328,
+      "step": 985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9509520951272864e-06,
+      "loss": 0.1494,
+      "step": 986
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.933330707590174e-06,
+      "loss": 0.1299,
+      "step": 987
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9157530428149683e-06,
+      "loss": 0.1473,
+      "step": 988
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.898219209557864e-06,
+      "loss": 0.134,
+      "step": 989
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8807293163038572e-06,
+      "loss": 0.1327,
+      "step": 990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8632834712660882e-06,
+      "loss": 0.1027,
+      "step": 991
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.845881782385156e-06,
+      "loss": 0.1169,
+      "step": 992
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.828524357328456e-06,
+      "loss": 0.119,
+      "step": 993
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8112113034895273e-06,
+      "loss": 0.1217,
+      "step": 994
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7939427279873543e-06,
+      "loss": 0.1499,
+      "step": 995
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7767187376657502e-06,
+      "loss": 0.1459,
+      "step": 996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7595394390926557e-06,
+      "loss": 0.1046,
+      "step": 997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.742404938559502e-06,
+      "loss": 0.1093,
+      "step": 998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.72531534208054e-06,
+      "loss": 0.1437,
+      "step": 999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.708270755392207e-06,
+      "loss": 0.1722,
+      "step": 1000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6912712839524336e-06,
+      "loss": 0.153,
+      "step": 1001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.674317032940035e-06,
+      "loss": 0.0967,
+      "step": 1002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.657408107254027e-06,
+      "loss": 0.1166,
+      "step": 1003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.640544611512993e-06,
+      "loss": 0.1004,
+      "step": 1004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6237266500544344e-06,
+      "loss": 0.1589,
+      "step": 1005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.60695432693412e-06,
+      "loss": 0.1299,
+      "step": 1006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5902277459254533e-06,
+      "loss": 0.1389,
+      "step": 1007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5735470105188166e-06,
+      "loss": 0.1682,
+      "step": 1008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5569122239209366e-06,
+      "loss": 0.1118,
+      "step": 1009
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5403234890542483e-06,
+      "loss": 0.1663,
+      "step": 1010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.523780908556265e-06,
+      "loss": 0.1281,
+      "step": 1011
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.507284584778913e-06,
+      "loss": 0.1573,
+      "step": 1012
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.490834619787943e-06,
+      "loss": 0.1495,
+      "step": 1013
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4744311153622636e-06,
+      "loss": 0.1109,
+      "step": 1014
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4580741729933246e-06,
+      "loss": 0.1307,
+      "step": 1015
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4417638938844877e-06,
+      "loss": 0.1447,
+      "step": 1016
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.425500378950405e-06,
+      "loss": 0.1495,
+      "step": 1017
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4092837288163807e-06,
+      "loss": 0.1116,
+      "step": 1018
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3931140438177724e-06,
+      "loss": 0.1251,
+      "step": 1019
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3769914239993432e-06,
+      "loss": 0.1301,
+      "step": 1020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3609159691146577e-06,
+      "loss": 0.132,
+      "step": 1021
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3448877786254742e-06,
+      "loss": 0.1156,
+      "step": 1022
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3289069517010953e-06,
+      "loss": 0.1458,
+      "step": 1023
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.312973587217798e-06,
+      "loss": 0.117,
+      "step": 1024
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2970877837581917e-06,
+      "loss": 0.1222,
+      "step": 1025
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2812496396106155e-06,
+      "loss": 0.1165,
+      "step": 1026
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2654592527685305e-06,
+      "loss": 0.1253,
+      "step": 1027
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2497167209299277e-06,
+      "loss": 0.1566,
+      "step": 1028
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.234022141496689e-06,
+      "loss": 0.1167,
+      "step": 1029
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2183756115740274e-06,
+      "loss": 0.1401,
+      "step": 1030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.202777227969851e-06,
+      "loss": 0.1691,
