diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,7847 +1,15221 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.0,
+  "epoch": 2.0,
   "eval_steps": 500,
-  "global_step": 1303,
+  "global_step": 2532,
   "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,
+      "learning_rate": 2.6315789473684213e-07,
+      "loss": 0.7239,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 0.6365,
+      "learning_rate": 5.263157894736843e-07,
+      "loss": 0.7012,
       "step": 2
     },
     {
       "epoch": 0.0,
-      "learning_rate": 1.5e-06,
-      "loss": 0.6543,
+      "learning_rate": 7.894736842105263e-07,
+      "loss": 0.7109,
       "step": 3
     },
     {
       "epoch": 0.0,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.5972,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 0.6641,
       "step": 4
     },
     {
       "epoch": 0.0,
-      "learning_rate": 2.5e-06,
-      "loss": 0.5361,
+      "learning_rate": 1.3157894736842106e-06,
+      "loss": 0.6204,
       "step": 5
     },
     {
       "epoch": 0.0,
-      "learning_rate": 3e-06,
-      "loss": 0.5745,
+      "learning_rate": 1.5789473684210526e-06,
+      "loss": 0.6628,
       "step": 6
     },
     {
       "epoch": 0.01,
-      "learning_rate": 3.5e-06,
-      "loss": 0.5942,
+      "learning_rate": 1.8421052631578948e-06,
+      "loss": 0.5847,
       "step": 7
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 0.4918,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 0.5625,
       "step": 8
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.5e-06,
-      "loss": 0.4242,
+      "learning_rate": 2.368421052631579e-06,
+      "loss": 0.6177,
       "step": 9
     },
     {
       "epoch": 0.01,
-      "learning_rate": 5e-06,
-      "loss": 0.2893,
+      "learning_rate": 2.631578947368421e-06,
+      "loss": 0.4053,
       "step": 10
     },
     {
       "epoch": 0.01,
-      "learning_rate": 5.500000000000001e-06,
-      "loss": 0.3571,
+      "learning_rate": 2.8947368421052634e-06,
+      "loss": 0.4482,
       "step": 11
     },
     {
       "epoch": 0.01,
-      "learning_rate": 6e-06,
-      "loss": 0.3047,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 0.4109,
       "step": 12
     },
     {
       "epoch": 0.01,
-      "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.283,
+      "learning_rate": 3.421052631578948e-06,
+      "loss": 0.3458,
       "step": 13
     },
     {
       "epoch": 0.01,
-      "learning_rate": 7e-06,
-      "loss": 0.3107,
+      "learning_rate": 3.6842105263157896e-06,
+      "loss": 0.3153,
       "step": 14
     },
     {
       "epoch": 0.01,
-      "learning_rate": 7.500000000000001e-06,
-      "loss": 0.2729,
+      "learning_rate": 3.947368421052632e-06,
+      "loss": 0.3787,
       "step": 15
     },
     {
       "epoch": 0.01,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 0.2882,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 0.3359,
       "step": 16
     },
     {
       "epoch": 0.01,
-      "learning_rate": 8.5e-06,
-      "loss": 0.2396,
+      "learning_rate": 4.473684210526316e-06,
+      "loss": 0.3125,
       "step": 17
     },
     {
       "epoch": 0.01,
-      "learning_rate": 9e-06,
-      "loss": 0.292,
+      "learning_rate": 4.736842105263158e-06,
+      "loss": 0.3079,
       "step": 18
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 9.5e-06,
-      "loss": 0.2642,
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 0.2822,
       "step": 19
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1e-05,
-      "loss": 0.2579,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 0.3159,
       "step": 20
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.0500000000000001e-05,
-      "loss": 0.2875,
+      "learning_rate": 5.526315789473685e-06,
+      "loss": 0.2959,
       "step": 21
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.1000000000000001e-05,
-      "loss": 0.2772,
+      "learning_rate": 5.789473684210527e-06,
+      "loss": 0.2277,
       "step": 22
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.15e-05,
-      "loss": 0.2433,
+      "learning_rate": 6.0526315789473685e-06,
+      "loss": 0.2153,
       "step": 23
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.2e-05,
-      "loss": 0.2805,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 0.2682,
       "step": 24
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.25e-05,
-      "loss": 0.251,
+      "learning_rate": 6.578947368421054e-06,
+      "loss": 0.306,
       "step": 25
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.3000000000000001e-05,
-      "loss": 0.2667,
+      "learning_rate": 6.842105263157896e-06,
+      "loss": 0.2457,
       "step": 26
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.3500000000000001e-05,
-      "loss": 0.2556,
+      "learning_rate": 7.1052631578947375e-06,
+      "loss": 0.2581,
       "step": 27
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.4e-05,
-      "loss": 0.243,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 0.2606,
       "step": 28
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.45e-05,
-      "loss": 0.2465,
+      "learning_rate": 7.631578947368423e-06,
+      "loss": 0.26,
       "step": 29
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.5000000000000002e-05,
-      "loss": 0.2405,
+      "learning_rate": 7.894736842105265e-06,
+      "loss": 0.2887,
       "step": 30
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.55e-05,
-      "loss": 0.2587,
+      "learning_rate": 8.157894736842106e-06,
+      "loss": 0.2501,
       "step": 31
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.6000000000000003e-05,
-      "loss": 0.2276,
+      "epoch": 0.03,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 0.2876,
       "step": 32
     },
     {
       "epoch": 0.03,
-      "learning_rate": 1.65e-05,
-      "loss": 0.2637,
+      "learning_rate": 8.68421052631579e-06,
+      "loss": 0.2555,
       "step": 33
     },
     {
       "epoch": 0.03,
-      "learning_rate": 1.7e-05,
-      "loss": 0.2574,
+      "learning_rate": 8.947368421052632e-06,
+      "loss": 0.259,
       "step": 34
     },
     {
       "epoch": 0.03,
-      "learning_rate": 1.7500000000000002e-05,
-      "loss": 0.2114,
+      "learning_rate": 9.210526315789474e-06,
+      "loss": 0.2455,
       "step": 35
     },
     {
       "epoch": 0.03,
-      "learning_rate": 1.8e-05,
-      "loss": 0.2324,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.2361,
       "step": 36
     },
     {
       "epoch": 0.03,
-      "learning_rate": 1.8500000000000002e-05,
-      "loss": 0.201,
+      "learning_rate": 9.736842105263159e-06,
+      "loss": 0.2563,
       "step": 37
     },
     {
       "epoch": 0.03,
-      "learning_rate": 1.9e-05,
-      "loss": 0.1838,
+      "learning_rate": 1e-05,
+      "loss": 0.2723,
       "step": 38
     },
     {
       "epoch": 0.03,
-      "learning_rate": 1.95e-05,
-      "loss": 0.2421,
+      "learning_rate": 1.0263157894736844e-05,
+      "loss": 0.2354,
       "step": 39
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2e-05,
-      "loss": 0.2412,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.2926,
       "step": 40
     },
     {
       "epoch": 0.03,
-      "learning_rate": 1.9999969064095054e-05,
-      "loss": 0.2747,
+      "learning_rate": 1.0789473684210528e-05,
+      "loss": 0.2101,
       "step": 41
     },
     {
       "epoch": 0.03,
-      "learning_rate": 1.9999876256571612e-05,
-      "loss": 0.2628,
+      "learning_rate": 1.105263157894737e-05,
+      "loss": 0.2506,
       "step": 42
     },
     {
       "epoch": 0.03,
-      "learning_rate": 1.9999721578003894e-05,
-      "loss": 0.2518,
+      "learning_rate": 1.1315789473684212e-05,
+      "loss": 0.2449,
       "step": 43
     },
     {
       "epoch": 0.03,
-      "learning_rate": 1.9999505029348927e-05,
-      "loss": 0.1815,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 0.2094,
       "step": 44
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.9999226611946536e-05,
-      "loss": 0.2665,
+      "epoch": 0.04,
+      "learning_rate": 1.1842105263157895e-05,
+      "loss": 0.2664,
       "step": 45
     },
     {
       "epoch": 0.04,
-      "learning_rate": 1.9998886327519337e-05,
-      "loss": 0.2011,
+      "learning_rate": 1.2105263157894737e-05,
+      "loss": 0.1968,
       "step": 46
     },
     {
       "epoch": 0.04,
-      "learning_rate": 1.9998484178172735e-05,
-      "loss": 0.2118,
+      "learning_rate": 1.236842105263158e-05,
+      "loss": 0.2122,
       "step": 47
     },
     {
       "epoch": 0.04,
-      "learning_rate": 1.9998020166394898e-05,
-      "loss": 0.2552,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 0.2319,
       "step": 48
     },
     {
       "epoch": 0.04,
-      "learning_rate": 1.999749429505675e-05,
-      "loss": 0.2316,
+      "learning_rate": 1.2894736842105264e-05,
+      "loss": 0.2343,
       "step": 49
     },
     {
       "epoch": 0.04,
-      "learning_rate": 1.9996906567411957e-05,
-      "loss": 0.191,
+      "learning_rate": 1.3157894736842108e-05,
+      "loss": 0.1932,
       "step": 50
     },
     {
       "epoch": 0.04,
-      "learning_rate": 1.999625698709689e-05,
-      "loss": 0.1987,
+      "learning_rate": 1.3421052631578948e-05,
+      "loss": 0.2632,
       "step": 51
     },
     {
       "epoch": 0.04,
-      "learning_rate": 1.9995545558130624e-05,
-      "loss": 0.2211,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 0.2258,
       "step": 52
     },
     {
       "epoch": 0.04,
-      "learning_rate": 1.9994772284914897e-05,
-      "loss": 0.2299,
+      "learning_rate": 1.3947368421052631e-05,
+      "loss": 0.2462,
       "step": 53
     },
     {
       "epoch": 0.04,
-      "learning_rate": 1.9993937172234094e-05,
-      "loss": 0.203,
+      "learning_rate": 1.4210526315789475e-05,
+      "loss": 0.2561,
       "step": 54
     },
     {
       "epoch": 0.04,
-      "learning_rate": 1.9993040225255205e-05,
-      "loss": 0.2178,
+      "learning_rate": 1.4473684210526317e-05,
+      "loss": 0.2427,
       "step": 55
     },
     {
       "epoch": 0.04,
-      "learning_rate": 1.9992081449527804e-05,
-      "loss": 0.2125,
+      "learning_rate": 1.4736842105263159e-05,
+      "loss": 0.2552,
       "step": 56
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.9991060850984007e-05,
-      "loss": 0.2457,
+      "epoch": 0.05,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.2261,
       "step": 57
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.998997843593845e-05,
-      "loss": 0.2397,
+      "epoch": 0.05,
+      "learning_rate": 1.5263157894736846e-05,
+      "loss": 0.2158,
       "step": 58
     },
     {
       "epoch": 0.05,
-      "learning_rate": 1.998883421108822e-05,
-      "loss": 0.217,
+      "learning_rate": 1.5526315789473686e-05,
+      "loss": 0.2233,
       "step": 59
     },
     {
       "epoch": 0.05,
-      "learning_rate": 1.9987628183512854e-05,
-      "loss": 0.2744,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 0.2129,
       "step": 60
     },
     {
       "epoch": 0.05,
-      "learning_rate": 1.9986360360674252e-05,
-      "loss": 0.1821,
+      "learning_rate": 1.605263157894737e-05,
+      "loss": 0.2444,
       "step": 61
     },
     {
       "epoch": 0.05,
-      "learning_rate": 1.998503075041667e-05,
-      "loss": 0.2382,
+      "learning_rate": 1.6315789473684213e-05,
+      "loss": 0.2185,
       "step": 62
     },
     {
       "epoch": 0.05,
-      "learning_rate": 1.998363936096665e-05,
-      "loss": 0.2343,
+      "learning_rate": 1.6578947368421053e-05,
+      "loss": 0.1999,
       "step": 63
     },
     {
       "epoch": 0.05,
-      "learning_rate": 1.9982186200932964e-05,
-      "loss": 0.2371,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 0.1981,
       "step": 64
     },
     {
       "epoch": 0.05,
-      "learning_rate": 1.998067127930658e-05,
-      "loss": 0.2141,
+      "learning_rate": 1.7105263157894737e-05,
+      "loss": 0.218,
       "step": 65
     },
     {
       "epoch": 0.05,
-      "learning_rate": 1.997909460546059e-05,
-      "loss": 0.2029,
+      "learning_rate": 1.736842105263158e-05,
+      "loss": 0.1497,
       "step": 66
     },
     {
       "epoch": 0.05,
-      "learning_rate": 1.9977456189150164e-05,
-      "loss": 0.1962,
+      "learning_rate": 1.763157894736842e-05,
+      "loss": 0.2368,
       "step": 67
     },
     {
       "epoch": 0.05,
-      "learning_rate": 1.9975756040512477e-05,
-      "loss": 0.2177,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 0.2047,
       "step": 68
     },
     {
       "epoch": 0.05,
-      "learning_rate": 1.9973994170066654e-05,
-      "loss": 0.2207,
+      "learning_rate": 1.8157894736842107e-05,
+      "loss": 0.1449,
       "step": 69
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.9972170588713715e-05,
-      "loss": 0.1895,
+      "epoch": 0.06,
+      "learning_rate": 1.8421052631578947e-05,
+      "loss": 0.2118,
       "step": 70
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.997028530773648e-05,
-      "loss": 0.1954,
+      "epoch": 0.06,
+      "learning_rate": 1.868421052631579e-05,
+      "loss": 0.272,
       "step": 71
     },
     {
       "epoch": 0.06,
-      "learning_rate": 1.9968338338799525e-05,
-      "loss": 0.2328,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 0.1837,
       "step": 72
     },
     {
       "epoch": 0.06,
-      "learning_rate": 1.9966329693949098e-05,
-      "loss": 0.2108,
+      "learning_rate": 1.9210526315789474e-05,
+      "loss": 0.223,
       "step": 73
     },
     {
       "epoch": 0.06,
-      "learning_rate": 1.9964259385613053e-05,
-      "loss": 0.2124,
+      "learning_rate": 1.9473684210526318e-05,
+      "loss": 0.2188,
       "step": 74
     },
     {
       "epoch": 0.06,
-      "learning_rate": 1.996212742660076e-05,
-      "loss": 0.1819,
+      "learning_rate": 1.9736842105263158e-05,
+      "loss": 0.1758,
       "step": 75
     },
     {
       "epoch": 0.06,
-      "learning_rate": 1.9959933830103034e-05,
-      "loss": 0.2227,
+      "learning_rate": 2e-05,
+      "loss": 0.1987,
       "step": 76
     },
     {
       "epoch": 0.06,
-      "learning_rate": 1.9957678609692056e-05,
-      "loss": 0.196,
+      "learning_rate": 1.9999991818876183e-05,
+      "loss": 0.2074,
       "step": 77
     },
     {
       "epoch": 0.06,
-      "learning_rate": 1.995536177932128e-05,
-      "loss": 0.1704,
+      "learning_rate": 1.9999967275518118e-05,
+      "loss": 0.2114,
       "step": 78
     },
     {
       "epoch": 0.06,
-      "learning_rate": 1.9952983353325358e-05,
-      "loss": 0.1788,
+      "learning_rate": 1.999992636996596e-05,
+      "loss": 0.1865,
       "step": 79
     },
     {
       "epoch": 0.06,
-      "learning_rate": 1.9950543346420042e-05,
-      "loss": 0.2139,
+      "learning_rate": 1.9999869102286638e-05,
+      "loss": 0.1844,
       "step": 80
     },
     {
       "epoch": 0.06,
-      "learning_rate": 1.994804177370209e-05,
-      "loss": 0.1984,
+      "learning_rate": 1.9999795472573865e-05,
+      "loss": 0.2158,
       "step": 81
     },
     {
       "epoch": 0.06,
-      "learning_rate": 1.9945478650649192e-05,
-      "loss": 0.1946,
+      "learning_rate": 1.9999705480948107e-05,
+      "loss": 0.1862,
       "step": 82
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.9942853993119852e-05,
-      "loss": 0.2256,
+      "epoch": 0.07,
+      "learning_rate": 1.9999599127556614e-05,
+      "loss": 0.2003,
       "step": 83
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.99401678173533e-05,
-      "loss": 0.2323,
+      "epoch": 0.07,
+      "learning_rate": 1.99994764125734e-05,
+      "loss": 0.1917,
       "step": 84
     },
     {
       "epoch": 0.07,
-      "learning_rate": 1.9937420139969397e-05,
-      "loss": 0.1898,
+      "learning_rate": 1.9999337336199257e-05,
+      "loss": 0.2457,
       "step": 85
     },
     {
       "epoch": 0.07,
-      "learning_rate": 1.9934610977968514e-05,
-      "loss": 0.1957,
+      "learning_rate": 1.999918189866175e-05,
+      "loss": 0.2261,
       "step": 86
     },
     {
       "epoch": 0.07,
-      "learning_rate": 1.9931740348731447e-05,
-      "loss": 0.1924,
+      "learning_rate": 1.9999010100215202e-05,
+      "loss": 0.1499,
       "step": 87
     },
     {
       "epoch": 0.07,
-      "learning_rate": 1.9928808270019297e-05,
-      "loss": 0.2029,
+      "learning_rate": 1.9998821941140716e-05,
+      "loss": 0.2052,
       "step": 88
     },
     {
       "epoch": 0.07,
-      "learning_rate": 1.992581475997337e-05,
-      "loss": 0.2148,
+      "learning_rate": 1.9998617421746166e-05,
+      "loss": 0.1982,
       "step": 89
     },
     {
       "epoch": 0.07,
-      "learning_rate": 1.992275983711505e-05,
-      "loss": 0.2034,
+      "learning_rate": 1.9998396542366188e-05,
+      "loss": 0.1809,
       "step": 90
     },
     {
       "epoch": 0.07,
-      "learning_rate": 1.9919643520345698e-05,
-      "loss": 0.1851,
+      "learning_rate": 1.9998159303362193e-05,
+      "loss": 0.2004,
       "step": 91
     },
     {
       "epoch": 0.07,
-      "learning_rate": 1.9916465828946533e-05,
-      "loss": 0.1973,
+      "learning_rate": 1.9997905705122352e-05,
+      "loss": 0.1782,
       "step": 92
     },
     {
       "epoch": 0.07,
-      "learning_rate": 1.9913226782578506e-05,
-      "loss": 0.168,
+      "learning_rate": 1.9997635748061615e-05,
+      "loss": 0.1495,
       "step": 93
     },
     {
       "epoch": 0.07,
-      "learning_rate": 1.990992640128218e-05,
-      "loss": 0.2184,
+      "learning_rate": 1.999734943262169e-05,
+      "loss": 0.1998,
       "step": 94
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9906564705477616e-05,
-      "loss": 0.2015,
+      "epoch": 0.08,
+      "learning_rate": 1.9997046759271055e-05,
+      "loss": 0.2246,
       "step": 95
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9903141715964233e-05,
-      "loss": 0.1582,
+      "epoch": 0.08,
+      "learning_rate": 1.999672772850495e-05,
+      "loss": 0.214,
       "step": 96
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.989965745392068e-05,
-      "loss": 0.1948,
+      "epoch": 0.08,
+      "learning_rate": 1.999639234084538e-05,
+      "loss": 0.1931,
       "step": 97
     },
     {
       "epoch": 0.08,
-      "learning_rate": 1.9896111940904724e-05,
-      "loss": 0.2361,
+      "learning_rate": 1.9996040596841118e-05,
+      "loss": 0.1842,
       "step": 98
     },
     {
       "epoch": 0.08,
-      "learning_rate": 1.9892505198853097e-05,
-      "loss": 0.2212,
+      "learning_rate": 1.999567249706769e-05,
+      "loss": 0.1994,
       "step": 99
     },
     {
       "epoch": 0.08,
-      "learning_rate": 1.988883725008136e-05,
-      "loss": 0.1744,
+      "learning_rate": 1.9995288042127396e-05,
+      "loss": 0.2056,
       "step": 100
     },
     {
       "epoch": 0.08,
-      "learning_rate": 1.988510811728378e-05,
-      "loss": 0.1954,
+      "learning_rate": 1.999488723264929e-05,
+      "loss": 0.1942,
       "step": 101
     },
     {
       "epoch": 0.08,
-      "learning_rate": 1.9881317823533176e-05,
-      "loss": 0.174,
+      "learning_rate": 1.999447006928918e-05,
+      "loss": 0.2142,
       "step": 102
     },
     {
       "epoch": 0.08,
-      "learning_rate": 1.9877466392280773e-05,
-      "loss": 0.1631,
+      "learning_rate": 1.999403655272965e-05,
+      "loss": 0.2058,
       "step": 103
     },
     {
       "epoch": 0.08,
-      "learning_rate": 1.9873553847356087e-05,
-      "loss": 0.1951,
+      "learning_rate": 1.999358668368002e-05,
+      "loss": 0.1619,
       "step": 104
     },
     {
       "epoch": 0.08,
-      "learning_rate": 1.986958021296673e-05,
-      "loss": 0.1931,
+      "learning_rate": 1.9993120462876385e-05,
+      "loss": 0.1484,
       "step": 105
     },
     {
       "epoch": 0.08,
-      "learning_rate": 1.9865545513698304e-05,
-      "loss": 0.1638,
+      "learning_rate": 1.9992637891081585e-05,
+      "loss": 0.207,
       "step": 106
     },
     {
       "epoch": 0.08,
-      "learning_rate": 1.9861449774514225e-05,
-      "loss": 0.2454,
+      "learning_rate": 1.999213896908521e-05,
+      "loss": 0.2008,
       "step": 107
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9857293020755565e-05,
-      "loss": 0.1537,
+      "epoch": 0.09,
+      "learning_rate": 1.9991623697703613e-05,
+      "loss": 0.1827,
       "step": 108
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9853075278140913e-05,
-      "loss": 0.2051,
+      "epoch": 0.09,
+      "learning_rate": 1.9991092077779895e-05,
+      "loss": 0.1765,
       "step": 109
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9848796572766214e-05,
-      "loss": 0.1603,
+      "epoch": 0.09,
+      "learning_rate": 1.9990544110183907e-05,
+      "loss": 0.1802,
       "step": 110
     },
     {
       "epoch": 0.09,
-      "learning_rate": 1.9844456931104588e-05,
-      "loss": 0.1938,
+      "learning_rate": 1.998997979581224e-05,
+      "loss": 0.1995,
       "step": 111
     },
     {
       "epoch": 0.09,
-      "learning_rate": 1.984005638000618e-05,
-      "loss": 0.1907,
+      "learning_rate": 1.9989399135588246e-05,
+      "loss": 0.1694,
       "step": 112
     },
     {
       "epoch": 0.09,
-      "learning_rate": 1.9835594946698e-05,
-      "loss": 0.2059,
+      "learning_rate": 1.9988802130462017e-05,
+      "loss": 0.1765,
       "step": 113
     },
     {
       "epoch": 0.09,
-      "learning_rate": 1.9831072658783745e-05,
-      "loss": 0.2077,
+      "learning_rate": 1.9988188781410377e-05,
+      "loss": 0.1763,
       "step": 114
     },
     {
       "epoch": 0.09,
-      "learning_rate": 1.9826489544243623e-05,
-      "loss": 0.2125,
+      "learning_rate": 1.9987559089436917e-05,
+      "loss": 0.1678,
       "step": 115
     },
     {
       "epoch": 0.09,
-      "learning_rate": 1.9821845631434192e-05,
-      "loss": 0.1701,
+      "learning_rate": 1.998691305557194e-05,
+      "loss": 0.1971,
       "step": 116
     },
     {
       "epoch": 0.09,
-      "learning_rate": 1.9817140949088193e-05,
-      "loss": 0.2443,
+      "learning_rate": 1.9986250680872515e-05,
+      "loss": 0.2261,
       "step": 117
     },
     {
       "epoch": 0.09,
-      "learning_rate": 1.981237552631434e-05,
-      "loss": 0.2032,
+      "learning_rate": 1.998557196642243e-05,
+      "loss": 0.1641,
       "step": 118
     },
     {
       "epoch": 0.09,
-      "learning_rate": 1.980754939259716e-05,
-      "loss": 0.2058,
+      "learning_rate": 1.9984876913332215e-05,
+      "loss": 0.1747,
       "step": 119
     },
     {
       "epoch": 0.09,
-      "learning_rate": 1.9802662577796822e-05,
-      "loss": 0.2188,
+      "learning_rate": 1.9984165522739135e-05,
+      "loss": 0.2225,
       "step": 120
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9797715112148937e-05,
-      "loss": 0.1956,
+      "epoch": 0.1,
+      "learning_rate": 1.998343779580718e-05,
+      "loss": 0.1716,
       "step": 121
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9792707026264364e-05,
-      "loss": 0.1921,
+      "epoch": 0.1,
+      "learning_rate": 1.998269373372708e-05,
+      "loss": 0.1971,
       "step": 122
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.978763835112904e-05,
-      "loss": 0.1755,
+      "epoch": 0.1,
+      "learning_rate": 1.9981933337716288e-05,
+      "loss": 0.1938,
       "step": 123
     },
     {
       "epoch": 0.1,
-      "learning_rate": 1.9782509118103773e-05,
-      "loss": 0.1808,
+      "learning_rate": 1.9981156609018977e-05,
+      "loss": 0.1991,
       "step": 124
     },
     {
       "epoch": 0.1,
-      "learning_rate": 1.977731935892406e-05,
-      "loss": 0.1833,
+      "learning_rate": 1.9980363548906056e-05,
+      "loss": 0.2011,
       "step": 125
     },
     {
       "epoch": 0.1,
-      "learning_rate": 1.9772069105699878e-05,
-      "loss": 0.203,
+      "learning_rate": 1.9979554158675145e-05,
+      "loss": 0.1749,
       "step": 126
     },
     {
       "epoch": 0.1,
-      "learning_rate": 1.9766758390915494e-05,
-      "loss": 0.1536,
+      "learning_rate": 1.997872843965059e-05,
+      "loss": 0.1621,
       "step": 127
     },
     {
       "epoch": 0.1,
-      "learning_rate": 1.976138724742926e-05,
-      "loss": 0.1759,
+      "learning_rate": 1.9977886393183454e-05,
+      "loss": 0.194,
       "step": 128
     },
     {
       "epoch": 0.1,
-      "learning_rate": 1.9755955708473416e-05,
-      "loss": 0.1912,
+      "learning_rate": 1.9977028020651516e-05,
+      "loss": 0.1716,
       "step": 129
     },
     {
       "epoch": 0.1,
-      "learning_rate": 1.9750463807653873e-05,
-      "loss": 0.156,
+      "learning_rate": 1.9976153323459262e-05,
+      "loss": 0.222,
       "step": 130
     },
     {
       "epoch": 0.1,
-      "learning_rate": 1.9744911578950023e-05,
-      "loss": 0.2351,
+      "learning_rate": 1.9975262303037896e-05,
+      "loss": 0.176,
       "step": 131
     },
     {
       "epoch": 0.1,
-      "learning_rate": 1.9739299056714497e-05,
-      "loss": 0.1415,
+      "learning_rate": 1.9974354960845326e-05,
+      "loss": 0.1573,
       "step": 132
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.9733626275673e-05,
-      "loss": 0.1274,
+      "epoch": 0.11,
+      "learning_rate": 1.997343129836617e-05,
+      "loss": 0.1545,
       "step": 133
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.9727893270924043e-05,
-      "loss": 0.1941,
+      "epoch": 0.11,
+      "learning_rate": 1.9972491317111745e-05,
+      "loss": 0.1724,
       "step": 134
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.972210007793877e-05,
-      "loss": 0.1843,
+      "epoch": 0.11,
+      "learning_rate": 1.997153501862007e-05,
+      "loss": 0.1727,
       "step": 135
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.9716246732560715e-05,
-      "loss": 0.1786,
+      "epoch": 0.11,
+      "learning_rate": 1.9970562404455872e-05,
+      "loss": 0.2224,
       "step": 136
     },
     {
       "epoch": 0.11,
-      "learning_rate": 1.9710333271005586e-05,
-      "loss": 0.2245,
+      "learning_rate": 1.9969573476210558e-05,
+      "loss": 0.1988,
       "step": 137
     },
     {
       "epoch": 0.11,
-      "learning_rate": 1.9704359729861035e-05,
-      "loss": 0.1808,
+      "learning_rate": 1.996856823550224e-05,
+      "loss": 0.2096,
       "step": 138
     },
     {
       "epoch": 0.11,
-      "learning_rate": 1.9698326146086446e-05,
-      "loss": 0.1969,
+      "learning_rate": 1.996754668397572e-05,
+      "loss": 0.162,
       "step": 139
     },
     {
       "epoch": 0.11,
-      "learning_rate": 1.9692232557012693e-05,
-      "loss": 0.1827,
+      "learning_rate": 1.9966508823302484e-05,
+      "loss": 0.2034,
       "step": 140
     },
     {
       "epoch": 0.11,
-      "learning_rate": 1.9686079000341914e-05,
-      "loss": 0.1504,
+      "learning_rate": 1.9965454655180704e-05,
+      "loss": 0.1417,
       "step": 141
     },
     {
       "epoch": 0.11,
-      "learning_rate": 1.967986551414728e-05,
-      "loss": 0.2186,
+      "learning_rate": 1.9964384181335237e-05,
+      "loss": 0.1934,
       "step": 142
     },
     {
       "epoch": 0.11,
-      "learning_rate": 1.967359213687275e-05,
-      "loss": 0.1589,
+      "learning_rate": 1.996329740351762e-05,
+      "loss": 0.2012,
       "step": 143
     },
     {
       "epoch": 0.11,
-      "learning_rate": 1.966725890733285e-05,
-      "loss": 0.1551,
+      "learning_rate": 1.9962194323506064e-05,
+      "loss": 0.1806,
       "step": 144
     },
     {
       "epoch": 0.11,
-      "learning_rate": 1.9660865864712413e-05,
-      "loss": 0.1783,
+      "learning_rate": 1.9961074943105457e-05,
+      "loss": 0.1712,
       "step": 145
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9654413048566353e-05,
-      "loss": 0.2355,
+      "epoch": 0.12,
+      "learning_rate": 1.9959939264147355e-05,
+      "loss": 0.1501,
       "step": 146
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9647900498819412e-05,
-      "loss": 0.1827,
+      "epoch": 0.12,
+      "learning_rate": 1.9958787288489983e-05,
+      "loss": 0.1306,
       "step": 147
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9641328255765916e-05,
-      "loss": 0.2344,
+      "epoch": 0.12,
+      "learning_rate": 1.9957619018018243e-05,
+      "loss": 0.1934,
       "step": 148
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9634696360069517e-05,
-      "loss": 0.1632,
+      "epoch": 0.12,
+      "learning_rate": 1.9956434454643675e-05,
+      "loss": 0.218,
       "step": 149
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.9628004852762952e-05,
-      "loss": 0.1532,
+      "learning_rate": 1.9955233600304496e-05,
+      "loss": 0.1915,
       "step": 150
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.96212537752478e-05,
-      "loss": 0.1341,
+      "learning_rate": 1.995401645696557e-05,
+      "loss": 0.1661,
       "step": 151
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.9614443169294186e-05,
-      "loss": 0.1475,
+      "learning_rate": 1.9952783026618424e-05,
+      "loss": 0.1946,
       "step": 152
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.960757307704057e-05,
-      "loss": 0.1717,
+      "learning_rate": 1.995153331128122e-05,
+      "loss": 0.1627,
       "step": 153
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.9600643540993453e-05,
-      "loss": 0.2061,
+      "learning_rate": 1.9950267312998783e-05,
+      "loss": 0.176,
       "step": 154
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.959365460402713e-05,
-      "loss": 0.1491,
+      "learning_rate": 1.994898503384256e-05,
+      "loss": 0.1414,
       "step": 155
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.9586606309383415e-05,
-      "loss": 0.1643,
+      "learning_rate": 1.9947686475910656e-05,
+      "loss": 0.1778,
       "step": 156
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.9579498700671386e-05,
-      "loss": 0.1744,
+      "learning_rate": 1.9946371641327794e-05,
+      "loss": 0.218,
       "step": 157
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.9572331821867108e-05,
-      "loss": 0.1926,
+      "learning_rate": 1.9945040532245352e-05,
+      "loss": 0.2025,
       "step": 158
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.9565105717313352e-05,
-      "loss": 0.1363,
+      "epoch": 0.13,
+      "learning_rate": 1.9943693150841312e-05,
+      "loss": 0.2178,
       "step": 159
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.9557820431719333e-05,
-      "loss": 0.2054,
+      "epoch": 0.13,
+      "learning_rate": 1.9942329499320298e-05,
+      "loss": 0.1744,
       "step": 160
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.955047601016044e-05,
-      "loss": 0.1936,
+      "epoch": 0.13,
+      "learning_rate": 1.994094957991355e-05,
+      "loss": 0.1788,
       "step": 161
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.954307249807793e-05,
-      "loss": 0.1923,
+      "epoch": 0.13,
+      "learning_rate": 1.9939553394878926e-05,
+      "loss": 0.1949,
       "step": 162
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.9535609941278676e-05,
-      "loss": 0.1788,
+      "learning_rate": 1.99381409465009e-05,
+      "loss": 0.1616,
       "step": 163
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.952808838593487e-05,
-      "loss": 0.1501,
+      "learning_rate": 1.9936712237090554e-05,
+      "loss": 0.1937,
       "step": 164
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.9520507878583728e-05,
-      "loss": 0.1892,
+      "learning_rate": 1.9935267268985577e-05,
+      "loss": 0.1719,
       "step": 165
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.9512868466127232e-05,
-      "loss": 0.1671,
+      "learning_rate": 1.9933806044550262e-05,
+      "loss": 0.1735,
       "step": 166
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.95051701958318e-05,
-      "loss": 0.1625,
+      "learning_rate": 1.99323285661755e-05,
+      "loss": 0.1719,
       "step": 167
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.9497413115328028e-05,
-      "loss": 0.1973,
+      "learning_rate": 1.993083483627878e-05,
+      "loss": 0.157,
       "step": 168
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.9489597272610377e-05,
-      "loss": 0.1787,
+      "learning_rate": 1.9929324857304175e-05,
+      "loss": 0.1774,
       "step": 169
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.9481722716036883e-05,
-      "loss": 0.2179,
+      "learning_rate": 1.9927798631722353e-05,
+      "loss": 0.1601,
       "step": 170
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.9473789494328844e-05,
-      "loss": 0.168,
+      "epoch": 0.14,
+      "learning_rate": 1.9926256162030564e-05,
+      "loss": 0.2097,
       "step": 171
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.9465797656570546e-05,
-      "loss": 0.1288,
+      "epoch": 0.14,
+      "learning_rate": 1.9924697450752636e-05,
+      "loss": 0.1826,
       "step": 172
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.9457747252208936e-05,
-      "loss": 0.1256,
+      "epoch": 0.14,
+      "learning_rate": 1.9923122500438964e-05,
+      "loss": 0.1753,
       "step": 173
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.944963833105332e-05,
-      "loss": 0.1658,
+      "epoch": 0.14,
+      "learning_rate": 1.9921531313666526e-05,
+      "loss": 0.1631,
       "step": 174
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.944147094327506e-05,
-      "loss": 0.171,
+      "epoch": 0.14,
+      "learning_rate": 1.9919923893038863e-05,
+      "loss": 0.1866,
       "step": 175
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.9433245139407266e-05,
-      "loss": 0.1676,
+      "learning_rate": 1.991830024118607e-05,
+      "loss": 0.2166,
       "step": 176
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.942496097034447e-05,
-      "loss": 0.1388,
+      "learning_rate": 1.991666036076481e-05,
+      "loss": 0.1594,
       "step": 177
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.9416618487342333e-05,
-      "loss": 0.2108,
+      "learning_rate": 1.99150042544583e-05,
+      "loss": 0.1595,
       "step": 178
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.94082177420173e-05,
-      "loss": 0.173,
+      "learning_rate": 1.9913331924976295e-05,
+      "loss": 0.1654,
       "step": 179
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.9399758786346305e-05,
-      "loss": 0.1769,
+      "learning_rate": 1.991164337505511e-05,
+      "loss": 0.1832,
       "step": 180
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.9391241672666438e-05,
-      "loss": 0.1971,
+      "learning_rate": 1.990993860745758e-05,
+      "loss": 0.184,
       "step": 181
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.9382666453674625e-05,
-      "loss": 0.1708,
+      "learning_rate": 1.99082176249731e-05,
+      "loss": 0.1878,
       "step": 182
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.9374033182427297e-05,
-      "loss": 0.1697,
+      "learning_rate": 1.9906480430417575e-05,
+      "loss": 0.204,
       "step": 183
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.936534191234006e-05,
-      "loss": 0.1718,
+      "epoch": 0.15,
+      "learning_rate": 1.9904727026633453e-05,
+      "loss": 0.196,
       "step": 184
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.9356592697187383e-05,
-      "loss": 0.1803,
+      "epoch": 0.15,
+      "learning_rate": 1.9902957416489693e-05,
+      "loss": 0.1839,
       "step": 185
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.9347785591102244e-05,
-      "loss": 0.1874,
+      "epoch": 0.15,
+      "learning_rate": 1.9901171602881778e-05,
+      "loss": 0.1764,
       "step": 186
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.9338920648575798e-05,
-      "loss": 0.1672,
+      "epoch": 0.15,
+      "learning_rate": 1.9899369588731697e-05,
+      "loss": 0.1997,
       "step": 187
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.9329997924457046e-05,
-      "loss": 0.183,
+      "epoch": 0.15,
+      "learning_rate": 1.9897551376987948e-05,
+      "loss": 0.1846,
       "step": 188
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.93210174739525e-05,
-      "loss": 0.1642,
+      "learning_rate": 1.9895716970625544e-05,
+      "loss": 0.2019,
       "step": 189
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.9311979352625837e-05,
-      "loss": 0.1808,
+      "learning_rate": 1.9893866372645975e-05,
+      "loss": 0.1614,
       "step": 190
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.9302883616397546e-05,
-      "loss": 0.1992,
+      "learning_rate": 1.989199958607724e-05,
+      "loss": 0.2455,
       "step": 191
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.929373032154459e-05,
-      "loss": 0.1766,
+      "learning_rate": 1.9890116613973822e-05,
+      "loss": 0.163,
       "step": 192
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.928451952470007e-05,
-      "loss": 0.184,
+      "learning_rate": 1.9888217459416685e-05,
+      "loss": 0.2125,
       "step": 193
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.927525128285284e-05,
-      "loss": 0.1745,
+      "learning_rate": 1.9886302125513276e-05,
+      "loss": 0.1904,
       "step": 194
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.9265925653347198e-05,
-      "loss": 0.197,
+      "learning_rate": 1.9884370615397507e-05,
+      "loss": 0.1702,
       "step": 195
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.9256542693882505e-05,
-      "loss": 0.2009,
+      "learning_rate": 1.9882422932229765e-05,
+      "loss": 0.2197,
       "step": 196
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9247102462512823e-05,
-      "loss": 0.1737,
+      "epoch": 0.16,
+      "learning_rate": 1.9880459079196898e-05,
+      "loss": 0.2401,
       "step": 197
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9237605017646573e-05,
-      "loss": 0.1817,
+      "epoch": 0.16,
+      "learning_rate": 1.9878479059512212e-05,
+      "loss": 0.1961,
       "step": 198
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.922805041804617e-05,
-      "loss": 0.1366,
+      "epoch": 0.16,
+      "learning_rate": 1.987648287641546e-05,
+      "loss": 0.1503,
       "step": 199
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9218438722827644e-05,
-      "loss": 0.1855,
+      "epoch": 0.16,
+      "learning_rate": 1.987447053317285e-05,
+      "loss": 0.1893,
       "step": 200
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9208769991460298e-05,
-      "loss": 0.179,
+      "epoch": 0.16,
+      "learning_rate": 1.9872442033077027e-05,
+      "loss": 0.2006,
       "step": 201
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.919904428376632e-05,
-      "loss": 0.1719,
+      "learning_rate": 1.9870397379447074e-05,
+      "loss": 0.2062,
       "step": 202
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.9189261659920428e-05,
-      "loss": 0.1447,
+      "learning_rate": 1.98683365756285e-05,
+      "loss": 0.202,
       "step": 203
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.9179422180449478e-05,
-      "loss": 0.1615,
+      "learning_rate": 1.9866259624993246e-05,
+      "loss": 0.1645,
       "step": 204
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.916952590623212e-05,
-      "loss": 0.1361,
+      "learning_rate": 1.986416653093967e-05,
+      "loss": 0.2137,
       "step": 205
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.9159572898498387e-05,
-      "loss": 0.2001,
+      "learning_rate": 1.9862057296892546e-05,
+      "loss": 0.1686,
       "step": 206
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.9149563218829345e-05,
-      "loss": 0.1883,
+      "learning_rate": 1.985993192630305e-05,
+      "loss": 0.1646,
       "step": 207
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.9139496929156685e-05,
-      "loss": 0.1318,
+      "learning_rate": 1.9857790422648774e-05,
+      "loss": 0.1804,
       "step": 208
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.9129374091762375e-05,
-      "loss": 0.1536,
+      "epoch": 0.17,
+      "learning_rate": 1.9855632789433695e-05,
+      "loss": 0.215,
       "step": 209
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.911919476927823e-05,
-      "loss": 0.1636,
+      "epoch": 0.17,
+      "learning_rate": 1.9853459030188183e-05,
+      "loss": 0.1709,
       "step": 210
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.910895902468557e-05,
-      "loss": 0.1565,
+      "epoch": 0.17,
+      "learning_rate": 1.9851269148468998e-05,
+      "loss": 0.161,
       "step": 211
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.9098666921314793e-05,
-      "loss": 0.2114,
+      "epoch": 0.17,
+      "learning_rate": 1.9849063147859282e-05,
+      "loss": 0.1443,
       "step": 212
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.908831852284501e-05,
-      "loss": 0.1618,
+      "epoch": 0.17,
+      "learning_rate": 1.9846841031968545e-05,
+      "loss": 0.1774,
       "step": 213
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.907791389330363e-05,
-      "loss": 0.1819,
+      "epoch": 0.17,
+      "learning_rate": 1.9844602804432667e-05,
+      "loss": 0.1656,
       "step": 214
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.9067453097065986e-05,
-      "loss": 0.204,
+      "learning_rate": 1.9842348468913895e-05,
+      "loss": 0.1469,
       "step": 215
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.905693619885491e-05,
-      "loss": 0.2068,
+      "learning_rate": 1.9840078029100826e-05,
+      "loss": 0.1525,
       "step": 216
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.904636326374036e-05,
-      "loss": 0.1831,
+      "learning_rate": 1.983779148870841e-05,
+      "loss": 0.1416,
       "step": 217
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.9035734357139e-05,
-      "loss": 0.1766,
+      "learning_rate": 1.9835488851477943e-05,
+      "loss": 0.1882,
       "step": 218
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.9025049544813793e-05,
-      "loss": 0.2153,
+      "learning_rate": 1.983317012117705e-05,
+      "loss": 0.1964,
       "step": 219
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.9014308892873612e-05,
-      "loss": 0.1774,
+      "learning_rate": 1.983083530159971e-05,
+      "loss": 0.186,
       "step": 220
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.900351246777281e-05,
-      "loss": 0.2006,
+      "learning_rate": 1.9828484396566197e-05,
+      "loss": 0.167,
       "step": 221
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.899266033631083e-05,
-      "loss": 0.1797,
+      "epoch": 0.18,
+      "learning_rate": 1.982611740992313e-05,
+      "loss": 0.1727,
       "step": 222
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.8981752565631767e-05,
-      "loss": 0.1843,
+      "epoch": 0.18,
+      "learning_rate": 1.9823734345543422e-05,
+      "loss": 0.163,
       "step": 223
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.8970789223223978e-05,
-      "loss": 0.2014,
+      "epoch": 0.18,
+      "learning_rate": 1.982133520732631e-05,
+      "loss": 0.1827,
       "step": 224
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.895977037691964e-05,
-      "loss": 0.183,
+      "epoch": 0.18,
+      "learning_rate": 1.981891999919732e-05,
+      "loss": 0.167,
       "step": 225
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.8948696094894354e-05,
-      "loss": 0.1566,
+      "epoch": 0.18,
+      "learning_rate": 1.981648872510828e-05,
+      "loss": 0.2051,
       "step": 226
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.8937566445666707e-05,
-      "loss": 0.2221,
+      "epoch": 0.18,
+      "learning_rate": 1.9814041389037292e-05,
+      "loss": 0.1754,
       "step": 227
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.892638149809785e-05,
-      "loss": 0.1717,
+      "epoch": 0.18,
+      "learning_rate": 1.9811577994988755e-05,
+      "loss": 0.1783,
       "step": 228
     },
     {
       "epoch": 0.18,
-      "learning_rate": 1.8915141321391083e-05,
-      "loss": 0.1841,
+      "learning_rate": 1.9809098546993333e-05,
+      "loss": 0.1302,
       "step": 229
     },
     {
       "epoch": 0.18,
-      "learning_rate": 1.8903845985091406e-05,
-      "loss": 0.1818,
+      "learning_rate": 1.980660304910796e-05,
+      "loss": 0.1606,
       "step": 230
     },
     {
       "epoch": 0.18,
-      "learning_rate": 1.889249555908512e-05,
-      "loss": 0.1988,
+      "learning_rate": 1.9804091505415833e-05,
+      "loss": 0.1664,
       "step": 231
     },
     {
       "epoch": 0.18,
-      "learning_rate": 1.8881090113599353e-05,
-      "loss": 0.1559,
+      "learning_rate": 1.98015639200264e-05,
+      "loss": 0.1403,
       "step": 232
     },
     {
       "epoch": 0.18,
-      "learning_rate": 1.8869629719201668e-05,
-      "loss": 0.163,
+      "learning_rate": 1.979902029707536e-05,
+      "loss": 0.1882,
       "step": 233
     },
     {
       "epoch": 0.18,
-      "learning_rate": 1.88581144467996e-05,
-      "loss": 0.1536,
+      "learning_rate": 1.9796460640724646e-05,
+      "loss": 0.1729,
       "step": 234
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.8846544367640218e-05,
-      "loss": 0.1746,
+      "epoch": 0.19,
+      "learning_rate": 1.9793884955162442e-05,
+      "loss": 0.1339,
       "step": 235
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.8834919553309704e-05,
-      "loss": 0.1735,
+      "epoch": 0.19,
+      "learning_rate": 1.979129324460314e-05,
+      "loss": 0.2767,
       "step": 236
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.882324007573288e-05,
-      "loss": 0.2136,
+      "epoch": 0.19,
+      "learning_rate": 1.9788685513287368e-05,
+      "loss": 0.1669,
       "step": 237
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.881150600717279e-05,
-      "loss": 0.1572,
+      "epoch": 0.19,
+      "learning_rate": 1.9786061765481954e-05,
+      "loss": 0.1622,
       "step": 238
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.8799717420230242e-05,
-      "loss": 0.1514,
+      "epoch": 0.19,
+      "learning_rate": 1.9783422005479942e-05,
+      "loss": 0.1755,
       "step": 239
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.8787874387843358e-05,
-      "loss": 0.1613,
+      "epoch": 0.19,
+      "learning_rate": 1.9780766237600574e-05,
+      "loss": 0.1993,
       "step": 240
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.8775976983287117e-05,
-      "loss": 0.2228,
+      "epoch": 0.19,
+      "learning_rate": 1.977809446618928e-05,
+      "loss": 0.1592,
       "step": 241
     },
     {
       "epoch": 0.19,
-      "learning_rate": 1.876402528017292e-05,
-      "loss": 0.1721,
+      "learning_rate": 1.9775406695617677e-05,
+      "loss": 0.189,
       "step": 242
     },
     {
       "epoch": 0.19,
-      "learning_rate": 1.8752019352448114e-05,
-      "loss": 0.1595,
+      "learning_rate": 1.977270293028357e-05,
+      "loss": 0.1557,
       "step": 243
     },
     {
       "epoch": 0.19,
-      "learning_rate": 1.873995927439555e-05,
-      "loss": 0.2287,
+      "learning_rate": 1.9769983174610918e-05,
+      "loss": 0.2015,
       "step": 244
     },
     {
       "epoch": 0.19,
-      "learning_rate": 1.8727845120633108e-05,
-      "loss": 0.1858,
+      "learning_rate": 1.9767247433049858e-05,
+      "loss": 0.1606,
       "step": 245
     },
     {
       "epoch": 0.19,
-      "learning_rate": 1.8715676966113256e-05,
-      "loss": 0.1971,
+      "learning_rate": 1.9764495710076678e-05,
+      "loss": 0.1808,
       "step": 246
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.8703454886122568e-05,
-      "loss": 0.1451,
+      "epoch": 0.2,
+      "learning_rate": 1.9761728010193812e-05,
+      "loss": 0.184,
       "step": 247
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.869117895628126e-05,
-      "loss": 0.172,
+      "epoch": 0.2,
+      "learning_rate": 1.975894433792984e-05,
+      "loss": 0.1754,
       "step": 248
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.8678849252542733e-05,
-      "loss": 0.147,
+      "epoch": 0.2,
+      "learning_rate": 1.9756144697839477e-05,
+      "loss": 0.1304,
       "step": 249
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.86664658511931e-05,
-      "loss": 0.1631,
+      "epoch": 0.2,
+      "learning_rate": 1.9753329094503563e-05,
+      "loss": 0.1589,
       "step": 250
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.8654028828850705e-05,
-      "loss": 0.1617,
+      "epoch": 0.2,
+      "learning_rate": 1.9750497532529053e-05,
+      "loss": 0.1736,
       "step": 251
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.8641538262465656e-05,
-      "loss": 0.158,
+      "epoch": 0.2,
+      "learning_rate": 1.974765001654903e-05,
+      "loss": 0.1953,
       "step": 252
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.862899422931934e-05,
-      "loss": 0.1745,
+      "epoch": 0.2,
+      "learning_rate": 1.9744786551222658e-05,
+      "loss": 0.1297,
       "step": 253
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.8616396807023975e-05,
-      "loss": 0.1735,
+      "epoch": 0.2,
+      "learning_rate": 1.974190714123522e-05,
+      "loss": 0.1864,
       "step": 254
     },
     {
       "epoch": 0.2,
-      "learning_rate": 1.860374607352208e-05,
-      "loss": 0.1605,
+      "learning_rate": 1.9739011791298073e-05,
+      "loss": 0.1673,
       "step": 255
     },
     {
       "epoch": 0.2,
-      "learning_rate": 1.859104210708604e-05,
-      "loss": 0.2098,
+      "learning_rate": 1.9736100506148657e-05,
+      "loss": 0.1688,
       "step": 256
     },
     {
       "epoch": 0.2,
-      "learning_rate": 1.8578284986317592e-05,
-      "loss": 0.1746,
+      "learning_rate": 1.9733173290550494e-05,
+      "loss": 0.1959,
       "step": 257
     },
     {
       "epoch": 0.2,
-      "learning_rate": 1.8565474790147352e-05,
-      "loss": 0.1743,
+      "learning_rate": 1.9730230149293167e-05,
+      "loss": 0.1793,
       "step": 258
     },
     {
       "epoch": 0.2,
-      "learning_rate": 1.855261159783432e-05,
-      "loss": 0.1649,
+      "learning_rate": 1.9727271087192312e-05,
+      "loss": 0.1855,
       "step": 259
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.8539695488965396e-05,
-      "loss": 0.2058,
+      "epoch": 0.21,
+      "learning_rate": 1.9724296109089623e-05,
+      "loss": 0.2062,
       "step": 260
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.8526726543454883e-05,
-      "loss": 0.1575,
+      "epoch": 0.21,
+      "learning_rate": 1.9721305219852833e-05,
+      "loss": 0.1935,
       "step": 261
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.8513704841543997e-05,
-      "loss": 0.1764,
+      "epoch": 0.21,
+      "learning_rate": 1.971829842437571e-05,
+      "loss": 0.1766,
       "step": 262
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.8500630463800364e-05,
-      "loss": 0.1185,
+      "epoch": 0.21,
+      "learning_rate": 1.971527572757804e-05,
+      "loss": 0.1478,
       "step": 263
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.8487503491117523e-05,
-      "loss": 0.2048,
+      "epoch": 0.21,
+      "learning_rate": 1.971223713440564e-05,
+      "loss": 0.1705,
       "step": 264
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.847432400471443e-05,
-      "loss": 0.1627,
+      "epoch": 0.21,
+      "learning_rate": 1.970918264983034e-05,
+      "loss": 0.1625,
       "step": 265
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.8461092086134953e-05,
-      "loss": 0.1704,
+      "epoch": 0.21,
+      "learning_rate": 1.970611227884995e-05,
+      "loss": 0.1406,
       "step": 266
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.8447807817247368e-05,
-      "loss": 0.1682,
+      "epoch": 0.21,
+      "learning_rate": 1.9703026026488288e-05,
+      "loss": 0.2223,
       "step": 267
     },
     {
       "epoch": 0.21,
-      "learning_rate": 1.8434471280243854e-05,
-      "loss": 0.1464,
+      "learning_rate": 1.9699923897795165e-05,
+      "loss": 0.1574,
       "step": 268
     },
     {
       "epoch": 0.21,
-      "learning_rate": 1.8421082557639973e-05,
-      "loss": 0.1623,
+      "learning_rate": 1.9696805897846353e-05,
+      "loss": 0.1841,
       "step": 269
     },
     {
       "epoch": 0.21,
-      "learning_rate": 1.8407641732274174e-05,
-      "loss": 0.1785,
+      "learning_rate": 1.9693672031743604e-05,
+      "loss": 0.1978,
       "step": 270
     },
     {
       "epoch": 0.21,
-      "learning_rate": 1.8394148887307286e-05,
-      "loss": 0.1913,
+      "learning_rate": 1.9690522304614624e-05,
+      "loss": 0.1514,
       "step": 271
     },
     {
       "epoch": 0.21,
-      "learning_rate": 1.8380604106221973e-05,
-      "loss": 0.1482,
+      "learning_rate": 1.9687356721613084e-05,
+      "loss": 0.1738,
       "step": 272
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.8367007472822253e-05,
-      "loss": 0.1428,
+      "epoch": 0.22,
+      "learning_rate": 1.9684175287918576e-05,
+      "loss": 0.1966,
       "step": 273
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.8353359071232954e-05,
-      "loss": 0.14,
+      "epoch": 0.22,
+      "learning_rate": 1.968097800873665e-05,
+      "loss": 0.1744,
       "step": 274
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.833965898589921e-05,
-      "loss": 0.1552,
+      "epoch": 0.22,
+      "learning_rate": 1.9677764889298775e-05,
+      "loss": 0.2162,
       "step": 275
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.8325907301585924e-05,
-      "loss": 0.1795,
+      "epoch": 0.22,
+      "learning_rate": 1.9674535934862327e-05,
+      "loss": 0.1628,
       "step": 276
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.8312104103377266e-05,
-      "loss": 0.1943,
+      "epoch": 0.22,
+      "learning_rate": 1.967129115071061e-05,
+      "loss": 0.1842,
       "step": 277
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.8298249476676107e-05,
-      "loss": 0.2051,
+      "epoch": 0.22,
+      "learning_rate": 1.966803054215281e-05,
+      "loss": 0.1593,
       "step": 278
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.828434350720354e-05,
-      "loss": 0.1447,
+      "epoch": 0.22,
+      "learning_rate": 1.966475411452403e-05,
+      "loss": 0.1586,
       "step": 279
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.827038628099831e-05,
-      "loss": 0.1593,
+      "epoch": 0.22,
+      "learning_rate": 1.966146187318523e-05,
+      "loss": 0.134,
       "step": 280
     },
     {
       "epoch": 0.22,
-      "learning_rate": 1.8256377884416307e-05,
-      "loss": 0.2025,
+      "learning_rate": 1.9658153823523262e-05,
+      "loss": 0.2157,
       "step": 281
     },
     {
       "epoch": 0.22,
-      "learning_rate": 1.824231840413001e-05,
-      "loss": 0.1594,
+      "learning_rate": 1.9654829970950838e-05,
+      "loss": 0.1585,
       "step": 282
     },
     {
       "epoch": 0.22,
-      "learning_rate": 1.822820792712797e-05,
-      "loss": 0.2052,
+      "learning_rate": 1.965149032090653e-05,
+      "loss": 0.1978,
       "step": 283
     },
     {
       "epoch": 0.22,
-      "learning_rate": 1.8214046540714268e-05,
-      "loss": 0.1827,
+      "learning_rate": 1.9648134878854747e-05,
+      "loss": 0.1688,
       "step": 284
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.8199834332507955e-05,
-      "loss": 0.1561,
+      "epoch": 0.23,
+      "learning_rate": 1.9644763650285758e-05,
+      "loss": 0.1799,
       "step": 285
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.8185571390442542e-05,
-      "loss": 0.186,
+      "epoch": 0.23,
+      "learning_rate": 1.9641376640715646e-05,
+      "loss": 0.1987,
       "step": 286
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.8171257802765436e-05,
-      "loss": 0.1912,
+      "epoch": 0.23,
+      "learning_rate": 1.963797385568632e-05,
+      "loss": 0.1766,
       "step": 287
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.8156893658037386e-05,
-      "loss": 0.1793,
+      "epoch": 0.23,
+      "learning_rate": 1.96345553007655e-05,
+      "loss": 0.1429,
       "step": 288
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.8142479045131956e-05,
-      "loss": 0.1619,
+      "epoch": 0.23,
+      "learning_rate": 1.9631120981546713e-05,
+      "loss": 0.1824,
       "step": 289
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.812801405323497e-05,
-      "loss": 0.1592,
+      "epoch": 0.23,
+      "learning_rate": 1.9627670903649273e-05,
+      "loss": 0.1703,
       "step": 290
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.811349877184395e-05,
-      "loss": 0.1606,
+      "epoch": 0.23,
+      "learning_rate": 1.9624205072718285e-05,
+      "loss": 0.1364,
       "step": 291
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.809893329076757e-05,
-      "loss": 0.1774,
+      "epoch": 0.23,
+      "learning_rate": 1.9620723494424627e-05,
+      "loss": 0.1544,
       "step": 292
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.808431770012509e-05,
-      "loss": 0.1843,
+      "epoch": 0.23,
+      "learning_rate": 1.9617226174464945e-05,
+      "loss": 0.1559,
       "step": 293
     },
     {
       "epoch": 0.23,
-      "learning_rate": 1.8069652090345825e-05,
-      "loss": 0.1843,
+      "learning_rate": 1.9613713118561638e-05,
+      "loss": 0.2207,
       "step": 294
     },
     {
       "epoch": 0.23,
-      "learning_rate": 1.8054936552168548e-05,
-      "loss": 0.2021,
+      "learning_rate": 1.961018433246286e-05,
+      "loss": 0.2082,
       "step": 295
     },
     {
       "epoch": 0.23,
-      "learning_rate": 1.8040171176640967e-05,
-      "loss": 0.1893,
+      "learning_rate": 1.9606639821942496e-05,
+      "loss": 0.1779,
       "step": 296
     },
     {
       "epoch": 0.23,
-      "learning_rate": 1.8025356055119122e-05,
-      "loss": 0.1358,
+      "learning_rate": 1.9603079592800157e-05,
+      "loss": 0.1431,
       "step": 297
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.801049127926686e-05,
-      "loss": 0.1947,
+      "epoch": 0.24,
+      "learning_rate": 1.9599503650861183e-05,
+      "loss": 0.1925,
       "step": 298
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.7995576941055236e-05,
-      "loss": 0.1525,
+      "epoch": 0.24,
+      "learning_rate": 1.959591200197662e-05,
+      "loss": 0.1583,
       "step": 299
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.7980613132761956e-05,
-      "loss": 0.1676,
+      "epoch": 0.24,
+      "learning_rate": 1.9592304652023208e-05,
+      "loss": 0.1581,
       "step": 300
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.7965599946970814e-05,
-      "loss": 0.1277,
+      "epoch": 0.24,
+      "learning_rate": 1.9588681606903385e-05,
+      "loss": 0.1526,
       "step": 301
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.795053747657111e-05,
-      "loss": 0.152,
+      "epoch": 0.24,
+      "learning_rate": 1.9585042872545266e-05,
+      "loss": 0.2008,
       "step": 302
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.793542581475707e-05,
-      "loss": 0.1774,
+      "epoch": 0.24,
+      "learning_rate": 1.958138845490264e-05,
+      "loss": 0.1879,
       "step": 303
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.7920265055027285e-05,
-      "loss": 0.1532,
+      "epoch": 0.24,
+      "learning_rate": 1.9577718359954955e-05,
+      "loss": 0.1946,
       "step": 304
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.790505529118412e-05,
-      "loss": 0.1925,
+      "epoch": 0.24,
+      "learning_rate": 1.9574032593707314e-05,
+      "loss": 0.1762,
       "step": 305
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.788979661733313e-05,
-      "loss": 0.1935,
+      "epoch": 0.24,
+      "learning_rate": 1.957033116219045e-05,
+      "loss": 0.1531,
       "step": 306
     },
     {
       "epoch": 0.24,
-      "learning_rate": 1.7874489127882496e-05,
-      "loss": 0.1649,
+      "learning_rate": 1.956661407146075e-05,
+      "loss": 0.1547,
       "step": 307
     },
     {
       "epoch": 0.24,
-      "learning_rate": 1.7859132917542427e-05,
-      "loss": 0.1666,
+      "learning_rate": 1.9562881327600197e-05,
+      "loss": 0.1824,
       "step": 308
     },
     {
       "epoch": 0.24,
-      "learning_rate": 1.7843728081324575e-05,
-      "loss": 0.1657,
+      "learning_rate": 1.955913293671641e-05,
+      "loss": 0.1592,
       "step": 309
     },
     {
       "epoch": 0.24,
-      "learning_rate": 1.7828274714541445e-05,
-      "loss": 0.1985,
+      "learning_rate": 1.9555368904942593e-05,
+      "loss": 0.1917,
       "step": 310
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.781277291280582e-05,
-      "loss": 0.1566,
+      "epoch": 0.25,
+      "learning_rate": 1.9551589238437546e-05,
+      "loss": 0.1646,
       "step": 311
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.7797222772030155e-05,
-      "loss": 0.1545,
+      "epoch": 0.25,
+      "learning_rate": 1.954779394338566e-05,
+      "loss": 0.1711,
       "step": 312
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.7781624388425974e-05,
-      "loss": 0.1806,
+      "epoch": 0.25,
+      "learning_rate": 1.954398302599688e-05,
+      "loss": 0.165,
       "step": 313
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.7765977858503314e-05,
-      "loss": 0.1703,
+      "epoch": 0.25,
+      "learning_rate": 1.9540156492506734e-05,
+      "loss": 0.149,
       "step": 314
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.7750283279070077e-05,
-      "loss": 0.1627,
+      "epoch": 0.25,
+      "learning_rate": 1.9536314349176288e-05,
+      "loss": 0.1837,
       "step": 315
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.773454074723147e-05,
-      "loss": 0.188,
+      "epoch": 0.25,
+      "learning_rate": 1.9532456602292148e-05,
+      "loss": 0.144,
       "step": 316
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.771875036038939e-05,
-      "loss": 0.1873,
+      "epoch": 0.25,
+      "learning_rate": 1.952858325816646e-05,
+      "loss": 0.167,
       "step": 317
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.770291221624181e-05,
-      "loss": 0.1747,
+      "epoch": 0.25,
+      "learning_rate": 1.9524694323136883e-05,
+      "loss": 0.1427,
       "step": 318
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.76870264127822e-05,
-      "loss": 0.1897,
+      "epoch": 0.25,
+      "learning_rate": 1.952078980356659e-05,
+      "loss": 0.169,
       "step": 319
     },
     {
       "epoch": 0.25,
-      "learning_rate": 1.7671093048298905e-05,
-      "loss": 0.1751,
+      "learning_rate": 1.951686970584425e-05,
+      "loss": 0.1256,
       "step": 320
     },
     {
       "epoch": 0.25,
-      "learning_rate": 1.765511222137453e-05,
-      "loss": 0.1655,
+      "learning_rate": 1.9512934036384026e-05,
+      "loss": 0.1504,
       "step": 321
     },
     {
       "epoch": 0.25,
-      "learning_rate": 1.763908403088534e-05,
-      "loss": 0.1514,
+      "learning_rate": 1.9508982801625557e-05,
+      "loss": 0.1765,
       "step": 322
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.762300857600066e-05,
-      "loss": 0.1576,
+      "epoch": 0.26,
+      "learning_rate": 1.9505016008033953e-05,
+      "loss": 0.166,
       "step": 323
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.7606885956182228e-05,
-      "loss": 0.1656,
+      "epoch": 0.26,
+      "learning_rate": 1.950103366209978e-05,
+      "loss": 0.1697,
       "step": 324
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.759071627118362e-05,
-      "loss": 0.1844,
+      "epoch": 0.26,
+      "learning_rate": 1.949703577033905e-05,
+      "loss": 0.158,
       "step": 325
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.75744996210496e-05,
-      "loss": 0.1579,
+      "epoch": 0.26,
+      "learning_rate": 1.9493022339293207e-05,
+      "loss": 0.1729,
       "step": 326
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.7558236106115514e-05,
-      "loss": 0.1735,
+      "epoch": 0.26,
+      "learning_rate": 1.9488993375529137e-05,
+      "loss": 0.1765,
       "step": 327
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.754192582700668e-05,
-      "loss": 0.159,
+      "epoch": 0.26,
+      "learning_rate": 1.9484948885639122e-05,
+      "loss": 0.1379,
       "step": 328
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.752556888463774e-05,
-      "loss": 0.1721,
+      "epoch": 0.26,
+      "learning_rate": 1.948088887624086e-05,
+      "loss": 0.1697,
       "step": 329
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.750916538021206e-05,
-      "loss": 0.1626,
+      "epoch": 0.26,
+      "learning_rate": 1.9476813353977442e-05,
+      "loss": 0.175,
       "step": 330
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.7492715415221087e-05,
-      "loss": 0.1454,
+      "epoch": 0.26,
+      "learning_rate": 1.947272232551733e-05,
+      "loss": 0.1701,
       "step": 331
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.7476219091443738e-05,
-      "loss": 0.1846,
+      "epoch": 0.26,
+      "learning_rate": 1.9468615797554374e-05,
+      "loss": 0.1418,
       "step": 332
     },
     {
       "epoch": 0.26,
-      "learning_rate": 1.745967651094575e-05,
-      "loss": 0.17,
+      "learning_rate": 1.946449377680777e-05,
+      "loss": 0.1724,
       "step": 333
     },
     {
       "epoch": 0.26,
-      "learning_rate": 1.7443087776079068e-05,
-      "loss": 0.1989,
+      "learning_rate": 1.9460356270022073e-05,
+      "loss": 0.1642,
       "step": 334
     },
     {
       "epoch": 0.26,
-      "learning_rate": 1.7426452989481187e-05,
-      "loss": 0.1668,
+      "learning_rate": 1.945620328396718e-05,
+      "loss": 0.1498,
       "step": 335
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.740977225407455e-05,
-      "loss": 0.15,
+      "epoch": 0.27,
+      "learning_rate": 1.9452034825438302e-05,
+      "loss": 0.1409,
       "step": 336
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.739304567306588e-05,
-      "loss": 0.186,
+      "epoch": 0.27,
+      "learning_rate": 1.9447850901255975e-05,
+      "loss": 0.1533,
       "step": 337
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.7376273349945568e-05,
-      "loss": 0.2104,
+      "epoch": 0.27,
+      "learning_rate": 1.9443651518266044e-05,
+      "loss": 0.183,
       "step": 338
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.7359455388487008e-05,
-      "loss": 0.1669,
+      "epoch": 0.27,
+      "learning_rate": 1.943943668333964e-05,
+      "loss": 0.1507,
       "step": 339
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.7342591892745978e-05,
-      "loss": 0.1796,
+      "epoch": 0.27,
+      "learning_rate": 1.943520640337318e-05,