+      "step": 1031
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1872270871941824e-06,
+      "loss": 0.1343,
+      "step": 1032
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.171725285458559e-06,
+      "loss": 0.1263,
+      "step": 1033
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1562719186754287e-06,
+      "loss": 0.1284,
+      "step": 1034
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.140867082457575e-06,
+      "loss": 0.1513,
+      "step": 1035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1255108721175066e-06,
+      "loss": 0.1252,
+      "step": 1036
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.110203382666873e-06,
+      "loss": 0.1099,
+      "step": 1037
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0949447088158812e-06,
+      "loss": 0.1269,
+      "step": 1038
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.079734944972717e-06,
+      "loss": 0.1399,
+      "step": 1039
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.064574185242929e-06,
+      "loss": 0.1677,
+      "step": 1040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.049462523428891e-06,
+      "loss": 0.1197,
+      "step": 1041
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0344000530291875e-06,
+      "loss": 0.1111,
+      "step": 1042
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.019386867238048e-06,
+      "loss": 0.1371,
+      "step": 1043
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0044230589447698e-06,
+      "loss": 0.1318,
+      "step": 1044
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9895087207331422e-06,
+      "loss": 0.1212,
+      "step": 1045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9746439448808785e-06,
+      "loss": 0.1719,
+      "step": 1046
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.959828823359037e-06,
+      "loss": 0.1774,
+      "step": 1047
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.945063447831452e-06,
+      "loss": 0.1444,
+      "step": 1048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9303479096541767e-06,
+      "loss": 0.1477,
+      "step": 1049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9156822998749126e-06,
+      "loss": 0.1213,
+      "step": 1050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9010667092324342e-06,
+      "loss": 0.1494,
+      "step": 1051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8865012281560523e-06,
+      "loss": 0.1501,
+      "step": 1052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8719859467650314e-06,
+      "loss": 0.1611,
+      "step": 1053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8575209548680472e-06,
+      "loss": 0.1302,
+      "step": 1054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8431063419626172e-06,
+      "loss": 0.1354,
+      "step": 1055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8287421972345697e-06,
+      "loss": 0.1032,
+      "step": 1056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.814428609557458e-06,
+      "loss": 0.1342,
+      "step": 1057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8001656674920453e-06,
+      "loss": 0.1583,
+      "step": 1058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7859534592857353e-06,
+      "loss": 0.1392,
+      "step": 1059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7717920728720284e-06,
+      "loss": 0.1368,
+      "step": 1060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7576815958699932e-06,
+      "loss": 0.1336,
+      "step": 1061
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7436221155836953e-06,
+      "loss": 0.1251,
+      "step": 1062
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7296137190016916e-06,
+      "loss": 0.1131,
+      "step": 1063
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7156564927964637e-06,
+      "loss": 0.1281,
+      "step": 1064
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7017505233238962e-06,
+      "loss": 0.1192,
+      "step": 1065
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6878958966227366e-06,
+      "loss": 0.1315,
+      "step": 1066
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6740926984140771e-06,
+      "loss": 0.1454,
+      "step": 1067
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6603410141007913e-06,
+      "loss": 0.1409,
+      "step": 1068
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.646640928767047e-06,
+      "loss": 0.1392,
+      "step": 1069
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6329925271777492e-06,
+      "loss": 0.137,
+      "step": 1070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6193958937780285e-06,
+      "loss": 0.1085,
+      "step": 1071
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6058511126927178e-06,
+      "loss": 0.136,
+      "step": 1072
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5923582677258254e-06,
+      "loss": 0.1249,
+      "step": 1073
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5789174423600307e-06,
+      "loss": 0.1355,
+      "step": 1074
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5655287197561497e-06,
+      "loss": 0.1412,
+      "step": 1075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5521921827526332e-06,
+      "loss": 0.1488,
+      "step": 1076
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5389079138650464e-06,