+      "loss": 0.1285,
       "step": 340
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.732568296705997e-05,
-      "loss": 0.1844,
+      "epoch": 0.27,
+      "learning_rate": 1.9430960685288355e-05,
+      "loss": 0.1725,
       "step": 341
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.7308728716047568e-05,
-      "loss": 0.2021,
+      "epoch": 0.27,
+      "learning_rate": 1.942669953603211e-05,
+      "loss": 0.1523,
       "step": 342
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.72917292446078e-05,
-      "loss": 0.1658,
+      "epoch": 0.27,
+      "learning_rate": 1.9422422962576646e-05,
+      "loss": 0.1959,
       "step": 343
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.727468465791946e-05,
-      "loss": 0.178,
+      "epoch": 0.27,
+      "learning_rate": 1.94181309719194e-05,
+      "loss": 0.1661,
       "step": 344
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.7257595061440502e-05,
-      "loss": 0.1884,
+      "epoch": 0.27,
+      "learning_rate": 1.941382357108303e-05,
+      "loss": 0.1532,
       "step": 345
     },
     {
       "epoch": 0.27,
-      "learning_rate": 1.7240460560907345e-05,
-      "loss": 0.1221,
+      "learning_rate": 1.9409500767115414e-05,
+      "loss": 0.1627,
       "step": 346
     },
     {
       "epoch": 0.27,
-      "learning_rate": 1.7223281262334252e-05,
-      "loss": 0.1917,
+      "learning_rate": 1.9405162567089627e-05,
+      "loss": 0.1507,
       "step": 347
     },
     {
       "epoch": 0.27,
-      "learning_rate": 1.7206057272012647e-05,
-      "loss": 0.2018,
+      "learning_rate": 1.9400808978103948e-05,
+      "loss": 0.1634,
       "step": 348
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.7188788696510477e-05,
-      "loss": 0.1758,
+      "epoch": 0.28,
+      "learning_rate": 1.939644000728182e-05,
+      "loss": 0.1603,
       "step": 349
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.7171475642671546e-05,
-      "loss": 0.1558,
+      "epoch": 0.28,
+      "learning_rate": 1.939205566177186e-05,
+      "loss": 0.1818,
       "step": 350
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.7154118217614844e-05,
-      "loss": 0.1481,
+      "epoch": 0.28,
+      "learning_rate": 1.938765594874785e-05,
+      "loss": 0.1424,
       "step": 351
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.7136716528733916e-05,
-      "loss": 0.1893,
+      "epoch": 0.28,
+      "learning_rate": 1.93832408754087e-05,
+      "loss": 0.1697,
       "step": 352
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.7119270683696144e-05,
-      "loss": 0.1838,
+      "epoch": 0.28,
+      "learning_rate": 1.937881044897847e-05,
+      "loss": 0.1724,
       "step": 353
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.710178079044214e-05,
-      "loss": 0.1898,
+      "epoch": 0.28,
+      "learning_rate": 1.937436467670633e-05,
+      "loss": 0.1618,
       "step": 354
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.7084246957185036e-05,
-      "loss": 0.1707,
+      "epoch": 0.28,
+      "learning_rate": 1.9369903565866565e-05,
+      "loss": 0.2152,
       "step": 355
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.706666929240983e-05,
-      "loss": 0.1949,
+      "epoch": 0.28,
+      "learning_rate": 1.936542712375855e-05,
+      "loss": 0.1396,
       "step": 356
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.7049047904872715e-05,
-      "loss": 0.1887,
+      "epoch": 0.28,
+      "learning_rate": 1.9360935357706756e-05,
+      "loss": 0.1671,
       "step": 357
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.703138290360041e-05,
-      "loss": 0.1765,
+      "epoch": 0.28,
+      "learning_rate": 1.9356428275060722e-05,
+      "loss": 0.1516,
       "step": 358
     },
     {
       "epoch": 0.28,
-      "learning_rate": 1.701367439788947e-05,
-      "loss": 0.1746,
+      "learning_rate": 1.9351905883195044e-05,
+      "loss": 0.1405,
       "step": 359
     },
     {
       "epoch": 0.28,
-      "learning_rate": 1.699592249730563e-05,
-      "loss": 0.1614,
+      "learning_rate": 1.934736818950937e-05,
+      "loss": 0.1584,
       "step": 360
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.6978127311683103e-05,
-      "loss": 0.2043,
+      "epoch": 0.29,
+      "learning_rate": 1.9342815201428394e-05,
+      "loss": 0.1736,
       "step": 361
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.6960288951123933e-05,
-      "loss": 0.1299,
+      "epoch": 0.29,
+      "learning_rate": 1.9338246926401828e-05,
+      "loss": 0.1623,
       "step": 362
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.6942407525997277e-05,
-      "loss": 0.1686,
+      "epoch": 0.29,
+      "learning_rate": 1.9333663371904388e-05,
+      "loss": 0.1874,
       "step": 363
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.6924483146938756e-05,
-      "loss": 0.1363,
+      "epoch": 0.29,
+      "learning_rate": 1.9329064545435803e-05,
+      "loss": 0.1735,
       "step": 364
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.6906515924849745e-05,
-      "loss": 0.1557,
+      "epoch": 0.29,
+      "learning_rate": 1.932445045452079e-05,
+      "loss": 0.1627,
       "step": 365
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.6888505970896696e-05,
-      "loss": 0.1445,
+      "epoch": 0.29,
+      "learning_rate": 1.931982110670904e-05,
+      "loss": 0.1602,
       "step": 366
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.6870453396510456e-05,
-      "loss": 0.1707,
+      "epoch": 0.29,
+      "learning_rate": 1.9315176509575196e-05,
+      "loss": 0.1691,
       "step": 367
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.6852358313385568e-05,
-      "loss": 0.1439,
+      "epoch": 0.29,
+      "learning_rate": 1.9310516670718877e-05,
+      "loss": 0.1584,
       "step": 368
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.683422083347959e-05,
-      "loss": 0.1688,
+      "epoch": 0.29,
+      "learning_rate": 1.9305841597764615e-05,
+      "loss": 0.1337,
       "step": 369
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.681604106901239e-05,
-      "loss": 0.1727,
+      "epoch": 0.29,
+      "learning_rate": 1.9301151298361887e-05,
+      "loss": 0.1743,
       "step": 370
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.6797819132465462e-05,
-      "loss": 0.1892,
+      "epoch": 0.29,
+      "learning_rate": 1.9296445780185077e-05,
+      "loss": 0.1762,
       "step": 371
     },
     {
       "epoch": 0.29,
-      "learning_rate": 1.6779555136581228e-05,
-      "loss": 0.1981,
+      "learning_rate": 1.929172505093347e-05,
+      "loss": 0.1425,
       "step": 372
     },
     {
       "epoch": 0.29,
-      "learning_rate": 1.676124919436233e-05,
-      "loss": 0.1728,
+      "learning_rate": 1.928698911833123e-05,
+      "loss": 0.1342,
       "step": 373
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.674290141907095e-05,
-      "loss": 0.179,
+      "epoch": 0.3,
+      "learning_rate": 1.9282237990127425e-05,
+      "loss": 0.1722,
       "step": 374
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.6724511924228093e-05,
-      "loss": 0.1432,
+      "epoch": 0.3,
+      "learning_rate": 1.9277471674095952e-05,
+      "loss": 0.1671,
       "step": 375
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.6706080823612897e-05,
-      "loss": 0.1774,
+      "epoch": 0.3,
+      "learning_rate": 1.927269017803559e-05,
+      "loss": 0.1612,
       "step": 376
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.668760823126191e-05,
-      "loss": 0.1309,
+      "epoch": 0.3,
+      "learning_rate": 1.9267893509769927e-05,
+      "loss": 0.1655,
       "step": 377
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.6669094261468405e-05,
-      "loss": 0.1511,
+      "epoch": 0.3,
+      "learning_rate": 1.92630816771474e-05,
+      "loss": 0.153,
       "step": 378
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.665053902878167e-05,
-      "loss": 0.1712,
+      "epoch": 0.3,
+      "learning_rate": 1.925825468804125e-05,
+      "loss": 0.1446,
       "step": 379
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.6631942648006284e-05,
-      "loss": 0.1736,
+      "epoch": 0.3,
+      "learning_rate": 1.9253412550349507e-05,
+      "loss": 0.1342,
       "step": 380
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.661330523420142e-05,
-      "loss": 0.1523,
-      "step": 381
+      "epoch": 0.3,
+      "learning_rate": 1.9248555271995006e-05,
+      "loss": 0.142,
+      "step": 381
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.659462690268013e-05,
-      "loss": 0.1563,
+      "epoch": 0.3,
+      "learning_rate": 1.924368286092534e-05,
+      "loss": 0.1945,
       "step": 382
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.6575907769008633e-05,
-      "loss": 0.1881,
+      "epoch": 0.3,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 0.1688,
       "step": 383
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.6557147949005597e-05,
-      "loss": 0.139,
+      "epoch": 0.3,
+      "learning_rate": 1.92338926725547e-05,
+      "loss": 0.2001,
       "step": 384
     },
     {
       "epoch": 0.3,
-      "learning_rate": 1.6538347558741424e-05,
-      "loss": 0.1664,
+      "learning_rate": 1.9228974911272682e-05,
+      "loss": 0.1851,
       "step": 385
     },
     {
       "epoch": 0.3,
-      "learning_rate": 1.651950671453753e-05,
-      "loss": 0.1446,
+      "learning_rate": 1.922404204931337e-05,
+      "loss": 0.1499,
       "step": 386
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.650062553296563e-05,
-      "loss": 0.1598,
+      "epoch": 0.31,
+      "learning_rate": 1.9219094094748035e-05,
+      "loss": 0.1622,
       "step": 387
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.6481704130847013e-05,
-      "loss": 0.1639,
+      "epoch": 0.31,
+      "learning_rate": 1.9214131055672648e-05,
+      "loss": 0.1304,
       "step": 388
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.646274262525181e-05,
-      "loss": 0.1521,
+      "epoch": 0.31,
+      "learning_rate": 1.9209152940207846e-05,
+      "loss": 0.1463,
       "step": 389
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.6443741133498297e-05,
-      "loss": 0.1823,
+      "epoch": 0.31,
+      "learning_rate": 1.920415975649896e-05,
+      "loss": 0.204,
       "step": 390
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.642469977315214e-05,
-      "loss": 0.143,
+      "epoch": 0.31,
+      "learning_rate": 1.9199151512715948e-05,
+      "loss": 0.1337,
       "step": 391
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.6405618662025678e-05,
-      "loss": 0.1797,
+      "epoch": 0.31,
+      "learning_rate": 1.919412821705343e-05,
+      "loss": 0.1659,
       "step": 392
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.6386497918177208e-05,
-      "loss": 0.1504,
+      "epoch": 0.31,
+      "learning_rate": 1.918908987773064e-05,
+      "loss": 0.1411,
       "step": 393
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.6367337659910223e-05,
-      "loss": 0.1612,
+      "epoch": 0.31,
+      "learning_rate": 1.918403650299144e-05,
+      "loss": 0.1749,
       "step": 394
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.6348138005772717e-05,
-      "loss": 0.1428,
+      "epoch": 0.31,
+      "learning_rate": 1.9178968101104285e-05,
+      "loss": 0.1552,
       "step": 395
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.6328899074556415e-05,
-      "loss": 0.1461,
+      "epoch": 0.31,
+      "learning_rate": 1.917388468036222e-05,
+      "loss": 0.1846,
       "step": 396
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.6309620985296075e-05,
-      "loss": 0.1414,
+      "epoch": 0.31,
+      "learning_rate": 1.9168786249082862e-05,
+      "loss": 0.1744,
       "step": 397
     },
     {
       "epoch": 0.31,
-      "learning_rate": 1.6290303857268724e-05,
-      "loss": 0.1598,
+      "learning_rate": 1.9163672815608392e-05,
+      "loss": 0.1283,
       "step": 398
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.6270947809992924e-05,
-      "loss": 0.1652,
+      "epoch": 0.32,
+      "learning_rate": 1.9158544388305534e-05,
+      "loss": 0.1553,
       "step": 399
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.625155296322805e-05,
-      "loss": 0.1763,
+      "epoch": 0.32,
+      "learning_rate": 1.915340097556555e-05,
+      "loss": 0.1501,
       "step": 400
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.6232119436973525e-05,
-      "loss": 0.1559,
+      "epoch": 0.32,
+      "learning_rate": 1.914824258580422e-05,
+      "loss": 0.1497,
       "step": 401
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.6212647351468088e-05,
-      "loss": 0.158,
+      "epoch": 0.32,
+      "learning_rate": 1.914306922746183e-05,
+      "loss": 0.14,
       "step": 402
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.6193136827189067e-05,
-      "loss": 0.1583,
+      "epoch": 0.32,
+      "learning_rate": 1.9137880909003155e-05,
+      "loss": 0.1942,
       "step": 403
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.6173587984851596e-05,
-      "loss": 0.1823,
+      "epoch": 0.32,
+      "learning_rate": 1.913267763891745e-05,
+      "loss": 0.183,
       "step": 404
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.6154000945407914e-05,
-      "loss": 0.165,
+      "epoch": 0.32,
+      "learning_rate": 1.912745942571843e-05,
+      "loss": 0.1493,
       "step": 405
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.6134375830046566e-05,
-      "loss": 0.1606,
+      "epoch": 0.32,
+      "learning_rate": 1.9122226277944276e-05,
+      "loss": 0.1907,
       "step": 406
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.6114712760191697e-05,
-      "loss": 0.1712,
+      "epoch": 0.32,
+      "learning_rate": 1.9116978204157583e-05,
+      "loss": 0.1612,
       "step": 407
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.6095011857502282e-05,
-      "loss": 0.2072,
+      "epoch": 0.32,
+      "learning_rate": 1.9111715212945384e-05,
+      "loss": 0.1524,
       "step": 408
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.607527324387137e-05,
-      "loss": 0.1614,
+      "epoch": 0.32,
+      "learning_rate": 1.9106437312919116e-05,
+      "loss": 0.1364,
       "step": 409
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.6055497041425333e-05,
-      "loss": 0.1514,
+      "epoch": 0.32,
+      "learning_rate": 1.910114451271461e-05,
+      "loss": 0.1536,
       "step": 410
     },
     {
       "epoch": 0.32,
-      "learning_rate": 1.603568337252312e-05,
-      "loss": 0.1722,
+      "learning_rate": 1.9095836820992074e-05,
+      "loss": 0.1406,
       "step": 411
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.6015832359755483e-05,
-      "loss": 0.1444,
+      "epoch": 0.33,
+      "learning_rate": 1.9090514246436085e-05,
+      "loss": 0.1769,
       "step": 412
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.5995944125944235e-05,
-      "loss": 0.137,
+      "epoch": 0.33,
+      "learning_rate": 1.9085176797755575e-05,
+      "loss": 0.1453,
       "step": 413
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.597601879414147e-05,
-      "loss": 0.1322,
+      "epoch": 0.33,
+      "learning_rate": 1.9079824483683808e-05,
+      "loss": 0.1497,
       "step": 414
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.5956056487628832e-05,
-      "loss": 0.1699,
+      "epoch": 0.33,
+      "learning_rate": 1.9074457312978373e-05,
+      "loss": 0.1499,
       "step": 415
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.5936057329916718e-05,
-      "loss": 0.1318,
+      "epoch": 0.33,
+      "learning_rate": 1.9069075294421163e-05,
+      "loss": 0.1514,
       "step": 416
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.5916021444743535e-05,
-      "loss": 0.1461,
+      "epoch": 0.33,
+      "learning_rate": 1.9063678436818372e-05,
+      "loss": 0.1563,
       "step": 417
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.5895948956074937e-05,
-      "loss": 0.1544,
+      "epoch": 0.33,
+      "learning_rate": 1.905826674900048e-05,
+      "loss": 0.1949,
       "step": 418
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.587583998810304e-05,
-      "loss": 0.2183,
+      "epoch": 0.33,
+      "learning_rate": 1.9052840239822218e-05,
+      "loss": 0.1924,
       "step": 419
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.5855694665245665e-05,
-      "loss": 0.1257,
+      "epoch": 0.33,
+      "learning_rate": 1.904739891816257e-05,
+      "loss": 0.1266,
       "step": 420
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.5835513112145583e-05,
-      "loss": 0.1715,
+      "epoch": 0.33,
+      "learning_rate": 1.9041942792924767e-05,
+      "loss": 0.1591,
       "step": 421
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.5815295453669702e-05,
-      "loss": 0.1283,
+      "epoch": 0.33,
+      "learning_rate": 1.903647187303626e-05,
+      "loss": 0.1517,
       "step": 422
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.579504181490834e-05,
-      "loss": 0.1766,
+      "epoch": 0.33,
+      "learning_rate": 1.9030986167448696e-05,
+      "loss": 0.1684,
       "step": 423
     },
     {
       "epoch": 0.33,
-      "learning_rate": 1.5774752321174428e-05,
-      "loss": 0.1359,
+      "learning_rate": 1.902548568513793e-05,
+      "loss": 0.1654,
       "step": 424
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.5754427098002735e-05,
-      "loss": 0.1571,
+      "epoch": 0.34,
+      "learning_rate": 1.9019970435103978e-05,
+      "loss": 0.1402,
       "step": 425
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.573406627114909e-05,
-      "loss": 0.16,
+      "epoch": 0.34,
+      "learning_rate": 1.9014440426371034e-05,
+      "loss": 0.1564,
       "step": 426
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.571366996658962e-05,
-      "loss": 0.199,
+      "epoch": 0.34,
+      "learning_rate": 1.9008895667987434e-05,
+      "loss": 0.1695,
       "step": 427
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.5693238310519952e-05,
-      "loss": 0.1494,
+      "epoch": 0.34,
+      "learning_rate": 1.9003336169025655e-05,
+      "loss": 0.1497,
       "step": 428
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.5672771429354436e-05,
-      "loss": 0.1779,
+      "epoch": 0.34,
+      "learning_rate": 1.8997761938582277e-05,
+      "loss": 0.1688,
       "step": 429
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.5652269449725375e-05,
-      "loss": 0.1879,
+      "epoch": 0.34,
+      "learning_rate": 1.8992172985778002e-05,
+      "loss": 0.1687,
       "step": 430
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.5631732498482222e-05,
-      "loss": 0.1899,
+      "epoch": 0.34,
+      "learning_rate": 1.8986569319757605e-05,
+      "loss": 0.1805,
       "step": 431
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.561116070269082e-05,
-      "loss": 0.1733,
+      "epoch": 0.34,
+      "learning_rate": 1.8980950949689952e-05,
+      "loss": 0.1572,
       "step": 432
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.5590554189632585e-05,
-      "loss": 0.132,
+      "epoch": 0.34,
+      "learning_rate": 1.897531788476795e-05,
+      "loss": 0.1469,
       "step": 433
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.5569913086803744e-05,
-      "loss": 0.1451,
+      "epoch": 0.34,
+      "learning_rate": 1.896967013420857e-05,
+      "loss": 0.1357,
       "step": 434
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.5549237521914537e-05,
-      "loss": 0.1439,
+      "epoch": 0.34,
+      "learning_rate": 1.896400770725279e-05,
+      "loss": 0.1371,
       "step": 435
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.552852762288843e-05,
-      "loss": 0.1458,
+      "epoch": 0.34,
+      "learning_rate": 1.8958330613165622e-05,
+      "loss": 0.15,
       "step": 436
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.5507783517861308e-05,
-      "loss": 0.1711,
+      "epoch": 0.35,
+      "learning_rate": 1.8952638861236066e-05,
+      "loss": 0.2294,
       "step": 437
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.548700533518071e-05,
-      "loss": 0.1427,
+      "epoch": 0.35,
+      "learning_rate": 1.8946932460777105e-05,
+      "loss": 0.1645,
       "step": 438
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.5466193203405017e-05,
-      "loss": 0.1492,
+      "epoch": 0.35,
+      "learning_rate": 1.894121142112569e-05,
+      "loss": 0.1445,
       "step": 439
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.5445347251302642e-05,
-      "loss": 0.1391,
+      "epoch": 0.35,
+      "learning_rate": 1.8935475751642736e-05,
+      "loss": 0.1751,
       "step": 440
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.5424467607851274e-05,
-      "loss": 0.1373,
+      "epoch": 0.35,
+      "learning_rate": 1.8929725461713083e-05,
+      "loss": 0.1408,
       "step": 441
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.540355440223704e-05,
-      "loss": 0.1694,
+      "epoch": 0.35,
+      "learning_rate": 1.8923960560745495e-05,
+      "loss": 0.1505,
       "step": 442
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.5382607763853736e-05,
-      "loss": 0.1696,
+      "epoch": 0.35,
+      "learning_rate": 1.891818105817265e-05,
+      "loss": 0.162,
       "step": 443
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.5361627822301997e-05,
-      "loss": 0.1806,
+      "epoch": 0.35,
+      "learning_rate": 1.891238696345111e-05,
+      "loss": 0.1582,
       "step": 444
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.534061470738852e-05,
-      "loss": 0.1771,
+      "epoch": 0.35,
+      "learning_rate": 1.8906578286061325e-05,
+      "loss": 0.1551,
       "step": 445
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.5319568549125253e-05,
-      "loss": 0.1438,
+      "epoch": 0.35,
+      "learning_rate": 1.890075503550758e-05,
+      "loss": 0.1599,
       "step": 446
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.5298489477728587e-05,
-      "loss": 0.1574,
+      "epoch": 0.35,
+      "learning_rate": 1.8894917221318038e-05,
+      "loss": 0.136,
       "step": 447
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.527737762361855e-05,
-      "loss": 0.1293,
+      "epoch": 0.35,
+      "learning_rate": 1.888906485304467e-05,
+      "loss": 0.1523,
       "step": 448
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.5256233117418e-05,
-      "loss": 0.1812,
+      "epoch": 0.35,
+      "learning_rate": 1.888319794026326e-05,
+      "loss": 0.1505,
       "step": 449
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.523505608995183e-05,
-      "loss": 0.1493,
+      "epoch": 0.36,
+      "learning_rate": 1.88773164925734e-05,
+      "loss": 0.1717,
       "step": 450
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.5213846672246139e-05,
-      "loss": 0.1402,
+      "epoch": 0.36,
+      "learning_rate": 1.887142051959847e-05,
+      "loss": 0.169,
       "step": 451
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.5192604995527433e-05,
-      "loss": 0.1629,
+      "epoch": 0.36,
+      "learning_rate": 1.8865510030985588e-05,
+      "loss": 0.1859,
       "step": 452
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.5171331191221814e-05,
-      "loss": 0.1923,
+      "epoch": 0.36,
+      "learning_rate": 1.8859585036405653e-05,
+      "loss": 0.1559,
       "step": 453
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.5150025390954153e-05,
-      "loss": 0.1578,
+      "epoch": 0.36,
+      "learning_rate": 1.885364554555329e-05,
+      "loss": 0.1896,
       "step": 454
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.5128687726547297e-05,
-      "loss": 0.1849,
+      "epoch": 0.36,
+      "learning_rate": 1.884769156814684e-05,
+      "loss": 0.1495,
       "step": 455
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.5107318330021237e-05,
-      "loss": 0.1554,
+      "epoch": 0.36,
+      "learning_rate": 1.884172311392834e-05,
+      "loss": 0.1598,
       "step": 456
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.5085917333592299e-05,
-      "loss": 0.1608,
+      "epoch": 0.36,
+      "learning_rate": 1.883574019266353e-05,
+      "loss": 0.1559,
       "step": 457
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.5064484869672318e-05,
-      "loss": 0.1281,
+      "epoch": 0.36,
+      "learning_rate": 1.8829742814141813e-05,
+      "loss": 0.1823,
       "step": 458
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.5043021070867825e-05,
-      "loss": 0.2036,
+      "epoch": 0.36,
+      "learning_rate": 1.882373098817625e-05,
+      "loss": 0.1664,
       "step": 459
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.5021526069979232e-05,
-      "loss": 0.1752,
+      "epoch": 0.36,
+      "learning_rate": 1.8817704724603536e-05,
+      "loss": 0.2166,
       "step": 460
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.5000000000000002e-05,
-      "loss": 0.1619,
+      "epoch": 0.36,
+      "learning_rate": 1.8811664033283993e-05,
+      "loss": 0.1855,
       "step": 461
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.4978442994115821e-05,
-      "loss": 0.158,
+      "epoch": 0.36,
+      "learning_rate": 1.880560892410155e-05,
+      "loss": 0.1426,
       "step": 462
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.4956855185703787e-05,
-      "loss": 0.1711,
+      "epoch": 0.37,
+      "learning_rate": 1.879953940696373e-05,
+      "loss": 0.1498,
       "step": 463
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.4935236708331582e-05,
-      "loss": 0.174,
+      "epoch": 0.37,
+      "learning_rate": 1.8793455491801623e-05,
+      "loss": 0.1874,
       "step": 464
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.4913587695756632e-05,
-      "loss": 0.1359,
+      "epoch": 0.37,
+      "learning_rate": 1.878735718856988e-05,
+      "loss": 0.1571,
       "step": 465
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.48919082819253e-05,
-      "loss": 0.1531,
+      "epoch": 0.37,
+      "learning_rate": 1.8781244507246706e-05,
+      "loss": 0.1672,
       "step": 466
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.4870198600972043e-05,
-      "loss": 0.1546,
+      "epoch": 0.37,
+      "learning_rate": 1.877511745783381e-05,
+      "loss": 0.132,
       "step": 467
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.4848458787218583e-05,
-      "loss": 0.1509,
+      "epoch": 0.37,
+      "learning_rate": 1.8768976050356428e-05,
+      "loss": 0.1661,
       "step": 468
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.4826688975173085e-05,
-      "loss": 0.1888,
+      "epoch": 0.37,
+      "learning_rate": 1.876282029486328e-05,
+      "loss": 0.1756,
       "step": 469
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.480488929952932e-05,
-      "loss": 0.1388,
+      "epoch": 0.37,
+      "learning_rate": 1.8756650201426565e-05,
+      "loss": 0.1601,
       "step": 470
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.4783059895165818e-05,
-      "loss": 0.1703,
+      "epoch": 0.37,
+      "learning_rate": 1.8750465780141946e-05,
+      "loss": 0.1748,
       "step": 471
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.4761200897145063e-05,
-      "loss": 0.175,
+      "epoch": 0.37,
+      "learning_rate": 1.8744267041128528e-05,
+      "loss": 0.1384,
       "step": 472
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.4739312440712627e-05,
-      "loss": 0.1566,
+      "epoch": 0.37,
+      "learning_rate": 1.8738053994528835e-05,
+      "loss": 0.193,
       "step": 473
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.4717394661296353e-05,
-      "loss": 0.1714,
+      "epoch": 0.37,
+      "learning_rate": 1.8731826650508812e-05,
+      "loss": 0.1457,
       "step": 474
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.4695447694505512e-05,
-      "loss": 0.2009,
+      "epoch": 0.38,
+      "learning_rate": 1.8725585019257794e-05,
+      "loss": 0.1629,
       "step": 475
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.467347167612995e-05,
-      "loss": 0.1402,
+      "epoch": 0.38,
+      "learning_rate": 1.8719329110988487e-05,
+      "loss": 0.1481,
       "step": 476
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.4651466742139284e-05,
-      "loss": 0.1637,
+      "epoch": 0.38,
+      "learning_rate": 1.871305893593697e-05,
+      "loss": 0.1263,
       "step": 477
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.4629433028682014e-05,
-      "loss": 0.1657,
+      "epoch": 0.38,
+      "learning_rate": 1.8706774504362655e-05,
+      "loss": 0.173,
       "step": 478
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.4607370672084717e-05,
-      "loss": 0.1577,
+      "epoch": 0.38,
+      "learning_rate": 1.8700475826548285e-05,
+      "loss": 0.1913,
       "step": 479
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.458527980885118e-05,
-      "loss": 0.1331,
+      "epoch": 0.38,
+      "learning_rate": 1.8694162912799917e-05,
+      "loss": 0.1556,
       "step": 480
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.456316057566158e-05,
-      "loss": 0.1654,
+      "epoch": 0.38,
+      "learning_rate": 1.868783577344689e-05,
+      "loss": 0.176,
       "step": 481
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.454101310937161e-05,
-      "loss": 0.1318,
+      "epoch": 0.38,
+      "learning_rate": 1.8681494418841825e-05,
+      "loss": 0.1497,
       "step": 482
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.4518837547011657e-05,
-      "loss": 0.1051,
+      "epoch": 0.38,
+      "learning_rate": 1.867513885936061e-05,
+      "loss": 0.1416,
       "step": 483
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.4496634025785938e-05,
-      "loss": 0.1705,
+      "epoch": 0.38,
+      "learning_rate": 1.8668769105402366e-05,
+      "loss": 0.1512,
       "step": 484
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.4474402683071659e-05,
-      "loss": 0.1568,
+      "epoch": 0.38,
+      "learning_rate": 1.8662385167389443e-05,
+      "loss": 0.1766,
       "step": 485
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.4452143656418157e-05,
-      "loss": 0.1725,
+      "epoch": 0.38,
+      "learning_rate": 1.8655987055767396e-05,
+      "loss": 0.1298,
       "step": 486
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.4429857083546054e-05,
-      "loss": 0.1793,
+      "epoch": 0.38,
+      "learning_rate": 1.864957478100498e-05,
+      "loss": 0.1977,
       "step": 487
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.4407543102346423e-05,
-      "loss": 0.1368,
+      "epoch": 0.39,
+      "learning_rate": 1.864314835359411e-05,
+      "loss": 0.1721,
       "step": 488
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.4385201850879895e-05,
-      "loss": 0.199,
+      "epoch": 0.39,
+      "learning_rate": 1.8636707784049867e-05,
+      "loss": 0.1337,
       "step": 489
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.4362833467375839e-05,
-      "loss": 0.1596,
+      "epoch": 0.39,
+      "learning_rate": 1.8630253082910473e-05,
+      "loss": 0.2045,
       "step": 490
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.4340438090231492e-05,
-      "loss": 0.1677,
+      "epoch": 0.39,
+      "learning_rate": 1.8623784260737272e-05,
+      "loss": 0.1795,
       "step": 491
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.4318015858011108e-05,
-      "loss": 0.1423,
+      "epoch": 0.39,
+      "learning_rate": 1.8617301328114704e-05,
+      "loss": 0.1454,
       "step": 492
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.429556690944509e-05,
-      "loss": 0.1628,
+      "epoch": 0.39,
+      "learning_rate": 1.861080429565031e-05,
+      "loss": 0.1906,
       "step": 493
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.427309138342915e-05,
-      "loss": 0.1359,
+      "epoch": 0.39,
+      "learning_rate": 1.8604293173974694e-05,
+      "loss": 0.1819,
       "step": 494
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.4250589419023435e-05,
-      "loss": 0.1379,
+      "epoch": 0.39,
+      "learning_rate": 1.8597767973741514e-05,
+      "loss": 0.1821,
       "step": 495
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.4228061155451671e-05,
-      "loss": 0.1494,
+      "epoch": 0.39,
+      "learning_rate": 1.8591228705627464e-05,
+      "loss": 0.1418,
       "step": 496
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.4205506732100303e-05,
-      "loss": 0.188,
+      "epoch": 0.39,
+      "learning_rate": 1.8584675380332254e-05,
+      "loss": 0.1899,
       "step": 497
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.4182926288517628e-05,
-      "loss": 0.1556,
+      "epoch": 0.39,
+      "learning_rate": 1.8578108008578603e-05,
+      "loss": 0.1476,
       "step": 498
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.4160319964412943e-05,
-      "loss": 0.1462,
+      "epoch": 0.39,
+      "learning_rate": 1.8571526601112202e-05,
+      "loss": 0.1467,
       "step": 499
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.4137687899655662e-05,
-      "loss": 0.1661,
+      "epoch": 0.39,
+      "learning_rate": 1.8564931168701713e-05,
+      "loss": 0.1395,
       "step": 500
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.4115030234274461e-05,
-      "loss": 0.1377,
+      "epoch": 0.4,
+      "learning_rate": 1.855832172213875e-05,
+      "loss": 0.1514,
       "step": 501
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.4092347108456425e-05,
-      "loss": 0.1414,
+      "epoch": 0.4,
+      "learning_rate": 1.855169827223785e-05,
+      "loss": 0.1458,
       "step": 502
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.4069638662546158e-05,
-      "loss": 0.1763,
+      "epoch": 0.4,
+      "learning_rate": 1.854506082983647e-05,
+      "loss": 0.1214,
       "step": 503
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.404690503704492e-05,
-      "loss": 0.1616,
+      "epoch": 0.4,
+      "learning_rate": 1.8538409405794952e-05,
+      "loss": 0.1198,
       "step": 504
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.402414637260977e-05,
-      "loss": 0.1792,
+      "epoch": 0.4,
+      "learning_rate": 1.8531744010996525e-05,
+      "loss": 0.1476,
       "step": 505
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.4001362810052684e-05,
-      "loss": 0.1776,
+      "epoch": 0.4,
+      "learning_rate": 1.8525064656347265e-05,
+      "loss": 0.1599,
       "step": 506
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.3978554490339687e-05,
-      "loss": 0.1617,
+      "epoch": 0.4,
+      "learning_rate": 1.8518371352776107e-05,
+      "loss": 0.1392,
       "step": 507
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.3955721554589979e-05,
-      "loss": 0.1276,
+      "epoch": 0.4,
+      "learning_rate": 1.85116641112348e-05,
+      "loss": 0.119,
       "step": 508
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.3932864144075064e-05,
-      "loss": 0.1528,
+      "epoch": 0.4,
+      "learning_rate": 1.850494294269789e-05,
+      "loss": 0.1575,
       "step": 509
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.3909982400217885e-05,
-      "loss": 0.1324,
+      "epoch": 0.4,
+      "learning_rate": 1.8498207858162724e-05,
+      "loss": 0.1932,
       "step": 510
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.388707646459193e-05,
-      "loss": 0.1289,
+      "epoch": 0.4,
+      "learning_rate": 1.8491458868649417e-05,
+      "loss": 0.1801,
       "step": 511
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.3864146478920366e-05,
-      "loss": 0.1123,
+      "epoch": 0.4,
+      "learning_rate": 1.8484695985200832e-05,
+      "loss": 0.1568,
       "step": 512
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.3841192585075164e-05,
-      "loss": 0.1663,
+      "epoch": 0.41,
+      "learning_rate": 1.847791921888256e-05,
+      "loss": 0.1716,
       "step": 513
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.3818214925076226e-05,
-      "loss": 0.1562,
+      "epoch": 0.41,
+      "learning_rate": 1.8471128580782923e-05,
+      "loss": 0.1691,
       "step": 514
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.3795213641090483e-05,
-      "loss": 0.1518,
+      "epoch": 0.41,
+      "learning_rate": 1.8464324082012926e-05,
+      "loss": 0.129,
       "step": 515
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.3772188875431054e-05,
-      "loss": 0.1689,
+      "epoch": 0.41,
+      "learning_rate": 1.845750573370626e-05,
+      "loss": 0.1507,
       "step": 516
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.3749140770556322e-05,
-      "loss": 0.1257,
+      "epoch": 0.41,
+      "learning_rate": 1.8450673547019273e-05,
+      "loss": 0.1345,
       "step": 517
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.3726069469069083e-05,
-      "loss": 0.1672,
+      "epoch": 0.41,
+      "learning_rate": 1.844382753313096e-05,
+      "loss": 0.1349,
       "step": 518
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.3702975113715661e-05,
-      "loss": 0.1568,
+      "epoch": 0.41,
+      "learning_rate": 1.8436967703242938e-05,
+      "loss": 0.1661,
       "step": 519
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.367985784738501e-05,
-      "loss": 0.1649,
+      "epoch": 0.41,
+      "learning_rate": 1.843009406857943e-05,
+      "loss": 0.141,
       "step": 520
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.365671781310784e-05,
-      "loss": 0.1476,
+      "epoch": 0.41,
+      "learning_rate": 1.842320664038725e-05,
+      "loss": 0.1467,
       "step": 521
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.3633555154055733e-05,
-      "loss": 0.16,
+      "epoch": 0.41,
+      "learning_rate": 1.8416305429935776e-05,
+      "loss": 0.1582,
       "step": 522
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.361037001354025e-05,
-      "loss": 0.1575,
+      "epoch": 0.41,
+      "learning_rate": 1.840939044851694e-05,
+      "loss": 0.1425,
       "step": 523
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.3587162535012051e-05,
-      "loss": 0.1755,
+      "epoch": 0.41,
+      "learning_rate": 1.8402461707445206e-05,
+      "loss": 0.1522,
       "step": 524
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.356393286206001e-05,
-      "loss": 0.1418,
+      "epoch": 0.41,
+      "learning_rate": 1.839551921805755e-05,
+      "loss": 0.1858,
       "step": 525
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.3540681138410317e-05,
-      "loss": 0.1648,
-      "step": 526
+      "epoch": 0.42,
+      "learning_rate": 1.8388562991713447e-05,
+      "loss": 0.1921,
+      "step": 526
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.3517407507925593e-05,
-      "loss": 0.1678,
+      "epoch": 0.42,
+      "learning_rate": 1.8381593039794846e-05,
+      "loss": 0.1597,
       "step": 527
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.3494112114604001e-05,
-      "loss": 0.1302,
+      "epoch": 0.42,
+      "learning_rate": 1.8374609373706156e-05,
+      "loss": 0.1163,
       "step": 528
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.3470795102578358e-05,
-      "loss": 0.1494,
+      "epoch": 0.42,
+      "learning_rate": 1.8367612004874224e-05,
+      "loss": 0.1674,
       "step": 529
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.3447456616115238e-05,
-      "loss": 0.15,
+      "epoch": 0.42,
+      "learning_rate": 1.8360600944748316e-05,
+      "loss": 0.1451,
       "step": 530
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.342409679961408e-05,
-      "loss": 0.1729,
+      "epoch": 0.42,
+      "learning_rate": 1.8353576204800106e-05,
+      "loss": 0.1906,
       "step": 531
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.3400715797606293e-05,
-      "loss": 0.1377,
+      "epoch": 0.42,
+      "learning_rate": 1.8346537796523643e-05,
+      "loss": 0.1694,
       "step": 532
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.3377313754754373e-05,
-      "loss": 0.1398,
+      "epoch": 0.42,
+      "learning_rate": 1.833948573143535e-05,
+      "loss": 0.1291,
       "step": 533
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.3353890815850993e-05,
-      "loss": 0.1501,
+      "epoch": 0.42,
+      "learning_rate": 1.8332420021073992e-05,
+      "loss": 0.1578,
       "step": 534
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.3330447125818115e-05,
-      "loss": 0.1294,
+      "epoch": 0.42,
+      "learning_rate": 1.8325340677000646e-05,
+      "loss": 0.1716,
       "step": 535
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.3306982829706093e-05,
-      "loss": 0.1425,
+      "epoch": 0.42,
+      "learning_rate": 1.8318247710798728e-05,
+      "loss": 0.1471,
       "step": 536
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.3283498072692771e-05,
-      "loss": 0.1606,
+      "epoch": 0.42,
+      "learning_rate": 1.831114113407391e-05,
+      "loss": 0.1838,
       "step": 537
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.3259993000082599e-05,
-      "loss": 0.1425,
+      "epoch": 0.42,
+      "learning_rate": 1.8304020958454156e-05,
+      "loss": 0.1962,
       "step": 538
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.3236467757305704e-05,
-      "loss": 0.1227,
+      "epoch": 0.43,
+      "learning_rate": 1.8296887195589678e-05,
+      "loss": 0.18,
       "step": 539
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.321292248991703e-05,
-      "loss": 0.1601,
+      "epoch": 0.43,
+      "learning_rate": 1.8289739857152903e-05,
+      "loss": 0.1538,
       "step": 540
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.3189357343595405e-05,
-      "loss": 0.1459,
+      "epoch": 0.43,
+      "learning_rate": 1.8282578954838493e-05,
+      "loss": 0.1528,
       "step": 541
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.3165772464142654e-05,
-      "loss": 0.1351,
+      "epoch": 0.43,
+      "learning_rate": 1.8275404500363293e-05,
+      "loss": 0.1511,
       "step": 542
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.3142167997482693e-05,
-      "loss": 0.1617,
+      "epoch": 0.43,
+      "learning_rate": 1.8268216505466318e-05,
+      "loss": 0.1474,
       "step": 543
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.3118544089660635e-05,
-      "loss": 0.1509,
+      "epoch": 0.43,
+      "learning_rate": 1.826101498190875e-05,
+      "loss": 0.1587,
       "step": 544
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.3094900886841867e-05,
-      "loss": 0.1467,
+      "epoch": 0.43,
+      "learning_rate": 1.8253799941473894e-05,
+      "loss": 0.194,
       "step": 545
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.3071238535311164e-05,
-      "loss": 0.1607,
+      "epoch": 0.43,
+      "learning_rate": 1.824657139596718e-05,
+      "loss": 0.1669,
       "step": 546
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.3047557181471784e-05,
-      "loss": 0.1551,
+      "epoch": 0.43,
+      "learning_rate": 1.8239329357216135e-05,
+      "loss": 0.1796,
       "step": 547
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.3023856971844543e-05,
-      "loss": 0.1519,
+      "epoch": 0.43,
+      "learning_rate": 1.823207383707036e-05,
+      "loss": 0.1567,
       "step": 548
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.3000138053066931e-05,
-      "loss": 0.1434,
+      "epoch": 0.43,
+      "learning_rate": 1.8224804847401518e-05,
+      "loss": 0.1602,
       "step": 549
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.2976400571892189e-05,
-      "loss": 0.1422,
+      "epoch": 0.43,
+      "learning_rate": 1.821752240010331e-05,
+      "loss": 0.1985,
       "step": 550
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.2952644675188413e-05,
-      "loss": 0.1603,
+      "epoch": 0.44,
+      "learning_rate": 1.8210226507091454e-05,
+      "loss": 0.1497,
       "step": 551
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.2928870509937633e-05,
-      "loss": 0.1939,
+      "epoch": 0.44,
+      "learning_rate": 1.8202917180303673e-05,
+      "loss": 0.1635,
       "step": 552
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.2905078223234907e-05,
-      "loss": 0.1395,
+      "epoch": 0.44,
+      "learning_rate": 1.819559443169967e-05,
+      "loss": 0.1484,
       "step": 553
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.288126796228743e-05,
-      "loss": 0.1453,
+      "epoch": 0.44,
+      "learning_rate": 1.8188258273261104e-05,
+      "loss": 0.1664,
       "step": 554
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.2857439874413589e-05,
-      "loss": 0.1609,
+      "epoch": 0.44,
+      "learning_rate": 1.818090871699158e-05,
+      "loss": 0.1544,
       "step": 555
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.2833594107042078e-05,
-      "loss": 0.1581,
+      "epoch": 0.44,
+      "learning_rate": 1.8173545774916628e-05,
+      "loss": 0.1241,
       "step": 556
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.2809730807710973e-05,
-      "loss": 0.1776,
+      "epoch": 0.44,
+      "learning_rate": 1.8166169459083673e-05,
+      "loss": 0.1646,
       "step": 557
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.2785850124066829e-05,
-      "loss": 0.1639,
+      "epoch": 0.44,
+      "learning_rate": 1.8158779781562022e-05,
+      "loss": 0.1454,
       "step": 558
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.2761952203863759e-05,
-      "loss": 0.1727,
+      "epoch": 0.44,
+      "learning_rate": 1.8151376754442856e-05,
+      "loss": 0.1318,
       "step": 559
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.2738037194962516e-05,
-      "loss": 0.1725,
+      "epoch": 0.44,
+      "learning_rate": 1.8143960389839184e-05,
+      "loss": 0.1398,
       "step": 560
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.2714105245329591e-05,
-      "loss": 0.1625,
+      "epoch": 0.44,
+      "learning_rate": 1.8136530699885852e-05,
+      "loss": 0.1682,
       "step": 561
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.2690156503036288e-05,
-      "loss": 0.1178,
+      "epoch": 0.44,
+      "learning_rate": 1.8129087696739497e-05,
+      "loss": 0.1721,
       "step": 562
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.266619111625781e-05,
-      "loss": 0.1387,
+      "epoch": 0.44,
+      "learning_rate": 1.8121631392578545e-05,
+      "loss": 0.133,
       "step": 563
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.264220923327234e-05,
-      "loss": 0.1333,
+      "epoch": 0.45,
+      "learning_rate": 1.8114161799603195e-05,
+      "loss": 0.1346,
       "step": 564
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.2618211002460135e-05,
-      "loss": 0.1608,
+      "epoch": 0.45,
+      "learning_rate": 1.810667893003537e-05,
+      "loss": 0.1724,
       "step": 565
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.2594196572302586e-05,
-      "loss": 0.1705,
+      "epoch": 0.45,
+      "learning_rate": 1.8099182796118727e-05,
+      "loss": 0.1505,
       "step": 566
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.257016609138132e-05,
-      "loss": 0.1398,
+      "epoch": 0.45,
+      "learning_rate": 1.8091673410118633e-05,
+      "loss": 0.1233,
       "step": 567
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.2546119708377273e-05,
-      "loss": 0.1556,
+      "epoch": 0.45,
+      "learning_rate": 1.8084150784322123e-05,
+      "loss": 0.1431,
       "step": 568
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.2522057572069772e-05,
-      "loss": 0.1579,
+      "epoch": 0.45,
+      "learning_rate": 1.8076614931037908e-05,
+      "loss": 0.168,
       "step": 569
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.2497979831335604e-05,
-      "loss": 0.1472,
+      "epoch": 0.45,
+      "learning_rate": 1.8069065862596338e-05,
+      "loss": 0.1528,
       "step": 570
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.2473886635148109e-05,
-      "loss": 0.1385,
+      "epoch": 0.45,
+      "learning_rate": 1.8061503591349386e-05,
+      "loss": 0.1581,
       "step": 571
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.2449778132576256e-05,
-      "loss": 0.1447,
+      "epoch": 0.45,
+      "learning_rate": 1.8053928129670624e-05,
+      "loss": 0.1442,
       "step": 572
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.2425654472783712e-05,
-      "loss": 0.1469,
+      "epoch": 0.45,
+      "learning_rate": 1.8046339489955214e-05,
+      "loss": 0.1739,
       "step": 573
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.2401515805027924e-05,
-      "loss": 0.1749,
+      "epoch": 0.45,
+      "learning_rate": 1.8038737684619874e-05,
+      "loss": 0.137,
       "step": 574
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.2377362278659202e-05,
-      "loss": 0.1406,
+      "epoch": 0.45,
+      "learning_rate": 1.8031122726102868e-05,
+      "loss": 0.1691,
       "step": 575
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.2353194043119783e-05,
-      "loss": 0.1743,
+      "epoch": 0.45,
+      "learning_rate": 1.8023494626863976e-05,
+      "loss": 0.1083,
       "step": 576
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.2329011247942915e-05,
-      "loss": 0.1695,
+      "epoch": 0.46,
+      "learning_rate": 1.8015853399384488e-05,
+      "loss": 0.1469,
       "step": 577
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.2304814042751927e-05,
-      "loss": 0.1807,
+      "epoch": 0.46,
+      "learning_rate": 1.8008199056167167e-05,
+      "loss": 0.1541,
       "step": 578
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.2280602577259312e-05,
-      "loss": 0.1356,
+      "epoch": 0.46,
+      "learning_rate": 1.8000531609736236e-05,
+      "loss": 0.1178,
       "step": 579
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.2256377001265785e-05,
-      "loss": 0.1706,
+      "epoch": 0.46,
+      "learning_rate": 1.7992851072637366e-05,
+      "loss": 0.1793,
       "step": 580
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.2232137464659368e-05,
-      "loss": 0.1539,
+      "epoch": 0.46,
+      "learning_rate": 1.798515745743764e-05,
+      "loss": 0.176,
       "step": 581
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.2207884117414465e-05,
-      "loss": 0.1608,
+      "epoch": 0.46,
+      "learning_rate": 1.797745077672554e-05,
+      "loss": 0.1898,
       "step": 582
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.2183617109590925e-05,
-      "loss": 0.1359,
+      "epoch": 0.46,
+      "learning_rate": 1.7969731043110928e-05,
+      "loss": 0.1929,
       "step": 583
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.215933659133311e-05,
-      "loss": 0.142,
+      "epoch": 0.46,
+      "learning_rate": 1.7961998269225024e-05,
+      "loss": 0.1379,
       "step": 584
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.213504271286899e-05,
-      "loss": 0.1407,
+      "epoch": 0.46,
+      "learning_rate": 1.7954252467720386e-05,
+      "loss": 0.1708,
       "step": 585
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.2110735624509184e-05,
-      "loss": 0.175,
+      "epoch": 0.46,
+      "learning_rate": 1.7946493651270883e-05,
+      "loss": 0.1544,
       "step": 586
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.2086415476646047e-05,
-      "loss": 0.1622,
+      "epoch": 0.46,
+      "learning_rate": 1.7938721832571688e-05,
+      "loss": 0.1524,
       "step": 587
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.2062082419752737e-05,
-      "loss": 0.1487,
+      "epoch": 0.46,
+      "learning_rate": 1.793093702433924e-05,
+      "loss": 0.1609,
       "step": 588
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.2037736604382279e-05,
-      "loss": 0.1431,
+      "epoch": 0.47,
+      "learning_rate": 1.792313923931123e-05,
+      "loss": 0.1651,
       "step": 589
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.201337818116664e-05,
-      "loss": 0.1998,
+      "epoch": 0.47,
+      "learning_rate": 1.7915328490246594e-05,
+      "loss": 0.1479,
       "step": 590
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.1989007300815793e-05,
-      "loss": 0.1755,
+      "epoch": 0.47,
+      "learning_rate": 1.7907504789925473e-05,
+      "loss": 0.164,
       "step": 591
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.1964624114116784e-05,
-      "loss": 0.1401,
+      "epoch": 0.47,
+      "learning_rate": 1.789966815114919e-05,
+      "loss": 0.1995,
       "step": 592
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.1940228771932808e-05,
-      "loss": 0.1581,
+      "epoch": 0.47,
+      "learning_rate": 1.789181858674026e-05,
+      "loss": 0.1652,
       "step": 593
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.1915821425202258e-05,
-      "loss": 0.1665,
+      "epoch": 0.47,
+      "learning_rate": 1.788395610954233e-05,
+      "loss": 0.1424,
       "step": 594
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.1891402224937805e-05,
-      "loss": 0.1521,
+      "epoch": 0.47,
+      "learning_rate": 1.7876080732420176e-05,
+      "loss": 0.1594,
       "step": 595
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.1866971322225462e-05,
-      "loss": 0.149,
+      "epoch": 0.47,
+      "learning_rate": 1.7868192468259686e-05,
+      "loss": 0.1821,
       "step": 596
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.1842528868223645e-05,
-      "loss": 0.1561,
+      "epoch": 0.47,
+      "learning_rate": 1.7860291329967842e-05,
+      "loss": 0.1387,
       "step": 597
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.1818075014162243e-05,
-      "loss": 0.1748,
+      "epoch": 0.47,
+      "learning_rate": 1.7852377330472668e-05,
+      "loss": 0.1544,
       "step": 598
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.1793609911341672e-05,
-      "loss": 0.1284,
+      "epoch": 0.47,
+      "learning_rate": 1.7844450482723258e-05,
+      "loss": 0.1847,
       "step": 599
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.176913371113196e-05,
-      "loss": 0.16,
+      "epoch": 0.47,
+      "learning_rate": 1.783651079968971e-05,
+      "loss": 0.1418,
       "step": 600
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.1744646564971777e-05,
-      "loss": 0.15,
+      "epoch": 0.47,
+      "learning_rate": 1.782855829436313e-05,
+      "loss": 0.1604,
       "step": 601
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.1720148624367533e-05,
-      "loss": 0.1507,
+      "epoch": 0.48,
+      "learning_rate": 1.7820592979755605e-05,
+      "loss": 0.1713,
       "step": 602
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.1695640040892424e-05,
-      "loss": 0.1599,
+      "epoch": 0.48,
+      "learning_rate": 1.7812614868900185e-05,
+      "loss": 0.1426,
       "step": 603
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.1671120966185486e-05,
-      "loss": 0.1293,
+      "epoch": 0.48,
+      "learning_rate": 1.7804623974850844e-05,
+      "loss": 0.1388,
       "step": 604
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.1646591551950677e-05,
-      "loss": 0.1552,
+      "epoch": 0.48,
+      "learning_rate": 1.779662031068249e-05,
+      "loss": 0.1506,
       "step": 605
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.1622051949955915e-05,
-      "loss": 0.1371,
+      "epoch": 0.48,
+      "learning_rate": 1.7788603889490907e-05,
+      "loss": 0.1573,
       "step": 606
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.159750231203217e-05,
-      "loss": 0.1745,
+      "epoch": 0.48,
+      "learning_rate": 1.778057472439277e-05,
+      "loss": 0.1416,
       "step": 607
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.1572942790072486e-05,
-      "loss": 0.1589,
+      "epoch": 0.48,
+      "learning_rate": 1.7772532828525593e-05,
+      "loss": 0.1426,
       "step": 608
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.154837353603107e-05,
-      "loss": 0.1422,
+      "epoch": 0.48,
+      "learning_rate": 1.7764478215047725e-05,
+      "loss": 0.1471,
       "step": 609
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.1523794701922351e-05,
-      "loss": 0.1644,
+      "epoch": 0.48,
+      "learning_rate": 1.7756410897138326e-05,
+      "loss": 0.1268,
       "step": 610
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.149920643982002e-05,
-      "loss": 0.1649,
+      "epoch": 0.48,
+      "learning_rate": 1.7748330887997344e-05,
+      "loss": 0.175,
       "step": 611
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.1474608901856103e-05,
-      "loss": 0.1237,
+      "epoch": 0.48,
+      "learning_rate": 1.7740238200845485e-05,
+      "loss": 0.2045,
       "step": 612
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.145000224022002e-05,
-      "loss": 0.1696,
+      "epoch": 0.48,
+      "learning_rate": 1.7732132848924206e-05,
+      "loss": 0.1685,
       "step": 613
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.1425386607157642e-05,
-      "loss": 0.1442,
+      "epoch": 0.48,
+      "learning_rate": 1.7724014845495684e-05,
+      "loss": 0.1867,
       "step": 614
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.1400762154970346e-05,
-      "loss": 0.1315,
+      "epoch": 0.49,
+      "learning_rate": 1.77158842038428e-05,
+      "loss": 0.1667,
       "step": 615
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.1376129036014073e-05,
-      "loss": 0.1257,
+      "epoch": 0.49,
+      "learning_rate": 1.7707740937269108e-05,
+      "loss": 0.1667,
       "step": 616
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.1351487402698388e-05,
-      "loss": 0.1306,
+      "epoch": 0.49,
+      "learning_rate": 1.769958505909882e-05,
+      "loss": 0.1607,
       "step": 617
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.1326837407485537e-05,
-      "loss": 0.1968,
+      "epoch": 0.49,
+      "learning_rate": 1.7691416582676792e-05,
+      "loss": 0.1376,
       "step": 618
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.1302179202889505e-05,
-      "loss": 0.1537,
+      "epoch": 0.49,
+      "learning_rate": 1.7683235521368484e-05,
+      "loss": 0.1528,
       "step": 619
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.1277512941475059e-05,
-      "loss": 0.1595,
+      "epoch": 0.49,
+      "learning_rate": 1.7675041888559952e-05,
+      "loss": 0.1846,
       "step": 620
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.1252838775856829e-05,
-      "loss": 0.1656,
+      "epoch": 0.49,
+      "learning_rate": 1.7666835697657824e-05,
+      "loss": 0.1604,
       "step": 621
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.1228156858698344e-05,
-      "loss": 0.1618,
+      "epoch": 0.49,
+      "learning_rate": 1.7658616962089262e-05,
+      "loss": 0.1447,
       "step": 622
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.1203467342711087e-05,
-      "loss": 0.1572,
+      "epoch": 0.49,
+      "learning_rate": 1.765038569530198e-05,
+      "loss": 0.1818,
       "step": 623
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.1178770380653567e-05,
-      "loss": 0.1497,
+      "epoch": 0.49,
+      "learning_rate": 1.7642141910764164e-05,
+      "loss": 0.1523,
       "step": 624
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.1154066125330358e-05,
-      "loss": 0.1502,
+      "epoch": 0.49,
+      "learning_rate": 1.7633885621964516e-05,
+      "loss": 0.1609,
       "step": 625
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.1129354729591158e-05,
-      "loss": 0.1497,
+      "epoch": 0.49,
+      "learning_rate": 1.7625616842412166e-05,
+      "loss": 0.1875,
       "step": 626
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.1104636346329838e-05,
-      "loss": 0.1498,
+      "epoch": 0.5,
+      "learning_rate": 1.76173355856367e-05,
+      "loss": 0.1528,
       "step": 627
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.107991112848352e-05,
-      "loss": 0.1829,
+      "epoch": 0.5,
+      "learning_rate": 1.7609041865188122e-05,
+      "loss": 0.1871,
       "step": 628
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.1055179229031598e-05,
-      "loss": 0.1269,
+      "epoch": 0.5,
+      "learning_rate": 1.7600735694636814e-05,
+      "loss": 0.1522,
       "step": 629
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.1030440800994806e-05,
-      "loss": 0.177,
+      "epoch": 0.5,
+      "learning_rate": 1.759241708757354e-05,
+      "loss": 0.1302,
       "step": 630
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.100569599743428e-05,
-      "loss": 0.1429,
+      "epoch": 0.5,
+      "learning_rate": 1.7584086057609413e-05,
+      "loss": 0.1486,
       "step": 631
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.09809449714506e-05,
-      "loss": 0.1397,
+      "epoch": 0.5,
+      "learning_rate": 1.757574261837587e-05,
+      "loss": 0.1343,
       "step": 632
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.0956187876182837e-05,
-      "loss": 0.1304,
+      "epoch": 0.5,
+      "learning_rate": 1.7567386783524655e-05,
+      "loss": 0.1525,
       "step": 633
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.0931424864807624e-05,
-      "loss": 0.1517,
+      "epoch": 0.5,
+      "learning_rate": 1.7559018566727788e-05,
+      "loss": 0.1338,
       "step": 634
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.0906656090538196e-05,
-      "loss": 0.1763,
+      "epoch": 0.5,
+      "learning_rate": 1.755063798167755e-05,
+      "loss": 0.1312,
       "step": 635
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.0881881706623443e-05,
-      "loss": 0.1377,
+      "epoch": 0.5,
+      "learning_rate": 1.754224504208647e-05,
+      "loss": 0.1557,
       "step": 636
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.0857101866346953e-05,
-      "loss": 0.1616,
+      "epoch": 0.5,
+      "learning_rate": 1.7533839761687278e-05,
+      "loss": 0.1354,
       "step": 637
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.0832316723026092e-05,
-      "loss": 0.1833,
+      "epoch": 0.5,
+      "learning_rate": 1.7525422154232906e-05,
+      "loss": 0.1221,
       "step": 638
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.0807526430011028e-05,
-      "loss": 0.1581,
+      "epoch": 0.5,
+      "learning_rate": 1.7516992233496443e-05,
+      "loss": 0.1337,
       "step": 639
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.0782731140683786e-05,
-      "loss": 0.1369,
+      "epoch": 0.51,
+      "learning_rate": 1.7508550013271146e-05,
+      "loss": 0.1564,
       "step": 640
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.0757931008457308e-05,
-      "loss": 0.1349,
+      "epoch": 0.51,
+      "learning_rate": 1.7500095507370376e-05,
+      "loss": 0.1733,
       "step": 641
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.0733126186774504e-05,
-      "loss": 0.1469,
+      "epoch": 0.51,
+      "learning_rate": 1.749162872962761e-05,
+      "loss": 0.1567,
       "step": 642
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.0708316829107295e-05,
-      "loss": 0.1632,
+      "epoch": 0.51,
+      "learning_rate": 1.7483149693896396e-05,
+      "loss": 0.1643,
       "step": 643
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.0683503088955663e-05,
-      "loss": 0.1455,
+      "epoch": 0.51,
+      "learning_rate": 1.7474658414050344e-05,
+      "loss": 0.1685,
       "step": 644
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.0658685119846713e-05,
-      "loss": 0.178,
+      "epoch": 0.51,
+      "learning_rate": 1.7466154903983092e-05,
+      "loss": 0.1618,
       "step": 645
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.0633863075333713e-05,
-      "loss": 0.1727,
+      "epoch": 0.51,
+      "learning_rate": 1.74576391776083e-05,
+      "loss": 0.1741,
       "step": 646
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.0609037108995143e-05,
-      "loss": 0.1544,
+      "epoch": 0.51,
+      "learning_rate": 1.744911124885961e-05,
+      "loss": 0.1597,
       "step": 647
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.058420737443375e-05,
-      "loss": 0.1403,
+      "epoch": 0.51,
+      "learning_rate": 1.7440571131690626e-05,
+      "loss": 0.1419,
       "step": 648
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.0559374025275597e-05,
-      "loss": 0.1364,
+      "epoch": 0.51,
+      "learning_rate": 1.7432018840074905e-05,
+      "loss": 0.1508,
       "step": 649
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.0534537215169106e-05,
-      "loss": 0.1556,
+      "epoch": 0.51,
+      "learning_rate": 1.742345438800591e-05,
+      "loss": 0.1378,
       "step": 650
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.0509697097784122e-05,
-      "loss": 0.1201,
+      "epoch": 0.51,
+      "learning_rate": 1.7414877789497017e-05,
+      "loss": 0.1457,
       "step": 651
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.0484853826810942e-05,
-      "loss": 0.1265,
+      "epoch": 0.52,
+      "learning_rate": 1.7406289058581466e-05,
+      "loss": 0.1361,
       "step": 652
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.0460007555959384e-05,
-      "loss": 0.1558,
+      "epoch": 0.52,
+      "learning_rate": 1.739768820931235e-05,
+      "loss": 0.1902,
       "step": 653
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.043515843895782e-05,
-      "loss": 0.1304,
+      "epoch": 0.52,
+      "learning_rate": 1.7389075255762592e-05,
+      "loss": 0.192,
       "step": 654
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.0410306629552231e-05,
-      "loss": 0.1529,
+      "epoch": 0.52,
+      "learning_rate": 1.7380450212024924e-05,
+      "loss": 0.1248,
       "step": 655
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.0385452281505269e-05,
-      "loss": 0.149,
+      "epoch": 0.52,
+      "learning_rate": 1.737181309221185e-05,
+      "loss": 0.1322,
       "step": 656
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.0360595548595274e-05,
-      "loss": 0.14,
+      "epoch": 0.52,
+      "learning_rate": 1.7363163910455646e-05,
+      "loss": 0.1592,
       "step": 657
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.0335736584615357e-05,
-      "loss": 0.1418,
+      "epoch": 0.52,
+      "learning_rate": 1.735450268090831e-05,
+      "loss": 0.142,
       "step": 658
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.0310875543372425e-05,
-      "loss": 0.1474,
+      "epoch": 0.52,
+      "learning_rate": 1.7345829417741564e-05,
+      "loss": 0.1517,
       "step": 659
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.0286012578686244e-05,
-      "loss": 0.2043,
+      "epoch": 0.52,
+      "learning_rate": 1.7337144135146818e-05,
+      "loss": 0.1356,
       "step": 660
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.0261147844388472e-05,
-      "loss": 0.1749,
+      "epoch": 0.52,
+      "learning_rate": 1.7328446847335142e-05,
+      "loss": 0.1697,