+      "loss": 0.1064,
+      "step": 1077
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5256759952855737e-06,
+      "loss": 0.1736,
+      "step": 1078
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5124965088824795e-06,
+      "loss": 0.1454,
+      "step": 1079
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4993695361996374e-06,
+      "loss": 0.1449,
+      "step": 1080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4862951584560037e-06,
+      "loss": 0.1262,
+      "step": 1081
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.473273456545119e-06,
+      "loss": 0.1353,
+      "step": 1082
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.460304511034606e-06,
+      "loss": 0.149,
+      "step": 1083
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.447388402165686e-06,
+      "loss": 0.1104,
+      "step": 1084
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4345252098526518e-06,
+      "loss": 0.1214,
+      "step": 1085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4217150136824121e-06,
+      "loss": 0.1336,
+      "step": 1086
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4089578929139635e-06,
+      "loss": 0.1774,
+      "step": 1087
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3962539264779206e-06,
+      "loss": 0.1284,
+      "step": 1088
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3836031929760296e-06,
+      "loss": 0.1526,
+      "step": 1089
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.371005770680659e-06,
+      "loss": 0.146,
+      "step": 1090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3584617375343478e-06,
+      "loss": 0.1228,
+      "step": 1091
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3459711711492962e-06,
+      "loss": 0.1277,
+      "step": 1092
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3335341488068997e-06,
+      "loss": 0.1296,
+      "step": 1093
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.321150747457265e-06,
+      "loss": 0.1346,
+      "step": 1094
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.308821043718743e-06,
+      "loss": 0.1183,
+      "step": 1095
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2965451138774343e-06,
+      "loss": 0.1393,
+      "step": 1096
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2843230338867452e-06,
+      "loss": 0.1581,
+      "step": 1097
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2721548793668936e-06,
+      "loss": 0.1227,
+      "step": 1098
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2600407256044544e-06,
+      "loss": 0.1638,
+      "step": 1099
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2479806475518896e-06,
+      "loss": 0.1273,
+      "step": 1100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2359747198270832e-06,
+      "loss": 0.1533,
+      "step": 1101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2240230167128863e-06,
+      "loss": 0.1843,
+      "step": 1102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.212125612156647e-06,
+      "loss": 0.1274,
+      "step": 1103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2002825797697604e-06,
+      "loss": 0.1323,
+      "step": 1104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.188493992827211e-06,
+      "loss": 0.1273,
+      "step": 1105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1767599242671245e-06,
+      "loss": 0.1308,
+      "step": 1106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1650804466902987e-06,
+      "loss": 0.1276,
+      "step": 1107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1534556323597824e-06,
+      "loss": 0.1516,
+      "step": 1108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1418855532004025e-06,
+      "loss": 0.134,
+      "step": 1109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.130370280798334e-06,
+      "loss": 0.1335,
+      "step": 1110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1189098864006488e-06,
+      "loss": 0.1317,
+      "step": 1111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.107504440914885e-06,
+      "loss": 0.1416,
+      "step": 1112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0961540149085958e-06,
+      "loss": 0.1387,
+      "step": 1113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.084858678608922e-06,
+      "loss": 0.1221,
+      "step": 1114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.073618501902154e-06,
+      "loss": 0.1387,
+      "step": 1115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0624335543332964e-06,
+      "loss": 0.125,
+      "step": 1116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0513039051056507e-06,
+      "loss": 0.1354,
+      "step": 1117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.040229623080362e-06,
+      "loss": 0.1358,
+      "step": 1118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0292107767760261e-06,
+      "loss": 0.1284,
+      "step": 1119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0182474343682346e-06,
+      "loss": 0.1293,
+      "step": 1120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.007339663689173e-06,
+      "loss": 0.1234,
+      "step": 1121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.964875322271895e-07,
+      "loss": 0.1043,
+      "step": 1122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.85691107126392e-07,