       "step": 661
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.023628149432172e-05,
-      "loss": 0.1605,
+      "epoch": 0.52,
+      "learning_rate": 1.731973756853726e-05,
+      "loss": 0.1617,
       "step": 662
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.0211413682338603e-05,
-      "loss": 0.1263,
+      "epoch": 0.52,
+      "learning_rate": 1.73110163130035e-05,
+      "loss": 0.1167,
       "step": 663
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.0186544562300766e-05,
-      "loss": 0.129,
+      "epoch": 0.52,
+      "learning_rate": 1.7302283095003807e-05,
+      "loss": 0.1485,
       "step": 664
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.0161674288077959e-05,
-      "loss": 0.1573,
+      "epoch": 0.53,
+      "learning_rate": 1.729353792882768e-05,
+      "loss": 0.2223,
       "step": 665
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.0136803013547071e-05,
-      "loss": 0.158,
+      "epoch": 0.53,
+      "learning_rate": 1.728478082878418e-05,
+      "loss": 0.1915,
       "step": 666
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.011193089259118e-05,
-      "loss": 0.1378,
+      "epoch": 0.53,
+      "learning_rate": 1.7276011809201896e-05,
+      "loss": 0.1484,
       "step": 667
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.0087058079098595e-05,
-      "loss": 0.1514,
+      "epoch": 0.53,
+      "learning_rate": 1.7267230884428905e-05,
+      "loss": 0.1437,
       "step": 668
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.0062184726961918e-05,
-      "loss": 0.1923,
+      "epoch": 0.53,
+      "learning_rate": 1.725843806883278e-05,
+      "loss": 0.1635,
       "step": 669
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.0037310990077083e-05,
-      "loss": 0.1526,
+      "epoch": 0.53,
+      "learning_rate": 1.7249633376800542e-05,
+      "loss": 0.169,
       "step": 670
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.0012437022342399e-05,
-      "loss": 0.1386,
+      "epoch": 0.53,
+      "learning_rate": 1.7240816822738646e-05,
+      "loss": 0.1634,
       "step": 671
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.987562977657605e-06,
-      "loss": 0.1846,
+      "epoch": 0.53,
+      "learning_rate": 1.7231988421072957e-05,
+      "loss": 0.1646,
       "step": 672
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.962689009922918e-06,
-      "loss": 0.1565,
+      "epoch": 0.53,
+      "learning_rate": 1.722314818624872e-05,
+      "loss": 0.1585,
       "step": 673
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.937815273038082e-06,
-      "loss": 0.1765,
+      "epoch": 0.53,
+      "learning_rate": 1.7214296132730555e-05,
+      "loss": 0.1649,
       "step": 674
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.912941920901408e-06,
-      "loss": 0.1717,
+      "epoch": 0.53,
+      "learning_rate": 1.7205432275002403e-05,
+      "loss": 0.1606,
       "step": 675
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.888069107408824e-06,
-      "loss": 0.1339,
+      "epoch": 0.53,
+      "learning_rate": 1.719655662756753e-05,
+      "loss": 0.1613,
       "step": 676
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.86319698645293e-06,
-      "loss": 0.1597,
+      "epoch": 0.53,
+      "learning_rate": 1.7187669204948495e-05,
+      "loss": 0.1891,
       "step": 677
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.838325711922045e-06,
-      "loss": 0.1493,
+      "epoch": 0.54,
+      "learning_rate": 1.7178770021687113e-05,
+      "loss": 0.1519,
       "step": 678
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.813455437699238e-06,
-      "loss": 0.1155,
+      "epoch": 0.54,
+      "learning_rate": 1.7169859092344448e-05,
+      "loss": 0.1976,
       "step": 679
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.788586317661404e-06,
-      "loss": 0.1222,
+      "epoch": 0.54,
+      "learning_rate": 1.7160936431500785e-05,
+      "loss": 0.1418,
       "step": 680
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.763718505678282e-06,
-      "loss": 0.162,
+      "epoch": 0.54,
+      "learning_rate": 1.7152002053755604e-05,
+      "loss": 0.1382,
       "step": 681
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.738852155611531e-06,
-      "loss": 0.1505,
+      "epoch": 0.54,
+      "learning_rate": 1.714305597372755e-05,
+      "loss": 0.1464,
       "step": 682
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.71398742131376e-06,
-      "loss": 0.1562,
+      "epoch": 0.54,
+      "learning_rate": 1.713409820605443e-05,
+      "loss": 0.1809,
       "step": 683
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.689124456627577e-06,
-      "loss": 0.1824,
+      "epoch": 0.54,
+      "learning_rate": 1.7125128765393157e-05,
+      "loss": 0.1667,
       "step": 684
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.664263415384644e-06,
-      "loss": 0.1526,
+      "epoch": 0.54,
+      "learning_rate": 1.7116147666419755e-05,
+      "loss": 0.1664,
       "step": 685
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.639404451404729e-06,
-      "loss": 0.1288,
+      "epoch": 0.54,
+      "learning_rate": 1.7107154923829317e-05,
+      "loss": 0.1418,
       "step": 686
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.614547718494736e-06,
-      "loss": 0.1289,
+      "epoch": 0.54,
+      "learning_rate": 1.7098150552335997e-05,
+      "loss": 0.1545,
       "step": 687
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.589693370447769e-06,
-      "loss": 0.149,
+      "epoch": 0.54,
+      "learning_rate": 1.708913456667297e-05,
+      "loss": 0.1762,
       "step": 688
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.564841561042187e-06,
-      "loss": 0.1466,
+      "epoch": 0.54,
+      "learning_rate": 1.7080106981592407e-05,
+      "loss": 0.1764,
       "step": 689
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.53999244404062e-06,
-      "loss": 0.1608,
+      "epoch": 0.55,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.1655,
       "step": 690
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.515146173189058e-06,
-      "loss": 0.1835,
+      "epoch": 0.55,
+      "learning_rate": 1.7062017072282285e-05,
+      "loss": 0.1686,
       "step": 691
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.490302902215882e-06,
-      "loss": 0.1229,
+      "epoch": 0.55,
+      "learning_rate": 1.705295477765188e-05,
+      "loss": 0.1635,
       "step": 692
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.465462784830895e-06,
-      "loss": 0.1499,
+      "epoch": 0.55,
+      "learning_rate": 1.7043880942802212e-05,
+      "loss": 0.1619,
       "step": 693
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.440625974724408e-06,
-      "loss": 0.1361,
+      "epoch": 0.55,
+      "learning_rate": 1.7034795582580118e-05,
+      "loss": 0.1416,
       "step": 694
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.415792625566253e-06,
-      "loss": 0.1902,
+      "epoch": 0.55,
+      "learning_rate": 1.7025698711851283e-05,
+      "loss": 0.1404,
       "step": 695
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.390962891004858e-06,
-      "loss": 0.1227,
+      "epoch": 0.55,
+      "learning_rate": 1.701659034550023e-05,
+      "loss": 0.1412,
       "step": 696
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 9.366136924666289e-06,
-      "loss": 0.1444,
+      "epoch": 0.55,
+      "learning_rate": 1.70074704984303e-05,
+      "loss": 0.1303,
       "step": 697
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.34131488015329e-06,
-      "loss": 0.1432,
+      "epoch": 0.55,
+      "learning_rate": 1.6998339185563614e-05,
+      "loss": 0.1233,
       "step": 698
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.31649691104434e-06,
-      "loss": 0.1489,
+      "epoch": 0.55,
+      "learning_rate": 1.6989196421841045e-05,
+      "loss": 0.13,
       "step": 699
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.291683170892712e-06,
-      "loss": 0.1693,
+      "epoch": 0.55,
+      "learning_rate": 1.6980042222222216e-05,
+      "loss": 0.197,
       "step": 700
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.266873813225498e-06,
-      "loss": 0.17,
+      "epoch": 0.55,
+      "learning_rate": 1.697087660168545e-05,
+      "loss": 0.145,
       "step": 701
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.242068991542694e-06,
-      "loss": 0.14,
+      "epoch": 0.55,
+      "learning_rate": 1.6961699575227767e-05,
+      "loss": 0.1773,
       "step": 702
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.217268859316219e-06,
-      "loss": 0.1736,
+      "epoch": 0.56,
+      "learning_rate": 1.695251115786484e-05,
+      "loss": 0.1573,
       "step": 703
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.192473569988975e-06,
-      "loss": 0.1615,
+      "epoch": 0.56,
+      "learning_rate": 1.694331136463099e-05,
+      "loss": 0.1636,
       "step": 704
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.167683276973906e-06,
-      "loss": 0.1528,
+      "epoch": 0.56,
+      "learning_rate": 1.6934100210579144e-05,
+      "loss": 0.1592,
       "step": 705
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.142898133653049e-06,
-      "loss": 0.1623,
+      "epoch": 0.56,
+      "learning_rate": 1.6924877710780818e-05,
+      "loss": 0.1599,
       "step": 706
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.11811829337656e-06,
-      "loss": 0.1485,
+      "epoch": 0.56,
+      "learning_rate": 1.69156438803261e-05,
+      "loss": 0.1722,
       "step": 707
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.093343909461807e-06,
-      "loss": 0.1675,
+      "epoch": 0.56,
+      "learning_rate": 1.690639873432361e-05,
+      "loss": 0.1322,
       "step": 708
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.068575135192377e-06,
-      "loss": 0.1639,
+      "epoch": 0.56,
+      "learning_rate": 1.6897142287900477e-05,
+      "loss": 0.1415,
       "step": 709
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.043812123817165e-06,
-      "loss": 0.1341,
+      "epoch": 0.56,
+      "learning_rate": 1.6887874556202342e-05,
+      "loss": 0.1096,
       "step": 710
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 9.019055028549403e-06,
-      "loss": 0.1243,
+      "epoch": 0.56,
+      "learning_rate": 1.687859555439329e-05,
+      "loss": 0.1475,
       "step": 711
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 8.994304002565723e-06,
-      "loss": 0.1823,
+      "epoch": 0.56,
+      "learning_rate": 1.686930529765585e-05,
+      "loss": 0.1526,
       "step": 712
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 8.969559199005196e-06,
-      "loss": 0.1382,
+      "epoch": 0.56,
+      "learning_rate": 1.6860003801190975e-05,
+      "loss": 0.1563,
       "step": 713
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 8.944820770968407e-06,
-      "loss": 0.151,
+      "epoch": 0.56,
+      "learning_rate": 1.6850691080218e-05,
+      "loss": 0.1914,
       "step": 714
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 8.920088871516482e-06,
-      "loss": 0.1578,
+      "epoch": 0.56,
+      "learning_rate": 1.6841367149974638e-05,
+      "loss": 0.155,
       "step": 715
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 8.895363653670162e-06,
-      "loss": 0.1764,
+      "epoch": 0.57,
+      "learning_rate": 1.683203202571692e-05,
+      "loss": 0.1391,
       "step": 716
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 8.870645270408849e-06,
-      "loss": 0.1588,
+      "epoch": 0.57,
+      "learning_rate": 1.6822685722719224e-05,
+      "loss": 0.1796,
       "step": 717
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 8.845933874669645e-06,
-      "loss": 0.1785,
+      "epoch": 0.57,
+      "learning_rate": 1.681332825627419e-05,
+      "loss": 0.1514,
       "step": 718
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 8.821229619346433e-06,
-      "loss": 0.1792,
+      "epoch": 0.57,
+      "learning_rate": 1.680395964169274e-05,
+      "loss": 0.1559,
       "step": 719
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 8.796532657288916e-06,
-      "loss": 0.1313,
+      "epoch": 0.57,
+      "learning_rate": 1.6794579894304043e-05,
+      "loss": 0.1548,
       "step": 720
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 8.771843141301659e-06,
-      "loss": 0.1253,
+      "epoch": 0.57,
+      "learning_rate": 1.6785189029455455e-05,
+      "loss": 0.1413,
       "step": 721
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 8.747161224143175e-06,
-      "loss": 0.1489,
+      "epoch": 0.57,
+      "learning_rate": 1.6775787062512557e-05,
+      "loss": 0.1298,
       "step": 722
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 8.722487058524945e-06,
-      "loss": 0.1428,
+      "epoch": 0.57,
+      "learning_rate": 1.676637400885907e-05,
+      "loss": 0.1871,
       "step": 723
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 8.697820797110499e-06,
-      "loss": 0.1637,
+      "epoch": 0.57,
+      "learning_rate": 1.6756949883896874e-05,
+      "loss": 0.1412,
       "step": 724
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 8.673162592514466e-06,
-      "loss": 0.162,
+      "epoch": 0.57,
+      "learning_rate": 1.6747514703045952e-05,
+      "loss": 0.1448,
       "step": 725
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 8.648512597301613e-06,
-      "loss": 0.1533,
+      "epoch": 0.57,
+      "learning_rate": 1.673806848174438e-05,
+      "loss": 0.1455,
       "step": 726
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 8.62387096398593e-06,
-      "loss": 0.0983,
+      "epoch": 0.57,
+      "learning_rate": 1.67286112354483e-05,
+      "loss": 0.1564,
       "step": 727
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 8.599237845029657e-06,
-      "loss": 0.1464,
+      "epoch": 0.58,
+      "learning_rate": 1.671914297963189e-05,
+      "loss": 0.1688,
       "step": 728
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 8.574613392842361e-06,
-      "loss": 0.1644,
+      "epoch": 0.58,
+      "learning_rate": 1.670966372978735e-05,
+      "loss": 0.1522,
       "step": 729
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 8.549997759779981e-06,
-      "loss": 0.1322,
+      "epoch": 0.58,
+      "learning_rate": 1.670017350142486e-05,
+      "loss": 0.1366,
       "step": 730
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 8.525391098143902e-06,
-      "loss": 0.1486,
+      "epoch": 0.58,
+      "learning_rate": 1.669067231007256e-05,
+      "loss": 0.1731,
       "step": 731
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 8.500793560179984e-06,
-      "loss": 0.1199,
+      "epoch": 0.58,
+      "learning_rate": 1.668116017127655e-05,
+      "loss": 0.1324,
       "step": 732
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 8.47620529807765e-06,
-      "loss": 0.157,
+      "epoch": 0.58,
+      "learning_rate": 1.667163710060082e-05,
+      "loss": 0.1663,
       "step": 733
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 8.451626463968932e-06,
-      "loss": 0.1303,
+      "epoch": 0.58,
+      "learning_rate": 1.6662103113627246e-05,
+      "loss": 0.1512,
       "step": 734
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 8.427057209927519e-06,
-      "loss": 0.1253,
+      "epoch": 0.58,
+      "learning_rate": 1.6652558225955582e-05,
+      "loss": 0.1547,
       "step": 735
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 8.402497687967837e-06,
-      "loss": 0.121,
+      "epoch": 0.58,
+      "learning_rate": 1.6643002453203405e-05,
+      "loss": 0.1785,
       "step": 736
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.377948050044087e-06,
-      "loss": 0.1411,
+      "epoch": 0.58,
+      "learning_rate": 1.6633435811006117e-05,
+      "loss": 0.1878,
       "step": 737
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.353408448049327e-06,
-      "loss": 0.1296,
+      "epoch": 0.58,
+      "learning_rate": 1.662385831501688e-05,
+      "loss": 0.1738,
       "step": 738
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.328879033814516e-06,
-      "loss": 0.1353,
+      "epoch": 0.58,
+      "learning_rate": 1.661426998090664e-05,
+      "loss": 0.1791,
       "step": 739
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.304359959107579e-06,
-      "loss": 0.1236,
+      "epoch": 0.58,
+      "learning_rate": 1.6604670824364067e-05,
+      "loss": 0.1755,
       "step": 740
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.279851375632469e-06,
-      "loss": 0.1898,
+      "epoch": 0.59,
+      "learning_rate": 1.6595060861095534e-05,
+      "loss": 0.1571,
       "step": 741
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.255353435028228e-06,
-      "loss": 0.1373,
+      "epoch": 0.59,
+      "learning_rate": 1.6585440106825107e-05,
+      "loss": 0.1475,
       "step": 742
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.230866288868045e-06,
-      "loss": 0.1348,
+      "epoch": 0.59,
+      "learning_rate": 1.6575808577294492e-05,
+      "loss": 0.1704,
       "step": 743
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.206390088658326e-06,
-      "loss": 0.1464,
+      "epoch": 0.59,
+      "learning_rate": 1.6566166288263046e-05,
+      "loss": 0.1656,
       "step": 744
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.181924985837762e-06,
-      "loss": 0.1536,
+      "epoch": 0.59,
+      "learning_rate": 1.6556513255507714e-05,
+      "loss": 0.172,
       "step": 745
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.157471131776356e-06,
-      "loss": 0.1459,
+      "epoch": 0.59,
+      "learning_rate": 1.6546849494823037e-05,
+      "loss": 0.1381,
       "step": 746
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.13302867777454e-06,
-      "loss": 0.1177,
+      "epoch": 0.59,
+      "learning_rate": 1.653717502202109e-05,
+      "loss": 0.1577,
       "step": 747
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.108597775062198e-06,
-      "loss": 0.1281,
+      "epoch": 0.59,
+      "learning_rate": 1.652748985293149e-05,
+      "loss": 0.1436,
       "step": 748
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 8.084178574797745e-06,
-      "loss": 0.1549,
+      "epoch": 0.59,
+      "learning_rate": 1.6517794003401345e-05,
+      "loss": 0.1476,
       "step": 749
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 8.059771228067197e-06,
-      "loss": 0.1098,
+      "epoch": 0.59,
+      "learning_rate": 1.650808748929525e-05,
+      "loss": 0.1315,
       "step": 750
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 8.035375885883217e-06,
-      "loss": 0.1377,
+      "epoch": 0.59,
+      "learning_rate": 1.6498370326495242e-05,
+      "loss": 0.1479,
       "step": 751
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 8.010992699184208e-06,
-      "loss": 0.136,
+      "epoch": 0.59,
+      "learning_rate": 1.648864253090078e-05,
+      "loss": 0.1329,
       "step": 752
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 7.986621818833364e-06,
-      "loss": 0.1519,
+      "epoch": 0.59,
+      "learning_rate": 1.6478904118428735e-05,
+      "loss": 0.1322,
       "step": 753
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 7.962263395617724e-06,
-      "loss": 0.1469,
+      "epoch": 0.6,
+      "learning_rate": 1.6469155105013324e-05,
+      "loss": 0.1688,
       "step": 754
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 7.937917580247265e-06,
-      "loss": 0.1231,
+      "epoch": 0.6,
+      "learning_rate": 1.6459395506606133e-05,
+      "loss": 0.1494,
       "step": 755
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 7.913584523353958e-06,
-      "loss": 0.1518,
+      "epoch": 0.6,
+      "learning_rate": 1.6449625339176056e-05,
+      "loss": 0.1414,
       "step": 756
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 7.88926437549082e-06,
-      "loss": 0.1262,
+      "epoch": 0.6,
+      "learning_rate": 1.6439844618709285e-05,
+      "loss": 0.1713,
       "step": 757
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 7.864957287131012e-06,
-      "loss": 0.1462,
+      "epoch": 0.6,
+      "learning_rate": 1.6430053361209274e-05,
+      "loss": 0.1348,
       "step": 758
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 7.840663408666893e-06,
-      "loss": 0.1082,
+      "epoch": 0.6,
+      "learning_rate": 1.642025158269672e-05,
+      "loss": 0.1366,
       "step": 759
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 7.81638289040908e-06,
-      "loss": 0.1541,
+      "epoch": 0.6,
+      "learning_rate": 1.641043929920954e-05,
+      "loss": 0.1754,
       "step": 760
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 7.792115882585536e-06,
-      "loss": 0.1821,
+      "epoch": 0.6,
+      "learning_rate": 1.6400616526802835e-05,
+      "loss": 0.1418,
       "step": 761
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 7.767862535340635e-06,
-      "loss": 0.1318,
+      "epoch": 0.6,
+      "learning_rate": 1.6390783281548865e-05,
+      "loss": 0.1283,
       "step": 762
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 7.743622998734217e-06,
-      "loss": 0.1402,
+      "epoch": 0.6,
+      "learning_rate": 1.6380939579537033e-05,
+      "loss": 0.1426,
       "step": 763
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 7.71939742274069e-06,
-      "loss": 0.1638,
+      "epoch": 0.6,
+      "learning_rate": 1.6371085436873847e-05,
+      "loss": 0.1527,
       "step": 764
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 7.695185957248074e-06,
-      "loss": 0.1342,
+      "epoch": 0.6,
+      "learning_rate": 1.6361220869682896e-05,
+      "loss": 0.1596,
       "step": 765
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 7.670988752057088e-06,
-      "loss": 0.131,
+      "epoch": 0.61,
+      "learning_rate": 1.635134589410483e-05,
+      "loss": 0.128,
       "step": 766
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 7.646805956880222e-06,
-      "loss": 0.1409,
+      "epoch": 0.61,
+      "learning_rate": 1.6341460526297335e-05,
+      "loss": 0.1287,
       "step": 767
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 7.622637721340801e-06,
-      "loss": 0.1612,
+      "epoch": 0.61,
+      "learning_rate": 1.6331564782435087e-05,
+      "loss": 0.1433,
       "step": 768
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 7.598484194972076e-06,
-      "loss": 0.1199,
+      "epoch": 0.61,
+      "learning_rate": 1.6321658678709752e-05,
+      "loss": 0.1503,
       "step": 769
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 7.574345527216293e-06,
-      "loss": 0.1552,
+      "epoch": 0.61,
+      "learning_rate": 1.6311742231329936e-05,
+      "loss": 0.1298,
       "step": 770
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 7.550221867423746e-06,
-      "loss": 0.1327,
+      "epoch": 0.61,
+      "learning_rate": 1.6301815456521185e-05,
+      "loss": 0.1495,
       "step": 771
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 7.526113364851891e-06,
-      "loss": 0.1688,
+      "epoch": 0.61,
+      "learning_rate": 1.6291878370525925e-05,
+      "loss": 0.1633,
       "step": 772
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 7.5020201686644e-06,
-      "loss": 0.138,
+      "epoch": 0.61,
+      "learning_rate": 1.6281930989603466e-05,
+      "loss": 0.1372,
       "step": 773
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 7.47794242793023e-06,
-      "loss": 0.1579,
+      "epoch": 0.61,
+      "learning_rate": 1.627197333002996e-05,
+      "loss": 0.1363,
       "step": 774
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 7.453880291622726e-06,
-      "loss": 0.1556,
+      "epoch": 0.61,
+      "learning_rate": 1.6262005408098378e-05,
+      "loss": 0.1655,
       "step": 775
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 7.429833908618682e-06,
-      "loss": 0.1631,
+      "epoch": 0.61,
+      "learning_rate": 1.6252027240118472e-05,
+      "loss": 0.1063,
       "step": 776
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 7.405803427697417e-06,
-      "loss": 0.1072,
+      "epoch": 0.61,
+      "learning_rate": 1.624203884241678e-05,
+      "loss": 0.1298,
       "step": 777
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 7.381788997539869e-06,
-      "loss": 0.1691,
+      "epoch": 0.61,
+      "learning_rate": 1.6232040231336556e-05,
+      "loss": 0.1661,
       "step": 778
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 7.357790766727663e-06,
-      "loss": 0.1448,
+      "epoch": 0.62,
+      "learning_rate": 1.6222031423237776e-05,
+      "loss": 0.1701,
       "step": 779
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 7.333808883742192e-06,
-      "loss": 0.1266,
+      "epoch": 0.62,
+      "learning_rate": 1.6212012434497103e-05,
+      "loss": 0.1606,
       "step": 780
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 7.309843496963715e-06,
-      "loss": 0.1395,
+      "epoch": 0.62,
+      "learning_rate": 1.620198328150785e-05,
+      "loss": 0.1,
       "step": 781
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 7.285894754670413e-06,
-      "loss": 0.1482,
+      "epoch": 0.62,
+      "learning_rate": 1.6191943980679975e-05,
+      "loss": 0.1417,
       "step": 782
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 7.261962805037486e-06,
-      "loss": 0.1088,
+      "epoch": 0.62,
+      "learning_rate": 1.6181894548440022e-05,
+      "loss": 0.1613,
       "step": 783
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 7.238047796136247e-06,
-      "loss": 0.1393,
+      "epoch": 0.62,
+      "learning_rate": 1.617183500123112e-05,
+      "loss": 0.133,
       "step": 784
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 7.214149875933173e-06,
-      "loss": 0.1528,
+      "epoch": 0.62,
+      "learning_rate": 1.6161765355512958e-05,
+      "loss": 0.1328,
       "step": 785
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 7.190269192289028e-06,
-      "loss": 0.1342,
+      "epoch": 0.62,
+      "learning_rate": 1.615168562776173e-05,
+      "loss": 0.1596,
       "step": 786
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 7.166405892957926e-06,
-      "loss": 0.1461,
+      "epoch": 0.62,
+      "learning_rate": 1.6141595834470142e-05,
+      "loss": 0.1782,
       "step": 787
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 7.142560125586413e-06,
-      "loss": 0.1223,
+      "epoch": 0.62,
+      "learning_rate": 1.6131495992147363e-05,
+      "loss": 0.166,
       "step": 788
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 7.11873203771257e-06,
-      "loss": 0.1385,
+      "epoch": 0.62,
+      "learning_rate": 1.6121386117319e-05,
+      "loss": 0.1752,
       "step": 789
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 7.094921776765095e-06,
-      "loss": 0.1687,
+      "epoch": 0.62,
+      "learning_rate": 1.6111266226527086e-05,
+      "loss": 0.1328,
       "step": 790
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 7.071129490062372e-06,
-      "loss": 0.146,
+      "epoch": 0.62,
+      "learning_rate": 1.6101136336330037e-05,
+      "loss": 0.1566,
       "step": 791
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 7.047355324811591e-06,
-      "loss": 0.1597,
+      "epoch": 0.63,
+      "learning_rate": 1.609099646330263e-05,
+      "loss": 0.1208,
       "step": 792
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 7.023599428107815e-06,
-      "loss": 0.1599,
+      "epoch": 0.63,
+      "learning_rate": 1.6080846624035972e-05,
+      "loss": 0.1494,
       "step": 793
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 6.999861946933072e-06,
-      "loss": 0.1318,
+      "epoch": 0.63,
+      "learning_rate": 1.6070686835137484e-05,
+      "loss": 0.1548,
       "step": 794
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 6.97614302815546e-06,
-      "loss": 0.1526,
+      "epoch": 0.63,
+      "learning_rate": 1.6060517113230866e-05,
+      "loss": 0.1539,
       "step": 795
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 6.95244281852822e-06,
-      "loss": 0.1495,
+      "epoch": 0.63,
+      "learning_rate": 1.605033747495607e-05,
+      "loss": 0.1616,
       "step": 796
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 6.928761464688836e-06,
-      "loss": 0.158,
+      "epoch": 0.63,
+      "learning_rate": 1.6040147936969263e-05,
+      "loss": 0.168,
       "step": 797
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 6.905099113158138e-06,
-      "loss": 0.1516,
+      "epoch": 0.63,
+      "learning_rate": 1.602994851594283e-05,
+      "loss": 0.192,
       "step": 798
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 6.881455910339369e-06,
-      "loss": 0.1445,
+      "epoch": 0.63,
+      "learning_rate": 1.6019739228565314e-05,
+      "loss": 0.1293,
       "step": 799
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 6.857832002517307e-06,
-      "loss": 0.1675,
+      "epoch": 0.63,
+      "learning_rate": 1.6009520091541403e-05,
+      "loss": 0.147,
       "step": 800
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 6.834227535857349e-06,
-      "loss": 0.1384,
+      "epoch": 0.63,
+      "learning_rate": 1.5999291121591894e-05,
+      "loss": 0.1683,
       "step": 801
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 6.8106426564045965e-06,
-      "loss": 0.1508,
+      "epoch": 0.63,
+      "learning_rate": 1.5989052335453695e-05,
+      "loss": 0.1624,
       "step": 802
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 6.78707751008297e-06,
-      "loss": 0.1315,
+      "epoch": 0.63,
+      "learning_rate": 1.5978803749879754e-05,
+      "loss": 0.1457,
       "step": 803
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 6.763532242694298e-06,
-      "loss": 0.152,
+      "epoch": 0.64,
+      "learning_rate": 1.596854538163906e-05,
+      "loss": 0.1184,
       "step": 804
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 6.740006999917406e-06,
-      "loss": 0.17,
+      "epoch": 0.64,
+      "learning_rate": 1.595827724751661e-05,
+      "loss": 0.146,
       "step": 805
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 6.716501927307231e-06,
-      "loss": 0.1817,
+      "epoch": 0.64,
+      "learning_rate": 1.5947999364313378e-05,
+      "loss": 0.1694,
       "step": 806
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 6.69301717029391e-06,
-      "loss": 0.1542,
+      "epoch": 0.64,
+      "learning_rate": 1.5937711748846292e-05,
+      "loss": 0.1581,
       "step": 807
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 6.669552874181888e-06,
-      "loss": 0.1599,
+      "epoch": 0.64,
+      "learning_rate": 1.5927414417948205e-05,
+      "loss": 0.1497,
       "step": 808
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 6.646109184149011e-06,
-      "loss": 0.1501,
+      "epoch": 0.64,
+      "learning_rate": 1.5917107388467866e-05,
+      "loss": 0.1393,
       "step": 809
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 6.6226862452456295e-06,
-      "loss": 0.1547,
+      "epoch": 0.64,
+      "learning_rate": 1.5906790677269887e-05,
+      "loss": 0.1601,
       "step": 810
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 6.599284202393709e-06,
-      "loss": 0.1898,
+      "epoch": 0.64,
+      "learning_rate": 1.589646430123473e-05,
+      "loss": 0.1525,
       "step": 811
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 6.575903200385927e-06,
-      "loss": 0.1584,
+      "epoch": 0.64,
+      "learning_rate": 1.5886128277258665e-05,
+      "loss": 0.178,
       "step": 812
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 6.552543383884766e-06,
-      "loss": 0.1512,
+      "epoch": 0.64,
+      "learning_rate": 1.587578262225375e-05,
+      "loss": 0.1186,
       "step": 813
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 6.529204897421644e-06,
-      "loss": 0.153,
+      "epoch": 0.64,
+      "learning_rate": 1.5865427353147805e-05,
+      "loss": 0.1531,
       "step": 814
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 6.505887885396003e-06,
-      "loss": 0.1422,
+      "epoch": 0.64,
+      "learning_rate": 1.5855062486884377e-05,
+      "loss": 0.1469,
       "step": 815
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 6.482592492074411e-06,
-      "loss": 0.1531,
+      "epoch": 0.64,
+      "learning_rate": 1.5844688040422714e-05,
+      "loss": 0.1353,
       "step": 816
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 6.4593188615896855e-06,
-      "loss": 0.1213,
+      "epoch": 0.65,
+      "learning_rate": 1.5834304030737744e-05,
+      "loss": 0.1458,
       "step": 817
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 6.436067137939994e-06,
-      "loss": 0.131,
+      "epoch": 0.65,
+      "learning_rate": 1.582391047482004e-05,
+      "loss": 0.1406,
       "step": 818
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 6.412837464987951e-06,
-      "loss": 0.1306,
+      "epoch": 0.65,
+      "learning_rate": 1.5813507389675796e-05,
+      "loss": 0.133,
       "step": 819
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 6.389629986459756e-06,
-      "loss": 0.1619,
+      "epoch": 0.65,
+      "learning_rate": 1.58030947923268e-05,
+      "loss": 0.1237,
       "step": 820
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 6.36644484594427e-06,
-      "loss": 0.1307,
+      "epoch": 0.65,
+      "learning_rate": 1.57926726998104e-05,
+      "loss": 0.1232,
       "step": 821