+      "loss": 0.1461,
+      "step": 1123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.749504551862087e-07,
+      "loss": 0.1273,
+      "step": 1124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.642656428610042e-07,
+      "loss": 0.1233,
+      "step": 1125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.536367362596422e-07,
+      "loss": 0.1323,
+      "step": 1126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.430638011450932e-07,
+      "loss": 0.1294,
+      "step": 1127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.325469029340195e-07,
+      "loss": 0.1316,
+      "step": 1128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.220861066963715e-07,
+      "loss": 0.125,
+      "step": 1129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.116814771549943e-07,
+      "loss": 0.1264,
+      "step": 1130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.013330786852093e-07,
+      "loss": 0.1618,
+      "step": 1131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.910409753144344e-07,
+      "loss": 0.1174,
+      "step": 1132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.808052307217707e-07,
+      "loss": 0.0969,
+      "step": 1133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.7062590823763e-07,
+      "loss": 0.1237,
+      "step": 1134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.605030708433149e-07,
+      "loss": 0.1344,
+      "step": 1135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.504367811706582e-07,
+      "loss": 0.1058,
+      "step": 1136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.404271015016152e-07,
+      "loss": 0.1223,
+      "step": 1137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.304740937678835e-07,
+      "loss": 0.1405,
+      "step": 1138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.205778195505221e-07,
+      "loss": 0.142,
+      "step": 1139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.107383400795754e-07,
+      "loss": 0.1165,
+      "step": 1140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.009557162336823e-07,
+      "loss": 0.1229,
+      "step": 1141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.912300085397051e-07,
+      "loss": 0.1359,
+      "step": 1142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.815612771723591e-07,
+      "loss": 0.1432,
+      "step": 1143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.719495819538325e-07,
+      "loss": 0.1377,
+      "step": 1144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.623949823534282e-07,
+      "loss": 0.1649,
+      "step": 1145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.528975374871783e-07,
+      "loss": 0.1356,
+      "step": 1146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.434573061174966e-07,
+      "loss": 0.1296,
+      "step": 1147
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.340743466528021e-07,
+      "loss": 0.1578,
+      "step": 1148
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.247487171471624e-07,
+      "loss": 0.1542,
+      "step": 1149
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.154804752999344e-07,
+      "loss": 0.1497,
+      "step": 1150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.062696784554124e-07,
+      "loss": 0.1151,
+      "step": 1151
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.971163836024575e-07,
+      "loss": 0.1234,
+      "step": 1152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.880206473741646e-07,
+      "loss": 0.1415,
+      "step": 1153
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.789825260475014e-07,
+      "loss": 0.1332,
+      "step": 1154
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.700020755429559e-07,
+      "loss": 0.1279,
+      "step": 1155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.610793514242075e-07,
+      "loss": 0.1436,
+      "step": 1156
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.522144088977578e-07,
+      "loss": 0.125,
+      "step": 1157
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.434073028126164e-07,
+      "loss": 0.1414,
+      "step": 1158
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.346580876599395e-07,
+      "loss": 0.1407,
+      "step": 1159
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.259668175727063e-07,
+      "loss": 0.1343,
+      "step": 1160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.17333546325376e-07,
+      "loss": 0.1296,
+      "step": 1161
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.08758327333564e-07,
+      "loss": 0.1446,
+      "step": 1162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.002412136536972e-07,
+      "loss": 0.1435,
+      "step": 1163
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.917822579827037e-07,
+      "loss": 0.1203,
+      "step": 1164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.833815126576714e-07,
+      "loss": 0.1339,
+      "step": 1165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.750390296555319e-07,
+      "loss": 0.1219,
+      "step": 1166
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.667548605927365e-07,
+      "loss": 0.1252,
+      "step": 1167
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.585290567249424e-07,
+      "loss": 0.1125,
+      "step": 1168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.503616689466829e-07,
+      "loss": 0.1358,