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 6.343282186892163e-06,
-      "loss": 0.1339,
+      "epoch": 0.65,
+      "learning_rate": 1.5782241129179482e-05,
+      "loss": 0.1457,
       "step": 822
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 6.3201421526149945e-06,
-      "loss": 0.1501,
+      "epoch": 0.65,
+      "learning_rate": 1.5771800097502437e-05,
+      "loss": 0.1308,
       "step": 823
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 6.297024886284342e-06,
-      "loss": 0.1694,
+      "epoch": 0.65,
+      "learning_rate": 1.5761349621863145e-05,
+      "loss": 0.1497,
       "step": 824
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 6.273930530930919e-06,
-      "loss": 0.1494,
+      "epoch": 0.65,
+      "learning_rate": 1.5750889719360927e-05,
+      "loss": 0.136,
       "step": 825
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 6.250859229443684e-06,
-      "loss": 0.1433,
+      "epoch": 0.65,
+      "learning_rate": 1.574042040711054e-05,
+      "loss": 0.1051,
       "step": 826
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 6.2278111245689495e-06,
-      "loss": 0.1362,
+      "epoch": 0.65,
+      "learning_rate": 1.572994170224213e-05,
+      "loss": 0.1459,
       "step": 827
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 6.204786358909516e-06,
-      "loss": 0.136,
+      "epoch": 0.65,
+      "learning_rate": 1.571945362190121e-05,
+      "loss": 0.1534,
       "step": 828
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 6.181785074923778e-06,
-      "loss": 0.1478,
+      "epoch": 0.65,
+      "learning_rate": 1.5708956183248644e-05,
+      "loss": 0.1234,
       "step": 829
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 6.1588074149248365e-06,
-      "loss": 0.1317,
+      "epoch": 0.66,
+      "learning_rate": 1.5698449403460593e-05,
+      "loss": 0.1586,
       "step": 830
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 6.135853521079636e-06,
-      "loss": 0.1588,
+      "epoch": 0.66,
+      "learning_rate": 1.5687933299728517e-05,
+      "loss": 0.1138,
       "step": 831
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 6.112923535408074e-06,
-      "loss": 0.1459,
+      "epoch": 0.66,
+      "learning_rate": 1.567740788925912e-05,
+      "loss": 0.143,
       "step": 832
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 6.0900175997821165e-06,
-      "loss": 0.1692,
+      "epoch": 0.66,
+      "learning_rate": 1.5666873189274344e-05,
+      "loss": 0.1641,
       "step": 833
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 6.067135855924937e-06,
-      "loss": 0.1084,
+      "epoch": 0.66,
+      "learning_rate": 1.5656329217011322e-05,
+      "loss": 0.1659,
       "step": 834
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 6.044278445410025e-06,
-      "loss": 0.1339,
+      "epoch": 0.66,
+      "learning_rate": 1.5645775989722366e-05,
+      "loss": 0.1471,
       "step": 835
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 6.021445509660315e-06,
-      "loss": 0.1735,
+      "epoch": 0.66,
+      "learning_rate": 1.563521352467493e-05,
+      "loss": 0.1544,
       "step": 836
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 5.9986371899473205e-06,
-      "loss": 0.1054,
+      "epoch": 0.66,
+      "learning_rate": 1.562464183915157e-05,
+      "loss": 0.1169,
       "step": 837
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 5.975853627390233e-06,
-      "loss": 0.1334,
+      "epoch": 0.66,
+      "learning_rate": 1.5614060950449948e-05,
+      "loss": 0.1423,
       "step": 838
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 5.953094962955081e-06,
-      "loss": 0.1395,
+      "epoch": 0.66,
+      "learning_rate": 1.560347087588278e-05,
+      "loss": 0.1516,
       "step": 839
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 5.930361337453847e-06,
-      "loss": 0.1121,
+      "epoch": 0.66,
+      "learning_rate": 1.5592871632777798e-05,
+      "loss": 0.1502,
       "step": 840
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.907652891543576e-06,
-      "loss": 0.153,
+      "epoch": 0.66,
+      "learning_rate": 1.5582263238477753e-05,
+      "loss": 0.1743,
       "step": 841
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.88496976572554e-06,
-      "loss": 0.1368,
+      "epoch": 0.67,
+      "learning_rate": 1.557164571034036e-05,
+      "loss": 0.1688,
       "step": 842
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.862312100344345e-06,
-      "loss": 0.1368,
+      "epoch": 0.67,
+      "learning_rate": 1.5561019065738282e-05,
+      "loss": 0.1461,
       "step": 843
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.839680035587061e-06,
-      "loss": 0.1158,
+      "epoch": 0.67,
+      "learning_rate": 1.55503833220591e-05,
+      "loss": 0.159,
       "step": 844
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.8170737114823715e-06,
-      "loss": 0.1136,
+      "epoch": 0.67,
+      "learning_rate": 1.5539738496705277e-05,
+      "loss": 0.1495,
       "step": 845
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.794493267899699e-06,
-      "loss": 0.1198,
+      "epoch": 0.67,
+      "learning_rate": 1.5529084607094144e-05,
+      "loss": 0.1511,
       "step": 846
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.771938844548331e-06,
-      "loss": 0.1435,
+      "epoch": 0.67,
+      "learning_rate": 1.5518421670657856e-05,
+      "loss": 0.152,
       "step": 847
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.749410580976569e-06,
-      "loss": 0.1306,
+      "epoch": 0.67,
+      "learning_rate": 1.550774970484337e-05,
+      "loss": 0.1362,
       "step": 848
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.7269086165708546e-06,
-      "loss": 0.1386,
+      "epoch": 0.67,
+      "learning_rate": 1.5497068727112435e-05,
+      "loss": 0.1707,
       "step": 849
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.704433090554912e-06,
-      "loss": 0.1523,
+      "epoch": 0.67,
+      "learning_rate": 1.5486378754941514e-05,
+      "loss": 0.1351,
       "step": 850
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.6819841419889e-06,
-      "loss": 0.1416,
+      "epoch": 0.67,
+      "learning_rate": 1.5475679805821814e-05,
+      "loss": 0.1357,
       "step": 851
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.659561909768509e-06,
-      "loss": 0.1427,
+      "epoch": 0.67,
+      "learning_rate": 1.546497189725922e-05,
+      "loss": 0.1607,
       "step": 852
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 5.6371665326241635e-06,
-      "loss": 0.1183,
+      "epoch": 0.67,
+      "learning_rate": 1.5454255046774273e-05,
+      "loss": 0.1497,
       "step": 853
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.614798149120108e-06,
-      "loss": 0.128,
+      "epoch": 0.67,
+      "learning_rate": 1.5443529271902155e-05,
+      "loss": 0.1337,
       "step": 854
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.5924568976535775e-06,
-      "loss": 0.1332,
+      "epoch": 0.68,
+      "learning_rate": 1.543279459019264e-05,
+      "loss": 0.1545,
       "step": 855
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.570142916453944e-06,
-      "loss": 0.1309,
+      "epoch": 0.68,
+      "learning_rate": 1.5422051019210082e-05,
+      "loss": 0.1332,
       "step": 856
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.547856343581848e-06,
-      "loss": 0.1473,
+      "epoch": 0.68,
+      "learning_rate": 1.5411298576533376e-05,
+      "loss": 0.155,
       "step": 857
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.525597316928344e-06,
-      "loss": 0.1506,
+      "epoch": 0.68,
+      "learning_rate": 1.5400537279755935e-05,
+      "loss": 0.1569,
       "step": 858
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.503365974214059e-06,
-      "loss": 0.1292,
+      "epoch": 0.68,
+      "learning_rate": 1.538976714648566e-05,
+      "loss": 0.1334,
       "step": 859
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.481162452988344e-06,
-      "loss": 0.1241,
+      "epoch": 0.68,
+      "learning_rate": 1.5378988194344913e-05,
+      "loss": 0.1639,
       "step": 860
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.458986890628389e-06,
-      "loss": 0.1215,
+      "epoch": 0.68,
+      "learning_rate": 1.5368200440970478e-05,
+      "loss": 0.1458,
       "step": 861
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.436839424338426e-06,
-      "loss": 0.1758,
+      "epoch": 0.68,
+      "learning_rate": 1.5357403904013546e-05,
+      "loss": 0.1395,
       "step": 862
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.414720191148822e-06,
-      "loss": 0.1288,
+      "epoch": 0.68,
+      "learning_rate": 1.5346598601139677e-05,
+      "loss": 0.1412,
       "step": 863
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.392629327915289e-06,
-      "loss": 0.107,
+      "epoch": 0.68,
+      "learning_rate": 1.533578455002878e-05,
+      "loss": 0.1465,
       "step": 864
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.37056697131799e-06,
-      "loss": 0.1331,
+      "epoch": 0.68,
+      "learning_rate": 1.5324961768375065e-05,
+      "loss": 0.1613,
       "step": 865
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 5.3485332578607205e-06,
-      "loss": 0.1093,
+      "epoch": 0.68,
+      "learning_rate": 1.531413027388704e-05,
+      "loss": 0.1522,
       "step": 866
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.326528323870049e-06,
-      "loss": 0.1636,
+      "epoch": 0.68,
+      "learning_rate": 1.5303290084287465e-05,
+      "loss": 0.1188,
       "step": 867
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.304552305494493e-06,
-      "loss": 0.1128,
+      "epoch": 0.69,
+      "learning_rate": 1.5292441217313324e-05,
+      "loss": 0.123,
       "step": 868
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.2826053387036495e-06,
-      "loss": 0.1729,
+      "epoch": 0.69,
+      "learning_rate": 1.5281583690715805e-05,
+      "loss": 0.1514,
       "step": 869
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.260687559287376e-06,
-      "loss": 0.1187,
+      "epoch": 0.69,
+      "learning_rate": 1.5270717522260264e-05,
+      "loss": 0.1119,
       "step": 870
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.238799102854941e-06,
-      "loss": 0.158,
+      "epoch": 0.69,
+      "learning_rate": 1.5259842729726186e-05,
+      "loss": 0.1356,
       "step": 871
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.2169401048341815e-06,
-      "loss": 0.1252,
+      "epoch": 0.69,
+      "learning_rate": 1.5248959330907186e-05,
+      "loss": 0.1479,
       "step": 872
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.1951107004706825e-06,
-      "loss": 0.1401,
+      "epoch": 0.69,
+      "learning_rate": 1.5238067343610943e-05,
+      "loss": 0.1341,
       "step": 873
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.173311024826916e-06,
-      "loss": 0.1047,
+      "epoch": 0.69,
+      "learning_rate": 1.52271667856592e-05,
+      "loss": 0.1339,
       "step": 874
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.15154121278142e-06,
-      "loss": 0.1689,
+      "epoch": 0.69,
+      "learning_rate": 1.5216257674887718e-05,
+      "loss": 0.1048,
       "step": 875
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.129801399027962e-06,
-      "loss": 0.151,
+      "epoch": 0.69,
+      "learning_rate": 1.5205340029146256e-05,
+      "loss": 0.1429,
       "step": 876
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.108091718074706e-06,
-      "loss": 0.1289,
+      "epoch": 0.69,
+      "learning_rate": 1.5194413866298536e-05,
+      "loss": 0.1288,
       "step": 877
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.08641230424337e-06,
-      "loss": 0.117,
+      "epoch": 0.69,
+      "learning_rate": 1.5183479204222216e-05,
+      "loss": 0.1715,
       "step": 878
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 5.064763291668425e-06,
-      "loss": 0.1656,
+      "epoch": 0.69,
+      "learning_rate": 1.5172536060808857e-05,
+      "loss": 0.153,
       "step": 879
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 5.043144814296214e-06,
-      "loss": 0.1221,
+      "epoch": 0.7,
+      "learning_rate": 1.5161584453963908e-05,
+      "loss": 0.1859,
       "step": 880
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 5.021557005884183e-06,
-      "loss": 0.1306,
+      "epoch": 0.7,
+      "learning_rate": 1.5150624401606658e-05,
+      "loss": 0.1927,
       "step": 881
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 5.000000000000003e-06,
-      "loss": 0.1522,
+      "epoch": 0.7,
+      "learning_rate": 1.5139655921670213e-05,
+      "loss": 0.1479,
       "step": 882
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.9784739300207675e-06,
-      "loss": 0.0971,
+      "epoch": 0.7,
+      "learning_rate": 1.5128679032101472e-05,
+      "loss": 0.139,
       "step": 883
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.956978929132177e-06,
-      "loss": 0.1313,
+      "epoch": 0.7,
+      "learning_rate": 1.5117693750861096e-05,
+      "loss": 0.1307,
       "step": 884
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.935515130327686e-06,
-      "loss": 0.1356,
+      "epoch": 0.7,
+      "learning_rate": 1.5106700095923471e-05,
+      "loss": 0.1604,
       "step": 885
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.914082666407705e-06,
-      "loss": 0.1385,
+      "epoch": 0.7,
+      "learning_rate": 1.5095698085276692e-05,
+      "loss": 0.1302,
       "step": 886
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.892681669978762e-06,
-      "loss": 0.1486,
+      "epoch": 0.7,
+      "learning_rate": 1.5084687736922514e-05,
+      "loss": 0.1457,
       "step": 887
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.871312273452706e-06,
-      "loss": 0.1226,
+      "epoch": 0.7,
+      "learning_rate": 1.5073669068876348e-05,
+      "loss": 0.1309,
       "step": 888
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.849974609045849e-06,
-      "loss": 0.1167,
+      "epoch": 0.7,
+      "learning_rate": 1.5062642099167208e-05,
+      "loss": 0.1853,
       "step": 889
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.828668808778194e-06,
-      "loss": 0.1118,
+      "epoch": 0.7,
+      "learning_rate": 1.5051606845837699e-05,
+      "loss": 0.1978,
       "step": 890
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.807395004472569e-06,
-      "loss": 0.1064,
+      "epoch": 0.7,
+      "learning_rate": 1.5040563326943974e-05,
+      "loss": 0.1522,
       "step": 891
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.786153327753865e-06,
-      "loss": 0.1519,
+      "epoch": 0.7,
+      "learning_rate": 1.5029511560555707e-05,
+      "loss": 0.1257,
       "step": 892
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.764943910048175e-06,
-      "loss": 0.1515,
+      "epoch": 0.71,
+      "learning_rate": 1.5018451564756078e-05,
+      "loss": 0.1222,
       "step": 893
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.743766882582006e-06,
-      "loss": 0.1672,
+      "epoch": 0.71,
+      "learning_rate": 1.5007383357641723e-05,
+      "loss": 0.1204,
       "step": 894
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.722622376381455e-06,
-      "loss": 0.1354,
+      "epoch": 0.71,
+      "learning_rate": 1.499630695732272e-05,
+      "loss": 0.1251,
       "step": 895
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.701510522271416e-06,
-      "loss": 0.1295,
+      "epoch": 0.71,
+      "learning_rate": 1.4985222381922543e-05,
+      "loss": 0.149,
       "step": 896
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.680431450874748e-06,
-      "loss": 0.1541,
+      "epoch": 0.71,
+      "learning_rate": 1.4974129649578058e-05,
+      "loss": 0.1439,
       "step": 897
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.659385292611479e-06,
-      "loss": 0.1615,
+      "epoch": 0.71,
+      "learning_rate": 1.496302877843946e-05,
+      "loss": 0.1432,
       "step": 898
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.638372177698007e-06,
-      "loss": 0.1323,
+      "epoch": 0.71,
+      "learning_rate": 1.4951919786670274e-05,
+      "loss": 0.1583,
       "step": 899
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.617392236146267e-06,
-      "loss": 0.1592,
+      "epoch": 0.71,
+      "learning_rate": 1.4940802692447306e-05,
+      "loss": 0.138,
       "step": 900
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.59644559776296e-06,
-      "loss": 0.1747,
+      "epoch": 0.71,
+      "learning_rate": 1.4929677513960621e-05,
+      "loss": 0.1772,
       "step": 901
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.575532392148729e-06,
-      "loss": 0.1382,
+      "epoch": 0.71,
+      "learning_rate": 1.4918544269413511e-05,
+      "loss": 0.1378,
       "step": 902
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.554652748697361e-06,
-      "loss": 0.1252,
+      "epoch": 0.71,
+      "learning_rate": 1.4907402977022465e-05,
+      "loss": 0.179,
       "step": 903
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.533806796594989e-06,
-      "loss": 0.1432,
+      "epoch": 0.71,
+      "learning_rate": 1.4896253655017146e-05,
+      "loss": 0.1574,
       "step": 904
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.5129946648192926e-06,
-      "loss": 0.1538,
+      "epoch": 0.71,
+      "learning_rate": 1.4885096321640346e-05,
+      "loss": 0.1359,
       "step": 905
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.4922164821386926e-06,
-      "loss": 0.1437,
+      "epoch": 0.72,
+      "learning_rate": 1.4873930995147971e-05,
+      "loss": 0.155,
       "step": 906
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.471472377111574e-06,
-      "loss": 0.1414,
+      "epoch": 0.72,
+      "learning_rate": 1.4862757693809009e-05,
+      "loss": 0.1668,
       "step": 907
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.450762478085465e-06,
-      "loss": 0.1168,
+      "epoch": 0.72,
+      "learning_rate": 1.4851576435905489e-05,
+      "loss": 0.158,
       "step": 908
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.43008691319626e-06,
-      "loss": 0.1405,
+      "epoch": 0.72,
+      "learning_rate": 1.484038723973246e-05,
+      "loss": 0.1247,
       "step": 909
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.409445810367421e-06,
-      "loss": 0.1372,
+      "epoch": 0.72,
+      "learning_rate": 1.4829190123597965e-05,
+      "loss": 0.1631,
       "step": 910
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.388839297309182e-06,
-      "loss": 0.134,
+      "epoch": 0.72,
+      "learning_rate": 1.4817985105823003e-05,
+      "loss": 0.1686,
       "step": 911
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.368267501517778e-06,
-      "loss": 0.1462,
+      "epoch": 0.72,
+      "learning_rate": 1.4806772204741503e-05,
+      "loss": 0.1631,
       "step": 912
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.347730550274628e-06,
-      "loss": 0.1419,
+      "epoch": 0.72,
+      "learning_rate": 1.4795551438700283e-05,
+      "loss": 0.1707,
       "step": 913
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.327228570645566e-06,
-      "loss": 0.124,
+      "epoch": 0.72,
+      "learning_rate": 1.4784322826059048e-05,
+      "loss": 0.1609,
       "step": 914
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.306761689480049e-06,
-      "loss": 0.1397,
+      "epoch": 0.72,
+      "learning_rate": 1.4773086385190328e-05,
+      "loss": 0.1569,
       "step": 915
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.286330033410384e-06,
-      "loss": 0.1179,
+      "epoch": 0.72,
+      "learning_rate": 1.4761842134479463e-05,
+      "loss": 0.1377,
       "step": 916
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.26593372885091e-06,
-      "loss": 0.1735,
+      "epoch": 0.72,
+      "learning_rate": 1.4750590092324579e-05,
+      "loss": 0.1211,
       "step": 917
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.2455729019972715e-06,
-      "loss": 0.1429,
+      "epoch": 0.73,
+      "learning_rate": 1.4739330277136546e-05,
+      "loss": 0.168,
       "step": 918
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.2252476788255735e-06,
-      "loss": 0.1198,
+      "epoch": 0.73,
+      "learning_rate": 1.4728062707338949e-05,
+      "loss": 0.177,
       "step": 919
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.204958185091662e-06,
-      "loss": 0.133,
+      "epoch": 0.73,
+      "learning_rate": 1.4716787401368067e-05,
+      "loss": 0.1469,
       "step": 920
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.184704546330302e-06,
-      "loss": 0.1344,
+      "epoch": 0.73,
+      "learning_rate": 1.4705504377672834e-05,
+      "loss": 0.1466,
       "step": 921
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.164486887854424e-06,
-      "loss": 0.1655,
+      "epoch": 0.73,
+      "learning_rate": 1.4694213654714816e-05,
+      "loss": 0.148,
       "step": 922
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.144305334754335e-06,
-      "loss": 0.1162,
+      "epoch": 0.73,
+      "learning_rate": 1.4682915250968169e-05,
+      "loss": 0.1362,
       "step": 923
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.124160011896965e-06,
-      "loss": 0.1405,
+      "epoch": 0.73,
+      "learning_rate": 1.4671609184919622e-05,
+      "loss": 0.1344,
       "step": 924
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.104051043925068e-06,
-      "loss": 0.1249,
+      "epoch": 0.73,
+      "learning_rate": 1.4660295475068443e-05,
+      "loss": 0.1714,
       "step": 925
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.083978555256466e-06,
-      "loss": 0.1201,
+      "epoch": 0.73,
+      "learning_rate": 1.4648974139926403e-05,
+      "loss": 0.1342,
       "step": 926
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.063942670083288e-06,
-      "loss": 0.1511,
+      "epoch": 0.73,
+      "learning_rate": 1.4637645198017745e-05,
+      "loss": 0.1521,
       "step": 927
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.043943512371171e-06,
-      "loss": 0.0911,
+      "epoch": 0.73,
+      "learning_rate": 1.4626308667879175e-05,
+      "loss": 0.1257,
       "step": 928
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.02398120585853e-06,
-      "loss": 0.1402,
+      "epoch": 0.73,
+      "learning_rate": 1.4614964568059795e-05,
+      "loss": 0.1433,
       "step": 929
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.004055874055769e-06,
-      "loss": 0.1402,
+      "epoch": 0.73,
+      "learning_rate": 1.4603612917121107e-05,
+      "loss": 0.1392,
       "step": 930
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 3.984167640244518e-06,
-      "loss": 0.1791,
+      "epoch": 0.74,
+      "learning_rate": 1.4592253733636961e-05,
+      "loss": 0.1647,
       "step": 931
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 3.964316627476883e-06,
-      "loss": 0.1511,
+      "epoch": 0.74,
+      "learning_rate": 1.4580887036193539e-05,
+      "loss": 0.157,
       "step": 932
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 3.9445029585746695e-06,
-      "loss": 0.1492,
+      "epoch": 0.74,
+      "learning_rate": 1.4569512843389306e-05,
+      "loss": 0.1591,
       "step": 933
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 3.924726756128632e-06,
-      "loss": 0.1592,
+      "epoch": 0.74,
+      "learning_rate": 1.4558131173835002e-05,
+      "loss": 0.1552,
       "step": 934
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 3.90498814249772e-06,
-      "loss": 0.1299,
+      "epoch": 0.74,
+      "learning_rate": 1.4546742046153596e-05,
+      "loss": 0.145,
       "step": 935
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 3.885287239808306e-06,
-      "loss": 0.1415,
+      "epoch": 0.74,
+      "learning_rate": 1.453534547898026e-05,
+      "loss": 0.137,
       "step": 936
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 3.8656241699534396e-06,
-      "loss": 0.1324,
+      "epoch": 0.74,
+      "learning_rate": 1.4523941490962342e-05,
+      "loss": 0.1602,
       "step": 937
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 3.845999054592092e-06,
-      "loss": 0.1171,
+      "epoch": 0.74,
+      "learning_rate": 1.451253010075933e-05,
+      "loss": 0.1409,
       "step": 938
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 3.826412015148403e-06,
-      "loss": 0.1458,
+      "epoch": 0.74,
+      "learning_rate": 1.4501111327042817e-05,
+      "loss": 0.1443,
       "step": 939
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 3.8068631728109364e-06,
-      "loss": 0.1119,
+      "epoch": 0.74,
+      "learning_rate": 1.4489685188496488e-05,
+      "loss": 0.1394,
       "step": 940
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 3.7873526485319135e-06,
-      "loss": 0.1619,
+      "epoch": 0.74,
+      "learning_rate": 1.447825170381607e-05,
+      "loss": 0.149,
       "step": 941
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 3.7678805630264802e-06,
-      "loss": 0.1647,
+      "epoch": 0.74,
+      "learning_rate": 1.446681089170932e-05,
+      "loss": 0.1287,
       "step": 942
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 3.7484470367719493e-06,
-      "loss": 0.1176,
+      "epoch": 0.74,
+      "learning_rate": 1.4455362770895976e-05,
+      "loss": 0.1429,
       "step": 943
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 3.7290521900070774e-06,
-      "loss": 0.1519,
+      "epoch": 0.75,
+      "learning_rate": 1.4443907360107734e-05,
+      "loss": 0.1815,
       "step": 944
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.709696142731277e-06,
-      "loss": 0.1189,
+      "epoch": 0.75,
+      "learning_rate": 1.4432444678088222e-05,
+      "loss": 0.1435,
       "step": 945
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.6903790147039286e-06,
-      "loss": 0.1123,
+      "epoch": 0.75,
+      "learning_rate": 1.4420974743592964e-05,
+      "loss": 0.1578,
       "step": 946
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.671100925443586e-06,
-      "loss": 0.1374,
+      "epoch": 0.75,
+      "learning_rate": 1.4409497575389352e-05,
+      "loss": 0.1548,
       "step": 947
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.6518619942272883e-06,
-      "loss": 0.1244,
+      "epoch": 0.75,
+      "learning_rate": 1.4398013192256615e-05,
+      "loss": 0.1373,
       "step": 948
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.6326623400897797e-06,
-      "loss": 0.1542,
+      "epoch": 0.75,
+      "learning_rate": 1.438652161298578e-05,
+      "loss": 0.1426,
       "step": 949
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.613502081822794e-06,
-      "loss": 0.1219,
+      "epoch": 0.75,
+      "learning_rate": 1.4375022856379657e-05,
+      "loss": 0.1429,
       "step": 950
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.594381337974322e-06,
-      "loss": 0.146,
+      "epoch": 0.75,
+      "learning_rate": 1.4363516941252795e-05,
+      "loss": 0.1265,
       "step": 951
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.575300226847863e-06,
-      "loss": 0.1529,
+      "epoch": 0.75,
+      "learning_rate": 1.4352003886431459e-05,
+      "loss": 0.1233,
       "step": 952
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.5562588665017063e-06,
-      "loss": 0.1328,
+      "epoch": 0.75,
+      "learning_rate": 1.434048371075359e-05,
+      "loss": 0.1118,
       "step": 953
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.5372573747481907e-06,
-      "loss": 0.1257,
+      "epoch": 0.75,
+      "learning_rate": 1.4328956433068789e-05,
+      "loss": 0.1289,
       "step": 954
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.5182958691529945e-06,
-      "loss": 0.1434,
-      "step": 955
+      "epoch": 0.75,
+      "learning_rate": 1.4317422072238271e-05,
+      "loss": 0.1605,
+      "step": 955
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.4993744670343723e-06,
-      "loss": 0.1478,
+      "epoch": 0.76,
+      "learning_rate": 1.4305880647134847e-05,
+      "loss": 0.1603,
       "step": 956
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.4804932854624716e-06,
-      "loss": 0.1598,
+      "epoch": 0.76,
+      "learning_rate": 1.4294332176642875e-05,
+      "loss": 0.0912,
       "step": 957
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.4616524412585797e-06,
-      "loss": 0.1108,
+      "epoch": 0.76,
+      "learning_rate": 1.4282776679658255e-05,
+      "loss": 0.1398,
       "step": 958
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.4428520509944065e-06,
-      "loss": 0.1122,
+      "epoch": 0.76,
+      "learning_rate": 1.4271214175088374e-05,
+      "loss": 0.1755,
       "step": 959
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.4240922309913715e-06,
-      "loss": 0.1046,
+      "epoch": 0.76,
+      "learning_rate": 1.425964468185209e-05,
+      "loss": 0.1638,
       "step": 960
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.405373097319875e-06,
-      "loss": 0.1217,
+      "epoch": 0.76,
+      "learning_rate": 1.4248068218879691e-05,
+      "loss": 0.1498,
       "step": 961
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.3866947657985827e-06,
-      "loss": 0.1311,
+      "epoch": 0.76,
+      "learning_rate": 1.4236484805112878e-05,
+      "loss": 0.1353,
       "step": 962
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.3680573519937187e-06,
-      "loss": 0.151,
+      "epoch": 0.76,
+      "learning_rate": 1.4224894459504717e-05,
+      "loss": 0.1404,
       "step": 963
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.3494609712183323e-06,
-      "loss": 0.1498,
+      "epoch": 0.76,
+      "learning_rate": 1.4213297201019618e-05,
+      "loss": 0.1255,
       "step": 964
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.3309057385315966e-06,
-      "loss": 0.1071,
+      "epoch": 0.76,
+      "learning_rate": 1.4201693048633302e-05,
+      "loss": 0.1509,
       "step": 965
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.312391768738095e-06,
-      "loss": 0.1489,
+      "epoch": 0.76,
+      "learning_rate": 1.419008202133277e-05,
+      "loss": 0.1329,
       "step": 966
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.2939191763871047e-06,
-      "loss": 0.1398,
+      "epoch": 0.76,
+      "learning_rate": 1.4178464138116272e-05,
+      "loss": 0.1636,
       "step": 967
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.2754880757719065e-06,
-      "loss": 0.148,
+      "epoch": 0.76,
+      "learning_rate": 1.4166839417993281e-05,
+      "loss": 0.1247,
       "step": 968
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.2570985809290525e-06,
-      "loss": 0.1647,
+      "epoch": 0.77,
+      "learning_rate": 1.4155207879984447e-05,
+      "loss": 0.1434,
       "step": 969
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.2387508056376726e-06,
-      "loss": 0.1411,
+      "epoch": 0.77,
+      "learning_rate": 1.414356954312158e-05,
+      "loss": 0.1357,
       "step": 970
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.2204448634187734e-06,
-      "loss": 0.1137,
+      "epoch": 0.77,
+      "learning_rate": 1.4131924426447621e-05,
+      "loss": 0.1359,
       "step": 971
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.202180867534541e-06,
-      "loss": 0.1285,
+      "epoch": 0.77,
+      "learning_rate": 1.4120272549016591e-05,
+      "loss": 0.1321,
       "step": 972
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.183958930987612e-06,
-      "loss": 0.122,
+      "epoch": 0.77,
+      "learning_rate": 1.4108613929893586e-05,
+      "loss": 0.1389,
       "step": 973
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.165779166520415e-06,
-      "loss": 0.113,
+      "epoch": 0.77,
+      "learning_rate": 1.4096948588154723e-05,
+      "loss": 0.1396,
       "step": 974
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.1476416866144343e-06,
-      "loss": 0.1422,
+      "epoch": 0.77,
+      "learning_rate": 1.4085276542887128e-05,
+      "loss": 0.119,
       "step": 975
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.1295466034895482e-06,
-      "loss": 0.1307,
+      "epoch": 0.77,
+      "learning_rate": 1.4073597813188884e-05,
+      "loss": 0.1449,
       "step": 976
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.1114940291033092e-06,
-      "loss": 0.1409,
+      "epoch": 0.77,
+      "learning_rate": 1.4061912418169024e-05,
+      "loss": 0.172,
       "step": 977
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.093484075150257e-06,
-      "loss": 0.1454,
+      "epoch": 0.77,
+      "learning_rate": 1.405022037694748e-05,
+      "loss": 0.1558,
       "step": 978
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.0755168530612444e-06,
-      "loss": 0.1587,
+      "epoch": 0.77,
+      "learning_rate": 1.4038521708655054e-05,
+      "loss": 0.155,
       "step": 979
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.057592474002723e-06,
-      "loss": 0.1316,
+      "epoch": 0.77,
+      "learning_rate": 1.40268164324334e-05,
+      "loss": 0.1298,
       "step": 980
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.0397110488760707e-06,
-      "loss": 0.1523,
+      "epoch": 0.77,
+      "learning_rate": 1.4015104567434981e-05,
+      "loss": 0.1406,
       "step": 981