+      "step": 1169
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.422527477910666e-07,
+      "loss": 0.121,
+      "step": 1170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.342023434294552e-07,
+      "loss": 0.132,
+      "step": 1171
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.262105056711575e-07,
+      "loss": 0.1108,
+      "step": 1172
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.182772839631223e-07,
+      "loss": 0.1111,
+      "step": 1173
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.104027273896239e-07,
+      "loss": 0.1494,
+      "step": 1174
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.025868846719728e-07,
+      "loss": 0.1277,
+      "step": 1175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.94829804168202e-07,
+      "loss": 0.1166,
+      "step": 1176
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.871315338727711e-07,
+      "loss": 0.1416,
+      "step": 1177
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.794921214162717e-07,
+      "loss": 0.1551,
+      "step": 1178
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7191161406513363e-07,
+      "loss": 0.1167,
+      "step": 1179
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6439005872132457e-07,
+      "loss": 0.1158,
+      "step": 1180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.569275019220709e-07,
+      "loss": 0.1466,
+      "step": 1181
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4952398983956205e-07,
+      "loss": 0.1682,
+      "step": 1182
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.421795682806662e-07,
+      "loss": 0.1123,
+      "step": 1183
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.34894282686652e-07,
+      "loss": 0.1434,
+      "step": 1184
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.27668178132894e-07,
+      "loss": 0.1271,
+      "step": 1185
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2050129932861394e-07,
+      "loss": 0.1254,
+      "step": 1186
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1339369061658696e-07,
+      "loss": 0.1512,
+      "step": 1187
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.063453959728747e-07,
+      "loss": 0.1283,
+      "step": 1188
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9935645900654906e-07,
+      "loss": 0.1514,
+      "step": 1189
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.924269229594335e-07,
+      "loss": 0.1357,
+      "step": 1190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.855568307058155e-07,
+      "loss": 0.1205,
+      "step": 1191
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7874622475220336e-07,
+      "loss": 0.1191,
+      "step": 1192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7199514723704754e-07,
+      "loss": 0.1203,
+      "step": 1193
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6530363993048654e-07,
+      "loss": 0.1058,
+      "step": 1194
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.58671744234087e-07,
+      "loss": 0.1151,
+      "step": 1195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5209950118058835e-07,
+      "loss": 0.1348,
+      "step": 1196
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.455869514336474e-07,
+      "loss": 0.1615,
+      "step": 1197
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3913413528758877e-07,
+      "loss": 0.1454,
+      "step": 1198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.327410926671526e-07,
+      "loss": 0.1103,
+      "step": 1199
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.264078631272505e-07,
+      "loss": 0.142,
+      "step": 1200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2013448585272333e-07,
+      "loss": 0.1428,
+      "step": 1201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.139209996580872e-07,
+      "loss": 0.1244,
+      "step": 1202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.07767442987309e-07,
+      "loss": 0.1048,
+      "step": 1203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.016738539135566e-07,
+      "loss": 0.1411,
+      "step": 1204
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9564027013896713e-07,
+      "loss": 0.1516,
+      "step": 1205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.896667289944166e-07,
+      "loss": 0.1236,
+      "step": 1206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.837532674392862e-07,
+      "loss": 0.0942,
+      "step": 1207
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7789992206123106e-07,
+      "loss": 0.1439,
+      "step": 1208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.721067290759594e-07,
+      "loss": 0.1613,
+      "step": 1209
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6637372432700483e-07,
+      "loss": 0.1216,
+      "step": 1210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6070094328550344e-07,
+      "loss": 0.1573,
+      "step": 1211
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5508842104998265e-07,
+      "loss": 0.1479,
+      "step": 1212
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4953619234612816e-07,
+      "loss": 0.1694,
+      "step": 1213
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4404429152658726e-07,
+      "loss": 0.129,
+      "step": 1214
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.386127525707427e-07,