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.021872688316896e-06,
-      "loss": 0.1177,
+      "epoch": 0.78,
+      "learning_rate": 1.400338613282304e-05,
+      "loss": 0.1364,
       "step": 982
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.0040775026943757e-06,
-      "loss": 0.1644,
+      "epoch": 0.78,
+      "learning_rate": 1.3991661147771574e-05,
+      "loss": 0.1317,
       "step": 983
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.98632560211053e-06,
-      "loss": 0.1648,
+      "epoch": 0.78,
+      "learning_rate": 1.3979929631465286e-05,
+      "loss": 0.1633,
       "step": 984
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.968617096399592e-06,
-      "loss": 0.1328,
+      "epoch": 0.78,
+      "learning_rate": 1.396819160309958e-05,
+      "loss": 0.1467,
       "step": 985
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.9509520951272864e-06,
-      "loss": 0.1494,
+      "epoch": 0.78,
+      "learning_rate": 1.3956447081880506e-05,
+      "loss": 0.1879,
       "step": 986
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.933330707590174e-06,
-      "loss": 0.1299,
+      "epoch": 0.78,
+      "learning_rate": 1.394469608702474e-05,
+      "loss": 0.1614,
       "step": 987
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.9157530428149683e-06,
-      "loss": 0.1473,
+      "epoch": 0.78,
+      "learning_rate": 1.3932938637759555e-05,
+      "loss": 0.1698,
       "step": 988
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.898219209557864e-06,
-      "loss": 0.134,
+      "epoch": 0.78,
+      "learning_rate": 1.3921174753322775e-05,
+      "loss": 0.1368,
       "step": 989
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.8807293163038572e-06,
-      "loss": 0.1327,
+      "epoch": 0.78,
+      "learning_rate": 1.390940445296276e-05,
+      "loss": 0.1457,
       "step": 990
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.8632834712660882e-06,
-      "loss": 0.1027,
+      "epoch": 0.78,
+      "learning_rate": 1.3897627755938372e-05,
+      "loss": 0.1345,
       "step": 991
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.845881782385156e-06,
-      "loss": 0.1169,
+      "epoch": 0.78,
+      "learning_rate": 1.388584468151893e-05,
+      "loss": 0.1436,
       "step": 992
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.828524357328456e-06,
-      "loss": 0.119,
+      "epoch": 0.78,
+      "learning_rate": 1.3874055248984191e-05,
+      "loss": 0.1765,
       "step": 993
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.8112113034895273e-06,
-      "loss": 0.1217,
+      "epoch": 0.79,
+      "learning_rate": 1.3862259477624317e-05,
+      "loss": 0.1442,
       "step": 994
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.7939427279873543e-06,
-      "loss": 0.1499,
+      "epoch": 0.79,
+      "learning_rate": 1.3850457386739846e-05,
+      "loss": 0.1166,
       "step": 995
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.7767187376657502e-06,
-      "loss": 0.1459,
+      "epoch": 0.79,
+      "learning_rate": 1.3838648995641645e-05,
+      "loss": 0.1272,
       "step": 996
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.7595394390926557e-06,
-      "loss": 0.1046,
+      "epoch": 0.79,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 0.1459,
       "step": 997
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.742404938559502e-06,
-      "loss": 0.1093,
+      "epoch": 0.79,
+      "learning_rate": 1.3815013390099068e-05,
+      "loss": 0.1337,
       "step": 998
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.72531534208054e-06,
-      "loss": 0.1437,
+      "epoch": 0.79,
+      "learning_rate": 1.3803186214327852e-05,
+      "loss": 0.1652,
       "step": 999
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.708270755392207e-06,
-      "loss": 0.1722,
+      "epoch": 0.79,
+      "learning_rate": 1.3791352815689174e-05,
+      "loss": 0.1694,
       "step": 1000
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.6912712839524336e-06,
-      "loss": 0.153,
+      "epoch": 0.79,
+      "learning_rate": 1.3779513213545132e-05,
+      "loss": 0.1169,
       "step": 1001
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.674317032940035e-06,
-      "loss": 0.0967,
+      "epoch": 0.79,
+      "learning_rate": 1.3767667427267976e-05,
+      "loss": 0.1857,
       "step": 1002
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.657408107254027e-06,
-      "loss": 0.1166,
+      "epoch": 0.79,
+      "learning_rate": 1.3755815476240076e-05,
+      "loss": 0.1148,
       "step": 1003
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.640544611512993e-06,
-      "loss": 0.1004,
+      "epoch": 0.79,
+      "learning_rate": 1.3743957379853885e-05,
+      "loss": 0.16,
       "step": 1004
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.6237266500544344e-06,
-      "loss": 0.1589,
+      "epoch": 0.79,
+      "learning_rate": 1.3732093157511914e-05,
+      "loss": 0.1822,
       "step": 1005
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.60695432693412e-06,
-      "loss": 0.1299,
+      "epoch": 0.79,
+      "learning_rate": 1.3720222828626699e-05,
+      "loss": 0.1564,
       "step": 1006
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.5902277459254533e-06,
-      "loss": 0.1389,
+      "epoch": 0.8,
+      "learning_rate": 1.3708346412620768e-05,
+      "loss": 0.1229,
       "step": 1007
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.5735470105188166e-06,
-      "loss": 0.1682,
+      "epoch": 0.8,
+      "learning_rate": 1.3696463928926602e-05,
+      "loss": 0.1334,
       "step": 1008
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.5569122239209366e-06,
-      "loss": 0.1118,
+      "epoch": 0.8,
+      "learning_rate": 1.3684575396986622e-05,
+      "loss": 0.1351,
       "step": 1009
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.5403234890542483e-06,
-      "loss": 0.1663,
+      "epoch": 0.8,
+      "learning_rate": 1.3672680836253129e-05,
+      "loss": 0.1243,
       "step": 1010
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.523780908556265e-06,
-      "loss": 0.1281,
+      "epoch": 0.8,
+      "learning_rate": 1.3660780266188306e-05,
+      "loss": 0.1445,
       "step": 1011
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.507284584778913e-06,
-      "loss": 0.1573,
+      "epoch": 0.8,
+      "learning_rate": 1.3648873706264159e-05,
+      "loss": 0.192,
       "step": 1012
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.490834619787943e-06,
-      "loss": 0.1495,
+      "epoch": 0.8,
+      "learning_rate": 1.363696117596249e-05,
+      "loss": 0.1471,
       "step": 1013
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.4744311153622636e-06,
-      "loss": 0.1109,
+      "epoch": 0.8,
+      "learning_rate": 1.3625042694774886e-05,
+      "loss": 0.1369,
       "step": 1014
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.4580741729933246e-06,
-      "loss": 0.1307,
+      "epoch": 0.8,
+      "learning_rate": 1.3613118282202653e-05,
+      "loss": 0.1237,
       "step": 1015
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.4417638938844877e-06,
-      "loss": 0.1447,
+      "epoch": 0.8,
+      "learning_rate": 1.3601187957756814e-05,
+      "loss": 0.1455,
       "step": 1016
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.425500378950405e-06,
-      "loss": 0.1495,
+      "epoch": 0.8,
+      "learning_rate": 1.358925174095806e-05,
+      "loss": 0.1332,
       "step": 1017
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.4092837288163807e-06,
-      "loss": 0.1116,
+      "epoch": 0.8,
+      "learning_rate": 1.3577309651336728e-05,
+      "loss": 0.1263,
       "step": 1018
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.3931140438177724e-06,
-      "loss": 0.1251,
+      "epoch": 0.8,
+      "learning_rate": 1.3565361708432754e-05,
+      "loss": 0.1415,
       "step": 1019
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.3769914239993432e-06,
-      "loss": 0.1301,
+      "epoch": 0.81,
+      "learning_rate": 1.3553407931795662e-05,
+      "loss": 0.1577,
       "step": 1020
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.3609159691146577e-06,
-      "loss": 0.132,
+      "epoch": 0.81,
+      "learning_rate": 1.3541448340984516e-05,
+      "loss": 0.1512,
       "step": 1021
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.3448877786254742e-06,
-      "loss": 0.1156,
+      "epoch": 0.81,
+      "learning_rate": 1.3529482955567896e-05,
+      "loss": 0.1255,
       "step": 1022
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.3289069517010953e-06,
-      "loss": 0.1458,
+      "epoch": 0.81,
+      "learning_rate": 1.3517511795123864e-05,
+      "loss": 0.1357,
       "step": 1023
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.312973587217798e-06,
-      "loss": 0.117,
+      "epoch": 0.81,
+      "learning_rate": 1.3505534879239923e-05,
+      "loss": 0.1377,
       "step": 1024
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.2970877837581917e-06,
-      "loss": 0.1222,
+      "epoch": 0.81,
+      "learning_rate": 1.3493552227513007e-05,
+      "loss": 0.1338,
       "step": 1025
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.2812496396106155e-06,
-      "loss": 0.1165,
+      "epoch": 0.81,
+      "learning_rate": 1.348156385954942e-05,
+      "loss": 0.1343,
       "step": 1026
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.2654592527685305e-06,
-      "loss": 0.1253,
+      "epoch": 0.81,
+      "learning_rate": 1.3469569794964832e-05,
+      "loss": 0.1231,
       "step": 1027
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.2497167209299277e-06,
-      "loss": 0.1566,
+      "epoch": 0.81,
+      "learning_rate": 1.3457570053384225e-05,
+      "loss": 0.1721,
       "step": 1028
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.234022141496689e-06,
-      "loss": 0.1167,
+      "epoch": 0.81,
+      "learning_rate": 1.3445564654441879e-05,
+      "loss": 0.1626,
       "step": 1029
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.2183756115740274e-06,
-      "loss": 0.1401,
+      "epoch": 0.81,
+      "learning_rate": 1.3433553617781318e-05,
+      "loss": 0.131,
       "step": 1030
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.202777227969851e-06,
-      "loss": 0.1691,
+      "epoch": 0.81,
+      "learning_rate": 1.3421536963055304e-05,
+      "loss": 0.1506,
       "step": 1031
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.1872270871941824e-06,
-      "loss": 0.1343,
+      "epoch": 0.82,
+      "learning_rate": 1.3409514709925777e-05,
+      "loss": 0.1159,
       "step": 1032
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.171725285458559e-06,
-      "loss": 0.1263,
+      "epoch": 0.82,
+      "learning_rate": 1.3397486878063852e-05,
+      "loss": 0.1342,
       "step": 1033
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.1562719186754287e-06,
-      "loss": 0.1284,
+      "epoch": 0.82,
+      "learning_rate": 1.3385453487149765e-05,
+      "loss": 0.1638,
       "step": 1034
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.140867082457575e-06,
-      "loss": 0.1513,
+      "epoch": 0.82,
+      "learning_rate": 1.3373414556872844e-05,
+      "loss": 0.1609,
       "step": 1035
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.1255108721175066e-06,
-      "loss": 0.1252,
+      "epoch": 0.82,
+      "learning_rate": 1.3361370106931486e-05,
+      "loss": 0.1479,
       "step": 1036
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.110203382666873e-06,
-      "loss": 0.1099,
+      "epoch": 0.82,
+      "learning_rate": 1.3349320157033121e-05,
+      "loss": 0.1416,
       "step": 1037
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.0949447088158812e-06,
-      "loss": 0.1269,
+      "epoch": 0.82,
+      "learning_rate": 1.3337264726894175e-05,
+      "loss": 0.1507,
       "step": 1038
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.079734944972717e-06,
-      "loss": 0.1399,
+      "epoch": 0.82,
+      "learning_rate": 1.3325203836240039e-05,
+      "loss": 0.1262,
       "step": 1039
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.064574185242929e-06,
-      "loss": 0.1677,
+      "epoch": 0.82,
+      "learning_rate": 1.3313137504805042e-05,
+      "loss": 0.0989,
       "step": 1040
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.049462523428891e-06,
-      "loss": 0.1197,
+      "epoch": 0.82,
+      "learning_rate": 1.3301065752332415e-05,
+      "loss": 0.1365,
       "step": 1041
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.0344000530291875e-06,
-      "loss": 0.1111,
+      "epoch": 0.82,
+      "learning_rate": 1.328898859857426e-05,
+      "loss": 0.1479,
       "step": 1042
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.019386867238048e-06,
-      "loss": 0.1371,
+      "epoch": 0.82,
+      "learning_rate": 1.3276906063291511e-05,
+      "loss": 0.1162,
       "step": 1043
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.0044230589447698e-06,
-      "loss": 0.1318,
+      "epoch": 0.82,
+      "learning_rate": 1.3264818166253917e-05,
+      "loss": 0.1306,
       "step": 1044
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.9895087207331422e-06,
-      "loss": 0.1212,
+      "epoch": 0.83,
+      "learning_rate": 1.3252724927239986e-05,
+      "loss": 0.145,
       "step": 1045
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.9746439448808785e-06,
-      "loss": 0.1719,
+      "epoch": 0.83,
+      "learning_rate": 1.3240626366036982e-05,
+      "loss": 0.1568,
       "step": 1046
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.959828823359037e-06,
-      "loss": 0.1774,
+      "epoch": 0.83,
+      "learning_rate": 1.3228522502440868e-05,
+      "loss": 0.1147,
       "step": 1047
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.945063447831452e-06,
-      "loss": 0.1444,
+      "epoch": 0.83,
+      "learning_rate": 1.3216413356256286e-05,
+      "loss": 0.1161,
       "step": 1048
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.9303479096541767e-06,
-      "loss": 0.1477,
+      "epoch": 0.83,
+      "learning_rate": 1.3204298947296521e-05,
+      "loss": 0.174,
       "step": 1049
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.9156822998749126e-06,
-      "loss": 0.1213,
+      "epoch": 0.83,
+      "learning_rate": 1.319217929538347e-05,
+      "loss": 0.1413,
       "step": 1050
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.9010667092324342e-06,
-      "loss": 0.1494,
+      "epoch": 0.83,
+      "learning_rate": 1.3180054420347603e-05,
+      "loss": 0.1118,
       "step": 1051
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.8865012281560523e-06,
-      "loss": 0.1501,
+      "epoch": 0.83,
+      "learning_rate": 1.3167924342027947e-05,
+      "loss": 0.1467,
       "step": 1052
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.8719859467650314e-06,
-      "loss": 0.1611,
+      "epoch": 0.83,
+      "learning_rate": 1.315578908027203e-05,
+      "loss": 0.1448,
       "step": 1053
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.8575209548680472e-06,
-      "loss": 0.1302,
+      "epoch": 0.83,
+      "learning_rate": 1.3143648654935875e-05,
+      "loss": 0.1202,
       "step": 1054
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.8431063419626172e-06,
-      "loss": 0.1354,
+      "epoch": 0.83,
+      "learning_rate": 1.313150308588394e-05,
+      "loss": 0.1325,
       "step": 1055
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.8287421972345697e-06,
-      "loss": 0.1032,
+      "epoch": 0.83,
+      "learning_rate": 1.311935239298911e-05,
+      "loss": 0.1569,
       "step": 1056
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.814428609557458e-06,
-      "loss": 0.1342,
+      "epoch": 0.83,
+      "learning_rate": 1.310719659613265e-05,
+      "loss": 0.1239,
       "step": 1057
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.8001656674920453e-06,
-      "loss": 0.1583,
+      "epoch": 0.84,
+      "learning_rate": 1.3095035715204171e-05,
+      "loss": 0.1354,
       "step": 1058
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.7859534592857353e-06,
-      "loss": 0.1392,
+      "epoch": 0.84,
+      "learning_rate": 1.3082869770101613e-05,
+      "loss": 0.1655,
       "step": 1059
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.7717920728720284e-06,
-      "loss": 0.1368,
+      "epoch": 0.84,
+      "learning_rate": 1.3070698780731194e-05,
+      "loss": 0.1343,
       "step": 1060
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.7576815958699932e-06,
-      "loss": 0.1336,
+      "epoch": 0.84,
+      "learning_rate": 1.305852276700739e-05,
+      "loss": 0.1479,
       "step": 1061
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.7436221155836953e-06,
-      "loss": 0.1251,
+      "epoch": 0.84,
+      "learning_rate": 1.304634174885289e-05,
+      "loss": 0.1374,
       "step": 1062
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.7296137190016916e-06,
-      "loss": 0.1131,
+      "epoch": 0.84,
+      "learning_rate": 1.3034155746198588e-05,
+      "loss": 0.1248,
       "step": 1063
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.7156564927964637e-06,
-      "loss": 0.1281,
+      "epoch": 0.84,
+      "learning_rate": 1.3021964778983513e-05,
+      "loss": 0.1748,
       "step": 1064
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.7017505233238962e-06,
-      "loss": 0.1192,
+      "epoch": 0.84,
+      "learning_rate": 1.3009768867154834e-05,
+      "loss": 0.1555,
       "step": 1065
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.6878958966227366e-06,
-      "loss": 0.1315,
+      "epoch": 0.84,
+      "learning_rate": 1.2997568030667802e-05,
+      "loss": 0.1353,
       "step": 1066
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.6740926984140771e-06,
-      "loss": 0.1454,
+      "epoch": 0.84,
+      "learning_rate": 1.2985362289485728e-05,
+      "loss": 0.1635,
       "step": 1067
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.6603410141007913e-06,
-      "loss": 0.1409,
+      "epoch": 0.84,
+      "learning_rate": 1.2973151663579948e-05,
+      "loss": 0.1552,
       "step": 1068
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.646640928767047e-06,
-      "loss": 0.1392,
+      "epoch": 0.84,
+      "learning_rate": 1.296093617292979e-05,
+      "loss": 0.1137,
       "step": 1069
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.6329925271777492e-06,
-      "loss": 0.137,
+      "epoch": 0.85,
+      "learning_rate": 1.2948715837522542e-05,
+      "loss": 0.1503,
       "step": 1070
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.6193958937780285e-06,
-      "loss": 0.1085,
+      "epoch": 0.85,
+      "learning_rate": 1.2936490677353422e-05,
+      "loss": 0.1843,
       "step": 1071
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.6058511126927178e-06,
-      "loss": 0.136,
+      "epoch": 0.85,
+      "learning_rate": 1.2924260712425536e-05,
+      "loss": 0.1334,
       "step": 1072
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.5923582677258254e-06,
-      "loss": 0.1249,
+      "epoch": 0.85,
+      "learning_rate": 1.2912025962749856e-05,
+      "loss": 0.1812,
       "step": 1073
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.5789174423600307e-06,
-      "loss": 0.1355,
+      "epoch": 0.85,
+      "learning_rate": 1.2899786448345186e-05,
+      "loss": 0.1695,
       "step": 1074
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.5655287197561497e-06,
-      "loss": 0.1412,
+      "epoch": 0.85,
+      "learning_rate": 1.288754218923812e-05,
+      "loss": 0.1286,
       "step": 1075
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.5521921827526332e-06,
-      "loss": 0.1488,
+      "epoch": 0.85,
+      "learning_rate": 1.2875293205463018e-05,
+      "loss": 0.1254,
       "step": 1076
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.5389079138650464e-06,
-      "loss": 0.1064,
+      "epoch": 0.85,
+      "learning_rate": 1.2863039517061968e-05,
+      "loss": 0.149,
       "step": 1077
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.5256759952855737e-06,
-      "loss": 0.1736,
+      "epoch": 0.85,
+      "learning_rate": 1.2850781144084763e-05,
+      "loss": 0.1678,
       "step": 1078
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.5124965088824795e-06,
-      "loss": 0.1454,
+      "epoch": 0.85,
+      "learning_rate": 1.2838518106588856e-05,
+      "loss": 0.1288,
       "step": 1079
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.4993695361996374e-06,
-      "loss": 0.1449,
+      "epoch": 0.85,
+      "learning_rate": 1.2826250424639329e-05,
+      "loss": 0.1497,
       "step": 1080
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.4862951584560037e-06,
-      "loss": 0.1262,
+      "epoch": 0.85,
+      "learning_rate": 1.2813978118308872e-05,
+      "loss": 0.1551,
       "step": 1081
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.473273456545119e-06,
-      "loss": 0.1353,
+      "epoch": 0.85,
+      "learning_rate": 1.2801701207677731e-05,
+      "loss": 0.1745,
       "step": 1082
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.460304511034606e-06,
-      "loss": 0.149,
+      "epoch": 0.86,
+      "learning_rate": 1.2789419712833698e-05,
+      "loss": 0.1373,
       "step": 1083
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.447388402165686e-06,
-      "loss": 0.1104,
+      "epoch": 0.86,
+      "learning_rate": 1.277713365387205e-05,
+      "loss": 0.1902,
       "step": 1084
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.4345252098526518e-06,
-      "loss": 0.1214,
+      "epoch": 0.86,
+      "learning_rate": 1.2764843050895548e-05,
+      "loss": 0.1599,
       "step": 1085
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.4217150136824121e-06,
-      "loss": 0.1336,
+      "epoch": 0.86,
+      "learning_rate": 1.2752547924014378e-05,
+      "loss": 0.1725,
       "step": 1086
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.4089578929139635e-06,
-      "loss": 0.1774,
+      "epoch": 0.86,
+      "learning_rate": 1.2740248293346134e-05,
+      "loss": 0.1307,
       "step": 1087
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.3962539264779206e-06,
-      "loss": 0.1284,
+      "epoch": 0.86,
+      "learning_rate": 1.2727944179015773e-05,
+      "loss": 0.1237,
       "step": 1088
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.3836031929760296e-06,
-      "loss": 0.1526,
+      "epoch": 0.86,
+      "learning_rate": 1.271563560115559e-05,
+      "loss": 0.1321,
       "step": 1089
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.371005770680659e-06,
-      "loss": 0.146,
+      "epoch": 0.86,
+      "learning_rate": 1.2703322579905191e-05,
+      "loss": 0.1249,
       "step": 1090
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.3584617375343478e-06,
-      "loss": 0.1228,
+      "epoch": 0.86,
+      "learning_rate": 1.269100513541144e-05,
+      "loss": 0.158,
       "step": 1091
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.3459711711492962e-06,
-      "loss": 0.1277,
+      "epoch": 0.86,
+      "learning_rate": 1.2678683287828451e-05,
+      "loss": 0.108,
       "step": 1092
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.3335341488068997e-06,
-      "loss": 0.1296,
+      "epoch": 0.86,
+      "learning_rate": 1.266635705731753e-05,
+      "loss": 0.1713,
       "step": 1093
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.321150747457265e-06,
-      "loss": 0.1346,
+      "epoch": 0.86,
+      "learning_rate": 1.2654026464047165e-05,
+      "loss": 0.1707,
       "step": 1094
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.308821043718743e-06,
-      "loss": 0.1183,
+      "epoch": 0.86,
+      "learning_rate": 1.2641691528192976e-05,
+      "loss": 0.1264,
       "step": 1095
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.2965451138774343e-06,
-      "loss": 0.1393,
+      "epoch": 0.87,
+      "learning_rate": 1.262935226993769e-05,
+      "loss": 0.1452,
       "step": 1096
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.2843230338867452e-06,
-      "loss": 0.1581,
-      "step": 1097
+      "epoch": 0.87,
+      "learning_rate": 1.261700870947111e-05,
+      "loss": 0.1494,
+      "step": 1097
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.2721548793668936e-06,
-      "loss": 0.1227,
+      "epoch": 0.87,
+      "learning_rate": 1.2604660866990072e-05,
+      "loss": 0.1679,
       "step": 1098
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.2600407256044544e-06,
-      "loss": 0.1638,
+      "epoch": 0.87,
+      "learning_rate": 1.2592308762698422e-05,
+      "loss": 0.1648,
       "step": 1099
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.2479806475518896e-06,
-      "loss": 0.1273,
+      "epoch": 0.87,
+      "learning_rate": 1.257995241680698e-05,
+      "loss": 0.1506,
       "step": 1100
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.2359747198270832e-06,
-      "loss": 0.1533,
+      "epoch": 0.87,
+      "learning_rate": 1.2567591849533507e-05,
+      "loss": 0.16,
       "step": 1101
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.2240230167128863e-06,
-      "loss": 0.1843,
+      "epoch": 0.87,
+      "learning_rate": 1.2555227081102663e-05,
+      "loss": 0.159,
       "step": 1102
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.212125612156647e-06,
-      "loss": 0.1274,
+      "epoch": 0.87,
+      "learning_rate": 1.2542858131745997e-05,
+      "loss": 0.1354,
       "step": 1103
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.2002825797697604e-06,
-      "loss": 0.1323,
+      "epoch": 0.87,
+      "learning_rate": 1.253048502170188e-05,
+      "loss": 0.1196,
       "step": 1104
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.188493992827211e-06,
-      "loss": 0.1273,
+      "epoch": 0.87,
+      "learning_rate": 1.2518107771215511e-05,
+      "loss": 0.1448,
       "step": 1105
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.1767599242671245e-06,
-      "loss": 0.1308,
+      "epoch": 0.87,
+      "learning_rate": 1.2505726400538849e-05,
+      "loss": 0.1692,
       "step": 1106
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.1650804466902987e-06,
-      "loss": 0.1276,
+      "epoch": 0.87,
+      "learning_rate": 1.24933409299306e-05,
+      "loss": 0.1514,
       "step": 1107
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.1534556323597824e-06,
-      "loss": 0.1516,
+      "epoch": 0.88,
+      "learning_rate": 1.2480951379656175e-05,
+      "loss": 0.1127,
       "step": 1108
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.1418855532004025e-06,
-      "loss": 0.134,
+      "epoch": 0.88,
+      "learning_rate": 1.2468557769987667e-05,
+      "loss": 0.1664,
       "step": 1109
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.130370280798334e-06,
-      "loss": 0.1335,
+      "epoch": 0.88,
+      "learning_rate": 1.2456160121203808e-05,
+      "loss": 0.1654,
       "step": 1110
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.1189098864006488e-06,
-      "loss": 0.1317,
+      "epoch": 0.88,
+      "learning_rate": 1.2443758453589934e-05,
+      "loss": 0.1454,
       "step": 1111
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.107504440914885e-06,
-      "loss": 0.1416,
+      "epoch": 0.88,
+      "learning_rate": 1.243135278743796e-05,
+      "loss": 0.1548,
       "step": 1112
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.0961540149085958e-06,
-      "loss": 0.1387,
+      "epoch": 0.88,
+      "learning_rate": 1.2418943143046346e-05,
+      "loss": 0.1539,
       "step": 1113
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.084858678608922e-06,
-      "loss": 0.1221,
+      "epoch": 0.88,
+      "learning_rate": 1.2406529540720063e-05,
+      "loss": 0.1516,
       "step": 1114
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.073618501902154e-06,
-      "loss": 0.1387,
+      "epoch": 0.88,
+      "learning_rate": 1.239411200077055e-05,
+      "loss": 0.1303,
       "step": 1115
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.0624335543332964e-06,
-      "loss": 0.125,
+      "epoch": 0.88,
+      "learning_rate": 1.2381690543515692e-05,
+      "loss": 0.1611,
       "step": 1116
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.0513039051056507e-06,
-      "loss": 0.1354,
+      "epoch": 0.88,
+      "learning_rate": 1.236926518927979e-05,
+      "loss": 0.1184,
       "step": 1117
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.040229623080362e-06,
-      "loss": 0.1358,
+      "epoch": 0.88,
+      "learning_rate": 1.2356835958393513e-05,
+      "loss": 0.1443,
       "step": 1118
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.0292107767760261e-06,
-      "loss": 0.1284,
+      "epoch": 0.88,
+      "learning_rate": 1.2344402871193876e-05,
+      "loss": 0.1376,
       "step": 1119
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.0182474343682346e-06,
-      "loss": 0.1293,
+      "epoch": 0.88,
+      "learning_rate": 1.2331965948024209e-05,
+      "loss": 0.1669,
       "step": 1120
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.007339663689173e-06,
-      "loss": 0.1234,
+      "epoch": 0.89,
+      "learning_rate": 1.2319525209234109e-05,
+      "loss": 0.1535,
       "step": 1121
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 9.964875322271895e-07,
-      "loss": 0.1043,
+      "epoch": 0.89,
+      "learning_rate": 1.230708067517942e-05,
+      "loss": 0.1348,
       "step": 1122
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 9.85691107126392e-07,
-      "loss": 0.1461,
+      "epoch": 0.89,
+      "learning_rate": 1.2294632366222201e-05,
+      "loss": 0.1432,
       "step": 1123
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 9.749504551862087e-07,
-      "loss": 0.1273,
+      "epoch": 0.89,
+      "learning_rate": 1.2282180302730683e-05,
+      "loss": 0.1752,
       "step": 1124
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 9.642656428610042e-07,
-      "loss": 0.1233,
+      "epoch": 0.89,
+      "learning_rate": 1.2269724505079234e-05,
+      "loss": 0.1378,
       "step": 1125
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 9.536367362596422e-07,
-      "loss": 0.1323,
+      "epoch": 0.89,
+      "learning_rate": 1.2257264993648345e-05,
+      "loss": 0.123,
       "step": 1126
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 9.430638011450932e-07,
-      "loss": 0.1294,
+      "epoch": 0.89,
+      "learning_rate": 1.2244801788824577e-05,
+      "loss": 0.1418,
       "step": 1127
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 9.325469029340195e-07,
-      "loss": 0.1316,
+      "epoch": 0.89,
+      "learning_rate": 1.223233491100053e-05,
+      "loss": 0.1347,
       "step": 1128
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 9.220861066963715e-07,
-      "loss": 0.125,
+      "epoch": 0.89,
+      "learning_rate": 1.2219864380574822e-05,
+      "loss": 0.136,
       "step": 1129
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 9.116814771549943e-07,
-      "loss": 0.1264,
+      "epoch": 0.89,
+      "learning_rate": 1.2207390217952044e-05,
+      "loss": 0.164,
       "step": 1130
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 9.013330786852093e-07,
-      "loss": 0.1618,
+      "epoch": 0.89,
+      "learning_rate": 1.2194912443542728e-05,
+      "loss": 0.1519,
       "step": 1131
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 8.910409753144344e-07,
-      "loss": 0.1174,
+      "epoch": 0.89,
+      "learning_rate": 1.2182431077763317e-05,
+      "loss": 0.1143,
       "step": 1132
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 8.808052307217707e-07,
-      "loss": 0.0969,
+      "epoch": 0.89,
+      "learning_rate": 1.2169946141036133e-05,
+      "loss": 0.1422,
       "step": 1133
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 8.7062590823763e-07,
-      "loss": 0.1237,
+      "epoch": 0.9,
+      "learning_rate": 1.2157457653789337e-05,
+      "loss": 0.1124,
       "step": 1134
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 8.605030708433149e-07,
-      "loss": 0.1344,
+      "epoch": 0.9,
+      "learning_rate": 1.2144965636456903e-05,
+      "loss": 0.127,
       "step": 1135
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 8.504367811706582e-07,
-      "loss": 0.1058,
+      "epoch": 0.9,
+      "learning_rate": 1.2132470109478577e-05,
+      "loss": 0.1226,
       "step": 1136
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 8.404271015016152e-07,
-      "loss": 0.1223,
+      "epoch": 0.9,
+      "learning_rate": 1.2119971093299852e-05,
+      "loss": 0.1287,
       "step": 1137
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 8.304740937678835e-07,
-      "loss": 0.1405,
+      "epoch": 0.9,
+      "learning_rate": 1.2107468608371924e-05,
+      "loss": 0.1746,
       "step": 1138
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 8.205778195505221e-07,
-      "loss": 0.142,
+      "epoch": 0.9,
+      "learning_rate": 1.209496267515167e-05,
+      "loss": 0.1093,
       "step": 1139
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 8.107383400795754e-07,
-      "loss": 0.1165,
+      "epoch": 0.9,
+      "learning_rate": 1.2082453314101607e-05,
+      "loss": 0.158,
       "step": 1140
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 8.009557162336823e-07,
-      "loss": 0.1229,