+      "loss": 0.1313,
+      "step": 1215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.332416090845102e-07,
+      "loss": 0.1096,
+      "step": 1216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.279308943001246e-07,
+      "loss": 0.1484,
+      "step": 1217
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2268064107594302e-07,
+      "loss": 0.1172,
+      "step": 1218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1749088189622848e-07,
+      "loss": 0.1484,
+      "step": 1219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1236164887096345e-07,
+      "loss": 0.1427,
+      "step": 1220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0729297373563995e-07,
+      "loss": 0.1105,
+      "step": 1221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0228488785106636e-07,
+      "loss": 0.1247,
+      "step": 1222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9733742220317876e-07,
+      "loss": 0.139,
+      "step": 1223
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9245060740284316e-07,
+      "loss": 0.1609,
+      "step": 1224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8762447368566582e-07,
+      "loss": 0.0999,
+      "step": 1225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.828590509118089e-07,
+      "loss": 0.1661,
+      "step": 1226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7815436856580714e-07,
+      "loss": 0.1401,
+      "step": 1227
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7351045575638047e-07,
+      "loss": 0.119,
+      "step": 1228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6892734121625954e-07,
+      "loss": 0.1142,
+      "step": 1229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.644050533020014e-07,
+      "loss": 0.1285,
+      "step": 1230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5994361999382202e-07,
+      "loss": 0.1219,
+      "step": 1231
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5554306889541514e-07,
+      "loss": 0.1527,
+      "step": 1232
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.51203427233787e-07,
+      "loss": 0.1191,
+      "step": 1233
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4692472185908635e-07,
+      "loss": 0.1099,
+      "step": 1234
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.42706979244438e-07,
+      "loss": 0.1492,
+      "step": 1235
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3855022548577845e-07,
+      "loss": 0.144,
+      "step": 1236
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.344544863016961e-07,
+      "loss": 0.1532,
+      "step": 1237
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.304197870332713e-07,
+      "loss": 0.1497,
+      "step": 1238
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2644615264391536e-07,
+      "loss": 0.1043,
+      "step": 1239
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.225336077192274e-07,
+      "loss": 0.0963,
+      "step": 1240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1868217646682889e-07,
+      "loss": 0.1232,
+      "step": 1241
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1489188271622154e-07,
+      "loss": 0.1072,
+      "step": 1242
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1116274991864073e-07,
+      "loss": 0.1263,
+      "step": 1243
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0749480114690458e-07,
+      "loss": 0.1107,
+      "step": 1244
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0388805909527621e-07,
+      "loss": 0.1743,
+      "step": 1245
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.003425460793217e-07,
+      "loss": 0.1223,
+      "step": 1246
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.685828403577124e-08,
+      "loss": 0.1193,
+      "step": 1247
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.343529452238598e-08,
+      "loss": 0.1379,
+      "step": 1248
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.00735987178214e-08,
+      "loss": 0.1318,
+      "step": 1249
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.677321742149525e-08,
+      "loss": 0.1224,
+      "step": 1250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.353417105346762e-08,
+      "loss": 0.1554,
+      "step": 1251
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.035647965430215e-08,
+      "loss": 0.1392,
+      "step": 1252
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.724016288495173e-08,
+      "loss": 0.1354,
+      "step": 1253
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.418524002663296e-08,
+      "loss": 0.1282,
+      "step": 1254
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.119172998070412e-08,
+      "loss": 0.1335,
+      "step": 1255
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.825965126855517e-08,
+      "loss": 0.16,
+      "step": 1256
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.53890220314879e-08,
+      "loss": 0.1734,
+      "step": 1257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.25798600306049e-08,
+      "loss": 0.1393,
+      "step": 1258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.983218264669965e-08,
+      "loss": 0.1155,
+      "step": 1259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.714600688014882e-08,
+      "loss": 0.1148,
+      "step": 1260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4521349350808996e-08,
+      "loss": 0.1144,
+      "step": 1261
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.195822629791125e-08,
+      "loss": 0.1296,
+      "step": 1262
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.945665357996121e-08,
+      "loss": 0.1425,
+      "step": 1263
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.701664667464245e-08,
+      "loss": 0.1125,
+      "step": 1264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.463822067871993e-08,
+      "loss": 0.1179,
+      "step": 1265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2321390307944464e-08,
+      "loss": 0.1641,
+      "step": 1266
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.006616989696621e-08,
+      "loss": 0.1232,
+      "step": 1267
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.787257339924133e-08,
+      "loss": 0.1457,
+      "step": 1268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5740614386947644e-08,
+      "loss": 0.1624,
+      "step": 1269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.367030605090249e-08,
+      "loss": 0.1359,
+      "step": 1270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.166166120047831e-08,
+      "loss": 0.1428,
+      "step": 1271
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.971469226352275e-08,
+      "loss": 0.1678,
+      "step": 1272
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7829411286287577e-08,
+      "loss": 0.1122,
+      "step": 1273
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.600582993334544e-08,
+      "loss": 0.1486,
+      "step": 1274
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4243959487524337e-08,
+      "loss": 0.1507,
+      "step": 1275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.254381084983659e-08,
+      "loss": 0.1028,
+      "step": 1276
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0905394539409984e-08,
+      "loss": 0.1249,
+      "step": 1277
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9328720693420065e-08,
+      "loss": 0.1239,
+      "step": 1278
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.781379906703573e-08,
+      "loss": 0.1354,
+      "step": 1279
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.636063903335039e-08,
+      "loss": 0.1522,
+      "step": 1280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.496924958332868e-08,
+      "loss": 0.1352,
+      "step": 1281
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3639639325748744e-08,
+      "loss": 0.1457,
+      "step": 1282
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2371816487150023e-08,
+      "loss": 0.131,
+      "step": 1283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1165788911781106e-08,
+      "loss": 0.1323,
+      "step": 1284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0021564061554189e-08,
+      "loss": 0.1082,
+      "step": 1285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.93914901599291e-09,
+      "loss": 0.1111,
+      "step": 1286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.918550472199027e-09,
+      "loss": 0.1308,
+      "step": 1287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.959774744796921e-09,
+      "loss": 0.1291,
+      "step": 1288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.062827765906942e-09,
+      "loss": 0.1315,
+      "step": 1289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.227715085103224e-09,
+      "loss": 0.1454,
+      "step": 1290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.454441869377047e-09,
+      "loss": 0.131,
+      "step": 1291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7430129031124085e-09,
+      "loss": 0.1419,
+      "step": 1292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0934325880460636e-09,
+      "loss": 0.1497,
+      "step": 1293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5057049432519744e-09,
+      "loss": 0.1219,
+      "step": 1294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.979833605104675e-09,
+      "loss": 0.129,
+      "step": 1295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5158218272670611e-09,
+      "loss": 0.1303,
+      "step": 1296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1136724806637411e-09,
+      "loss": 0.1225,
+      "step": 1297
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.733880534654959e-10,
+      "loss": 0.1472,
+      "step": 1298
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.94970651073734e-10,
+      "loss": 0.142,
+      "step": 1299
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.784219961060597e-10,
+      "loss": 0.1168,
+      "step": 1300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.237434283907213e-10,
+      "loss": 0.1114,
+      "step": 1301
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.09359049488478e-11,
+      "loss": 0.106,
+      "step": 1302
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.1397,
+      "step": 1303
+    },
+    {
+      "epoch": 1.0,
+      "step": 1303,
+      "total_flos": 223839164989440.0,
+      "train_loss": 0.15930285860004556,
+      "train_runtime": 5519.5021,
+      "train_samples_per_second": 30.197,
+      "train_steps_per_second": 0.236
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1303,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 800,
+  "total_flos": 223839164989440.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}