+      "epoch": 0.9,
+      "learning_rate": 1.2069940545689867e-05,
+      "loss": 0.2056,
       "step": 1141
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.912300085397051e-07,
-      "loss": 0.1359,
+      "epoch": 0.9,
+      "learning_rate": 1.2057424390390141e-05,
+      "loss": 0.1414,
       "step": 1142
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.815612771723591e-07,
-      "loss": 0.1432,
+      "epoch": 0.9,
+      "learning_rate": 1.2044904868681684e-05,
+      "loss": 0.1347,
       "step": 1143
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.719495819538325e-07,
-      "loss": 0.1377,
+      "epoch": 0.9,
+      "learning_rate": 1.203238200104924e-05,
+      "loss": 0.1404,
       "step": 1144
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.623949823534282e-07,
-      "loss": 0.1649,
+      "epoch": 0.9,
+      "learning_rate": 1.2019855807983036e-05,
+      "loss": 0.1248,
       "step": 1145
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.528975374871783e-07,
-      "loss": 0.1356,
+      "epoch": 0.91,
+      "learning_rate": 1.200732630997874e-05,
+      "loss": 0.1298,
       "step": 1146
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.434573061174966e-07,
-      "loss": 0.1296,
+      "epoch": 0.91,
+      "learning_rate": 1.1994793527537427e-05,
+      "loss": 0.1718,
       "step": 1147
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.340743466528021e-07,
-      "loss": 0.1578,
+      "epoch": 0.91,
+      "learning_rate": 1.1982257481165547e-05,
+      "loss": 0.1298,
       "step": 1148
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.247487171471624e-07,
-      "loss": 0.1542,
+      "epoch": 0.91,
+      "learning_rate": 1.1969718191374888e-05,
+      "loss": 0.1631,
       "step": 1149
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.154804752999344e-07,
-      "loss": 0.1497,
+      "epoch": 0.91,
+      "learning_rate": 1.1957175678682548e-05,
+      "loss": 0.1534,
       "step": 1150
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 7.062696784554124e-07,
-      "loss": 0.1151,
+      "epoch": 0.91,
+      "learning_rate": 1.1944629963610897e-05,
+      "loss": 0.1553,
       "step": 1151
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 6.971163836024575e-07,
-      "loss": 0.1234,
+      "epoch": 0.91,
+      "learning_rate": 1.1932081066687544e-05,
+      "loss": 0.1414,
       "step": 1152
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 6.880206473741646e-07,
-      "loss": 0.1415,
+      "epoch": 0.91,
+      "learning_rate": 1.1919529008445302e-05,
+      "loss": 0.1538,
       "step": 1153
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 6.789825260475014e-07,
-      "loss": 0.1332,
+      "epoch": 0.91,
+      "learning_rate": 1.1906973809422163e-05,
+      "loss": 0.1495,
       "step": 1154
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 6.700020755429559e-07,
-      "loss": 0.1279,
+      "epoch": 0.91,
+      "learning_rate": 1.1894415490161253e-05,
+      "loss": 0.1136,
       "step": 1155
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 6.610793514242075e-07,
-      "loss": 0.1436,
+      "epoch": 0.91,
+      "learning_rate": 1.1881854071210805e-05,
+      "loss": 0.137,
       "step": 1156
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 6.522144088977578e-07,
-      "loss": 0.125,
+      "epoch": 0.91,
+      "learning_rate": 1.1869289573124125e-05,
+      "loss": 0.1437,
       "step": 1157
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 6.434073028126164e-07,
-      "loss": 0.1414,
+      "epoch": 0.91,
+      "learning_rate": 1.1856722016459554e-05,
+      "loss": 0.1658,
       "step": 1158
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 6.346580876599395e-07,
-      "loss": 0.1407,
+      "epoch": 0.92,
+      "learning_rate": 1.1844151421780442e-05,
+      "loss": 0.1419,
       "step": 1159
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 6.259668175727063e-07,
-      "loss": 0.1343,
+      "epoch": 0.92,
+      "learning_rate": 1.1831577809655105e-05,
+      "loss": 0.1304,
       "step": 1160
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 6.17333546325376e-07,
-      "loss": 0.1296,
+      "epoch": 0.92,
+      "learning_rate": 1.18190012006568e-05,
+      "loss": 0.1425,
       "step": 1161
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 6.08758327333564e-07,
-      "loss": 0.1446,
+      "epoch": 0.92,
+      "learning_rate": 1.1806421615363685e-05,
+      "loss": 0.152,
       "step": 1162
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 6.002412136536972e-07,
-      "loss": 0.1435,
+      "epoch": 0.92,
+      "learning_rate": 1.179383907435879e-05,
+      "loss": 0.1345,
       "step": 1163
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 5.917822579827037e-07,
-      "loss": 0.1203,
+      "epoch": 0.92,
+      "learning_rate": 1.1781253598229982e-05,
+      "loss": 0.157,
       "step": 1164
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 5.833815126576714e-07,
-      "loss": 0.1339,
+      "epoch": 0.92,
+      "learning_rate": 1.1768665207569922e-05,
+      "loss": 0.1361,
       "step": 1165
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 5.750390296555319e-07,
-      "loss": 0.1219,
+      "epoch": 0.92,
+      "learning_rate": 1.1756073922976056e-05,
+      "loss": 0.1417,
       "step": 1166
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 5.667548605927365e-07,
-      "loss": 0.1252,
+      "epoch": 0.92,
+      "learning_rate": 1.1743479765050549e-05,
+      "loss": 0.1236,
       "step": 1167
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 5.585290567249424e-07,
-      "loss": 0.1125,
+      "epoch": 0.92,
+      "learning_rate": 1.1730882754400274e-05,
+      "loss": 0.1444,
       "step": 1168
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 5.503616689466829e-07,
-      "loss": 0.1358,
+      "epoch": 0.92,
+      "learning_rate": 1.1718282911636774e-05,
+      "loss": 0.1352,
       "step": 1169
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 5.422527477910666e-07,
-      "loss": 0.121,
+      "epoch": 0.92,
+      "learning_rate": 1.1705680257376224e-05,
+      "loss": 0.1525,
       "step": 1170
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 5.342023434294552e-07,
-      "loss": 0.132,
+      "epoch": 0.92,
+      "learning_rate": 1.1693074812239397e-05,
+      "loss": 0.165,
       "step": 1171
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 5.262105056711575e-07,
-      "loss": 0.1108,
+      "epoch": 0.93,
+      "learning_rate": 1.1680466596851635e-05,
+      "loss": 0.1423,
       "step": 1172
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 5.182772839631223e-07,
-      "loss": 0.1111,
+      "epoch": 0.93,
+      "learning_rate": 1.1667855631842815e-05,
+      "loss": 0.1107,
       "step": 1173
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 5.104027273896239e-07,
-      "loss": 0.1494,
+      "epoch": 0.93,
+      "learning_rate": 1.1655241937847305e-05,
+      "loss": 0.1421,
       "step": 1174
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 5.025868846719728e-07,
-      "loss": 0.1277,
+      "epoch": 0.93,
+      "learning_rate": 1.164262553550395e-05,
+      "loss": 0.1012,
       "step": 1175
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 4.94829804168202e-07,
-      "loss": 0.1166,
+      "epoch": 0.93,
+      "learning_rate": 1.1630006445456015e-05,
+      "loss": 0.1469,
       "step": 1176
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 4.871315338727711e-07,
-      "loss": 0.1416,
+      "epoch": 0.93,
+      "learning_rate": 1.161738468835117e-05,
+      "loss": 0.1332,
       "step": 1177
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 4.794921214162717e-07,
-      "loss": 0.1551,
+      "epoch": 0.93,
+      "learning_rate": 1.1604760284841446e-05,
+      "loss": 0.1307,
       "step": 1178
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 4.7191161406513363e-07,
-      "loss": 0.1167,
+      "epoch": 0.93,
+      "learning_rate": 1.1592133255583204e-05,
+      "loss": 0.1522,
       "step": 1179
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 4.6439005872132457e-07,
-      "loss": 0.1158,
+      "epoch": 0.93,
+      "learning_rate": 1.1579503621237102e-05,
+      "loss": 0.1559,
       "step": 1180
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 4.569275019220709e-07,
-      "loss": 0.1466,
+      "epoch": 0.93,
+      "learning_rate": 1.156687140246806e-05,
+      "loss": 0.1762,
       "step": 1181
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 4.4952398983956205e-07,
-      "loss": 0.1682,
+      "epoch": 0.93,
+      "learning_rate": 1.1554236619945229e-05,
+      "loss": 0.1346,
       "step": 1182
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 4.421795682806662e-07,
-      "loss": 0.1123,
+      "epoch": 0.93,
+      "learning_rate": 1.1541599294341952e-05,
+      "loss": 0.1354,
       "step": 1183
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 4.34894282686652e-07,
-      "loss": 0.1434,
+      "epoch": 0.94,
+      "learning_rate": 1.1528959446335735e-05,
+      "loss": 0.1233,
       "step": 1184
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 4.27668178132894e-07,
-      "loss": 0.1271,
+      "epoch": 0.94,
+      "learning_rate": 1.1516317096608207e-05,
+      "loss": 0.1531,
       "step": 1185
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 4.2050129932861394e-07,
-      "loss": 0.1254,
+      "epoch": 0.94,
+      "learning_rate": 1.1503672265845098e-05,
+      "loss": 0.1325,
       "step": 1186
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 4.1339369061658696e-07,
-      "loss": 0.1512,
+      "epoch": 0.94,
+      "learning_rate": 1.1491024974736191e-05,
+      "loss": 0.1183,
       "step": 1187
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 4.063453959728747e-07,
-      "loss": 0.1283,
+      "epoch": 0.94,
+      "learning_rate": 1.1478375243975298e-05,
+      "loss": 0.1235,
       "step": 1188
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 3.9935645900654906e-07,
-      "loss": 0.1514,
+      "epoch": 0.94,
+      "learning_rate": 1.1465723094260219e-05,
+      "loss": 0.1434,
       "step": 1189
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 3.924269229594335e-07,
-      "loss": 0.1357,
+      "epoch": 0.94,
+      "learning_rate": 1.1453068546292718e-05,
+      "loss": 0.1572,
       "step": 1190
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 3.855568307058155e-07,
-      "loss": 0.1205,
+      "epoch": 0.94,
+      "learning_rate": 1.1440411620778478e-05,
+      "loss": 0.1718,
       "step": 1191
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 3.7874622475220336e-07,
-      "loss": 0.1191,
+      "epoch": 0.94,
+      "learning_rate": 1.1427752338427075e-05,
+      "loss": 0.1393,
       "step": 1192
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.7199514723704754e-07,
-      "loss": 0.1203,
+      "epoch": 0.94,
+      "learning_rate": 1.141509071995194e-05,
+      "loss": 0.1469,
       "step": 1193
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.6530363993048654e-07,
-      "loss": 0.1058,
+      "epoch": 0.94,
+      "learning_rate": 1.1402426786070326e-05,
+      "loss": 0.1434,
       "step": 1194
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.58671744234087e-07,
-      "loss": 0.1151,
+      "epoch": 0.94,
+      "learning_rate": 1.1389760557503275e-05,
+      "loss": 0.1407,
       "step": 1195
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.5209950118058835e-07,
-      "loss": 0.1348,
+      "epoch": 0.94,
+      "learning_rate": 1.1377092054975586e-05,
+      "loss": 0.153,
       "step": 1196
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.455869514336474e-07,
-      "loss": 0.1615,
+      "epoch": 0.95,
+      "learning_rate": 1.1364421299215773e-05,
+      "loss": 0.1502,
       "step": 1197
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.3913413528758877e-07,
-      "loss": 0.1454,
+      "epoch": 0.95,
+      "learning_rate": 1.135174831095604e-05,
+      "loss": 0.1178,
       "step": 1198
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.327410926671526e-07,
-      "loss": 0.1103,
+      "epoch": 0.95,
+      "learning_rate": 1.133907311093225e-05,
+      "loss": 0.1195,
       "step": 1199
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.264078631272505e-07,
-      "loss": 0.142,
+      "epoch": 0.95,
+      "learning_rate": 1.1326395719883876e-05,
+      "loss": 0.1378,
       "step": 1200
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.2013448585272333e-07,
-      "loss": 0.1428,
+      "epoch": 0.95,
+      "learning_rate": 1.1313716158553978e-05,
+      "loss": 0.1202,
       "step": 1201
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.139209996580872e-07,
-      "loss": 0.1244,
+      "epoch": 0.95,
+      "learning_rate": 1.130103444768917e-05,
+      "loss": 0.1326,
       "step": 1202
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.07767442987309e-07,
-      "loss": 0.1048,
+      "epoch": 0.95,
+      "learning_rate": 1.1288350608039577e-05,
+      "loss": 0.1426,
       "step": 1203
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.016738539135566e-07,
-      "loss": 0.1411,
+      "epoch": 0.95,
+      "learning_rate": 1.1275664660358818e-05,
+      "loss": 0.1265,
       "step": 1204
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 2.9564027013896713e-07,
-      "loss": 0.1516,
+      "epoch": 0.95,
+      "learning_rate": 1.1262976625403954e-05,
+      "loss": 0.15,
       "step": 1205
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.896667289944166e-07,
-      "loss": 0.1236,
+      "epoch": 0.95,
+      "learning_rate": 1.1250286523935456e-05,
+      "loss": 0.1558,
       "step": 1206
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.837532674392862e-07,
-      "loss": 0.0942,
+      "epoch": 0.95,
+      "learning_rate": 1.1237594376717188e-05,
+      "loss": 0.1422,
       "step": 1207
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.7789992206123106e-07,
-      "loss": 0.1439,
+      "epoch": 0.95,
+      "learning_rate": 1.122490020451635e-05,
+      "loss": 0.1358,
       "step": 1208
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.721067290759594e-07,
-      "loss": 0.1613,
+      "epoch": 0.95,
+      "learning_rate": 1.1212204028103469e-05,
+      "loss": 0.1317,
       "step": 1209
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.6637372432700483e-07,
-      "loss": 0.1216,
+      "epoch": 0.96,
+      "learning_rate": 1.1199505868252336e-05,
+      "loss": 0.1228,
       "step": 1210
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.6070094328550344e-07,
-      "loss": 0.1573,
+      "epoch": 0.96,
+      "learning_rate": 1.1186805745739999e-05,
+      "loss": 0.1194,
       "step": 1211
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.5508842104998265e-07,
-      "loss": 0.1479,
+      "epoch": 0.96,
+      "learning_rate": 1.1174103681346711e-05,
+      "loss": 0.1345,
       "step": 1212
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.4953619234612816e-07,
-      "loss": 0.1694,
+      "epoch": 0.96,
+      "learning_rate": 1.1161399695855903e-05,
+      "loss": 0.1174,
       "step": 1213
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.4404429152658726e-07,
-      "loss": 0.129,
+      "epoch": 0.96,
+      "learning_rate": 1.1148693810054152e-05,
+      "loss": 0.1655,
       "step": 1214
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.386127525707427e-07,
-      "loss": 0.1313,
+      "epoch": 0.96,
+      "learning_rate": 1.1135986044731144e-05,
+      "loss": 0.1142,
       "step": 1215
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.332416090845102e-07,
-      "loss": 0.1096,
+      "epoch": 0.96,
+      "learning_rate": 1.112327642067964e-05,
+      "loss": 0.13,
       "step": 1216
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.279308943001246e-07,
-      "loss": 0.1484,
+      "epoch": 0.96,
+      "learning_rate": 1.111056495869544e-05,
+      "loss": 0.1433,
       "step": 1217
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.2268064107594302e-07,
-      "loss": 0.1172,
+      "epoch": 0.96,
+      "learning_rate": 1.1097851679577351e-05,
+      "loss": 0.1341,
       "step": 1218
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.1749088189622848e-07,
-      "loss": 0.1484,
+      "epoch": 0.96,
+      "learning_rate": 1.1085136604127161e-05,
+      "loss": 0.1867,
       "step": 1219
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.1236164887096345e-07,
-      "loss": 0.1427,
+      "epoch": 0.96,
+      "learning_rate": 1.1072419753149585e-05,
+      "loss": 0.1511,
       "step": 1220
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.0729297373563995e-07,
-      "loss": 0.1105,
+      "epoch": 0.96,
+      "learning_rate": 1.105970114745225e-05,
+      "loss": 0.1571,
       "step": 1221
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.0228488785106636e-07,
-      "loss": 0.1247,
+      "epoch": 0.97,
+      "learning_rate": 1.104698080784566e-05,
+      "loss": 0.1595,
       "step": 1222
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.9733742220317876e-07,
-      "loss": 0.139,
+      "epoch": 0.97,
+      "learning_rate": 1.1034258755143141e-05,
+      "loss": 0.1411,
       "step": 1223
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.9245060740284316e-07,
-      "loss": 0.1609,
+      "epoch": 0.97,
+      "learning_rate": 1.1021535010160838e-05,
+      "loss": 0.1298,
       "step": 1224
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.8762447368566582e-07,
-      "loss": 0.0999,
+      "epoch": 0.97,
+      "learning_rate": 1.1008809593717653e-05,
+      "loss": 0.1755,
       "step": 1225
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.828590509118089e-07,
-      "loss": 0.1661,
+      "epoch": 0.97,
+      "learning_rate": 1.0996082526635227e-05,
+      "loss": 0.1675,
       "step": 1226
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.7815436856580714e-07,
-      "loss": 0.1401,
+      "epoch": 0.97,
+      "learning_rate": 1.0983353829737909e-05,
+      "loss": 0.1505,
       "step": 1227
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.7351045575638047e-07,
-      "loss": 0.119,
+      "epoch": 0.97,
+      "learning_rate": 1.0970623523852699e-05,
+      "loss": 0.1953,
       "step": 1228
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.6892734121625954e-07,
-      "loss": 0.1142,
+      "epoch": 0.97,
+      "learning_rate": 1.0957891629809248e-05,
+      "loss": 0.1054,
       "step": 1229
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.644050533020014e-07,
-      "loss": 0.1285,
+      "epoch": 0.97,
+      "learning_rate": 1.0945158168439786e-05,
+      "loss": 0.1543,
       "step": 1230
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.5994361999382202e-07,
-      "loss": 0.1219,
+      "epoch": 0.97,
+      "learning_rate": 1.0932423160579126e-05,
+      "loss": 0.1967,
       "step": 1231
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.5554306889541514e-07,
-      "loss": 0.1527,
+      "epoch": 0.97,
+      "learning_rate": 1.0919686627064603e-05,
+      "loss": 0.1588,
       "step": 1232
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.51203427233787e-07,
-      "loss": 0.1191,
+      "epoch": 0.97,
+      "learning_rate": 1.0906948588736044e-05,
+      "loss": 0.1592,
       "step": 1233
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.4692472185908635e-07,
-      "loss": 0.1099,
+      "epoch": 0.97,
+      "learning_rate": 1.0894209066435746e-05,
+      "loss": 0.1662,
       "step": 1234
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.42706979244438e-07,
-      "loss": 0.1492,
+      "epoch": 0.98,
+      "learning_rate": 1.0881468081008428e-05,
+      "loss": 0.1637,
       "step": 1235
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.3855022548577845e-07,
-      "loss": 0.144,
+      "epoch": 0.98,
+      "learning_rate": 1.0868725653301206e-05,
+      "loss": 0.1854,
       "step": 1236
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.344544863016961e-07,
-      "loss": 0.1532,
+      "epoch": 0.98,
+      "learning_rate": 1.085598180416356e-05,
+      "loss": 0.1248,
       "step": 1237
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.304197870332713e-07,
-      "loss": 0.1497,
+      "epoch": 0.98,
+      "learning_rate": 1.0843236554447288e-05,
+      "loss": 0.131,
       "step": 1238
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.2644615264391536e-07,
-      "loss": 0.1043,
+      "epoch": 0.98,
+      "learning_rate": 1.0830489925006485e-05,
+      "loss": 0.1401,
       "step": 1239
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.225336077192274e-07,
-      "loss": 0.0963,
+      "epoch": 0.98,
+      "learning_rate": 1.0817741936697499e-05,
+      "loss": 0.1583,
       "step": 1240
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.1868217646682889e-07,
-      "loss": 0.1232,
+      "epoch": 0.98,
+      "learning_rate": 1.0804992610378907e-05,
+      "loss": 0.1215,
       "step": 1241
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.1489188271622154e-07,
-      "loss": 0.1072,
+      "epoch": 0.98,
+      "learning_rate": 1.0792241966911472e-05,
+      "loss": 0.148,
       "step": 1242
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.1116274991864073e-07,
-      "loss": 0.1263,
+      "epoch": 0.98,
+      "learning_rate": 1.077949002715811e-05,
+      "loss": 0.1633,
       "step": 1243
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.0749480114690458e-07,
-      "loss": 0.1107,
+      "epoch": 0.98,
+      "learning_rate": 1.0766736811983864e-05,
+      "loss": 0.1467,
       "step": 1244
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.0388805909527621e-07,
-      "loss": 0.1743,
+      "epoch": 0.98,
+      "learning_rate": 1.0753982342255863e-05,
+      "loss": 0.1205,
       "step": 1245
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.003425460793217e-07,
-      "loss": 0.1223,
+      "epoch": 0.98,
+      "learning_rate": 1.0741226638843276e-05,
+      "loss": 0.1495,
       "step": 1246
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 9.685828403577124e-08,
-      "loss": 0.1193,
+      "epoch": 0.98,
+      "learning_rate": 1.072846972261731e-05,
+      "loss": 0.1223,
       "step": 1247
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 9.343529452238598e-08,
-      "loss": 0.1379,
+      "epoch": 0.99,
+      "learning_rate": 1.0715711614451146e-05,
+      "loss": 0.1373,
       "step": 1248
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 9.00735987178214e-08,
-      "loss": 0.1318,
+      "epoch": 0.99,
+      "learning_rate": 1.0702952335219912e-05,
+      "loss": 0.1389,
       "step": 1249
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 8.677321742149525e-08,
-      "loss": 0.1224,
+      "epoch": 0.99,
+      "learning_rate": 1.0690191905800659e-05,
+      "loss": 0.1387,
       "step": 1250
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 8.353417105346762e-08,
-      "loss": 0.1554,
+      "epoch": 0.99,
+      "learning_rate": 1.067743034707232e-05,
+      "loss": 0.1396,
       "step": 1251
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 8.035647965430215e-08,
-      "loss": 0.1392,
+      "epoch": 0.99,
+      "learning_rate": 1.066466767991567e-05,
+      "loss": 0.1072,
       "step": 1252
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 7.724016288495173e-08,
-      "loss": 0.1354,
+      "epoch": 0.99,
+      "learning_rate": 1.0651903925213304e-05,
+      "loss": 0.1498,
       "step": 1253
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 7.418524002663296e-08,
-      "loss": 0.1282,
+      "epoch": 0.99,
+      "learning_rate": 1.0639139103849591e-05,
+      "loss": 0.1157,
       "step": 1254
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 7.119172998070412e-08,
-      "loss": 0.1335,
+      "epoch": 0.99,
+      "learning_rate": 1.062637323671065e-05,
+      "loss": 0.1442,
       "step": 1255
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 6.825965126855517e-08,
-      "loss": 0.16,
+      "epoch": 0.99,
+      "learning_rate": 1.0613606344684309e-05,
+      "loss": 0.114,
       "step": 1256
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 6.53890220314879e-08,
-      "loss": 0.1734,
+      "epoch": 0.99,
+      "learning_rate": 1.060083844866007e-05,
+      "loss": 0.1339,
       "step": 1257
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 6.25798600306049e-08,
-      "loss": 0.1393,
+      "epoch": 0.99,
+      "learning_rate": 1.0588069569529085e-05,
+      "loss": 0.1187,
       "step": 1258
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 5.983218264669965e-08,
-      "loss": 0.1155,
+      "epoch": 0.99,
+      "learning_rate": 1.0575299728184105e-05,
+      "loss": 0.1313,
       "step": 1259
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 5.714600688014882e-08,
-      "loss": 0.1148,
+      "epoch": 1.0,
+      "learning_rate": 1.0562528945519463e-05,
+      "loss": 0.131,
       "step": 1260
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 5.4521349350808996e-08,
-      "loss": 0.1144,
+      "epoch": 1.0,
+      "learning_rate": 1.0549757242431032e-05,
+      "loss": 0.1393,
       "step": 1261
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 5.195822629791125e-08,
-      "loss": 0.1296,
+      "epoch": 1.0,
+      "learning_rate": 1.0536984639816183e-05,
+      "loss": 0.1331,
       "step": 1262
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 4.945665357996121e-08,
-      "loss": 0.1425,
+      "epoch": 1.0,
+      "learning_rate": 1.0524211158573772e-05,
+      "loss": 0.1402,
       "step": 1263
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 4.701664667464245e-08,
-      "loss": 0.1125,
+      "epoch": 1.0,
+      "learning_rate": 1.0511436819604082e-05,
+      "loss": 0.1368,
       "step": 1264
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 4.463822067871993e-08,
-      "loss": 0.1179,
+      "epoch": 1.0,
+      "learning_rate": 1.0498661643808801e-05,
+      "loss": 0.1474,
       "step": 1265
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 4.2321390307944464e-08,
-      "loss": 0.1641,
+      "epoch": 1.0,
+      "learning_rate": 1.0485885652090992e-05,
+      "loss": 0.1522,
       "step": 1266
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 4.006616989696621e-08,
-      "loss": 0.1232,
+      "epoch": 1.0,
+      "learning_rate": 1.0473108865355046e-05,
+      "loss": 0.1065,
       "step": 1267
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 3.787257339924133e-08,
-      "loss": 0.1457,
+      "epoch": 1.0,
+      "learning_rate": 1.0460331304506658e-05,
+      "loss": 0.1284,
       "step": 1268
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 3.5740614386947644e-08,
-      "loss": 0.1624,
+      "epoch": 1.0,
+      "learning_rate": 1.044755299045279e-05,
+      "loss": 0.1083,
       "step": 1269
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 3.367030605090249e-08,
-      "loss": 0.1359,
+      "epoch": 1.0,
+      "learning_rate": 1.0434773944101637e-05,
+      "loss": 0.1154,
       "step": 1270
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 3.166166120047831e-08,
-      "loss": 0.1428,
+      "epoch": 1.0,
+      "learning_rate": 1.0421994186362591e-05,
+      "loss": 0.1031,
       "step": 1271
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 2.971469226352275e-08,
-      "loss": 0.1678,
+      "epoch": 1.0,
+      "learning_rate": 1.0409213738146207e-05,
+      "loss": 0.1217,
       "step": 1272
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 2.7829411286287577e-08,
-      "loss": 0.1122,
+      "epoch": 1.01,
+      "learning_rate": 1.039643262036417e-05,
+      "loss": 0.1375,
       "step": 1273
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 2.600582993334544e-08,
-      "loss": 0.1486,
+      "epoch": 1.01,
+      "learning_rate": 1.0383650853929261e-05,
+      "loss": 0.1135,
       "step": 1274
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 2.4243959487524337e-08,
-      "loss": 0.1507,
+      "epoch": 1.01,
+      "learning_rate": 1.0370868459755325e-05,
+      "loss": 0.1231,
       "step": 1275
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 2.254381084983659e-08,
-      "loss": 0.1028,
+      "epoch": 1.01,
+      "learning_rate": 1.0358085458757233e-05,
+      "loss": 0.0985,
       "step": 1276
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 2.0905394539409984e-08,
-      "loss": 0.1249,
+      "epoch": 1.01,
+      "learning_rate": 1.0345301871850843e-05,
+      "loss": 0.1025,
       "step": 1277
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.9328720693420065e-08,
-      "loss": 0.1239,
+      "epoch": 1.01,
+      "learning_rate": 1.0332517719952982e-05,
+      "loss": 0.1292,
       "step": 1278
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.781379906703573e-08,
-      "loss": 0.1354,
+      "epoch": 1.01,
+      "learning_rate": 1.0319733023981392e-05,
+      "loss": 0.1079,
       "step": 1279
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.636063903335039e-08,
-      "loss": 0.1522,
+      "epoch": 1.01,
+      "learning_rate": 1.030694780485471e-05,
+      "loss": 0.088,
       "step": 1280
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.496924958332868e-08,
-      "loss": 0.1352,
+      "epoch": 1.01,
+      "learning_rate": 1.0294162083492429e-05,
+      "loss": 0.1092,
       "step": 1281
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.3639639325748744e-08,
-      "loss": 0.1457,
+      "epoch": 1.01,
+      "learning_rate": 1.0281375880814864e-05,
+      "loss": 0.112,
       "step": 1282
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.2371816487150023e-08,
-      "loss": 0.131,
+      "epoch": 1.01,
+      "learning_rate": 1.0268589217743114e-05,
+      "loss": 0.123,
       "step": 1283
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.1165788911781106e-08,
-      "loss": 0.1323,
+      "epoch": 1.01,
+      "learning_rate": 1.0255802115199034e-05,
+      "loss": 0.1327,
       "step": 1284
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.0021564061554189e-08,
-      "loss": 0.1082,
+      "epoch": 1.02,
+      "learning_rate": 1.0243014594105201e-05,
+      "loss": 0.1243,
       "step": 1285
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 8.93914901599291e-09,
-      "loss": 0.1111,
+      "epoch": 1.02,
+      "learning_rate": 1.0230226675384868e-05,
+      "loss": 0.1259,
       "step": 1286
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 7.918550472199027e-09,
-      "loss": 0.1308,
+      "epoch": 1.02,
+      "learning_rate": 1.021743837996195e-05,
+      "loss": 0.1006,
       "step": 1287
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 6.959774744796921e-09,
-      "loss": 0.1291,
+      "epoch": 1.02,
+      "learning_rate": 1.0204649728760969e-05,
+      "loss": 0.1332,
       "step": 1288
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 6.062827765906942e-09,
-      "loss": 0.1315,
+      "epoch": 1.02,
+      "learning_rate": 1.0191860742707034e-05,
+      "loss": 0.1244,
       "step": 1289
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 5.227715085103224e-09,
-      "loss": 0.1454,
+      "epoch": 1.02,
+      "learning_rate": 1.0179071442725801e-05,
+      "loss": 0.125,
       "step": 1290
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 4.454441869377047e-09,
-      "loss": 0.131,
+      "epoch": 1.02,
+      "learning_rate": 1.0166281849743438e-05,
+      "loss": 0.0927,
       "step": 1291
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 3.7430129031124085e-09,
-      "loss": 0.1419,
+      "epoch": 1.02,
+      "learning_rate": 1.0153491984686595e-05,
+      "loss": 0.101,
       "step": 1292
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 3.0934325880460636e-09,
-      "loss": 0.1497,
+      "epoch": 1.02,
+      "learning_rate": 1.0140701868482365e-05,
+      "loss": 0.0988,
       "step": 1293
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 2.5057049432519744e-09,
-      "loss": 0.1219,
+      "epoch": 1.02,
+      "learning_rate": 1.0127911522058256e-05,
+      "loss": 0.1052,
       "step": 1294
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.979833605104675e-09,
-      "loss": 0.129,
+      "epoch": 1.02,
+      "learning_rate": 1.0115120966342145e-05,
+      "loss": 0.1111,
       "step": 1295
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.5158218272670611e-09,
-      "loss": 0.1303,
+      "epoch": 1.02,
+      "learning_rate": 1.0102330222262257e-05,
+      "loss": 0.104,
       "step": 1296
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.1136724806637411e-09,
-      "loss": 0.1225,
+      "epoch": 1.02,
+      "learning_rate": 1.0089539310747127e-05,
+      "loss": 0.1495,
       "step": 1297
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 7.733880534654959e-10,
-      "loss": 0.1472,
+      "epoch": 1.03,
+      "learning_rate": 1.007674825272556e-05,
+      "loss": 0.1198,
       "step": 1298
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 4.94970651073734e-10,
-      "loss": 0.142,
+      "epoch": 1.03,
+      "learning_rate": 1.0063957069126602e-05,
+      "loss": 0.0834,
       "step": 1299
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 2.784219961060597e-10,
-      "loss": 0.1168,
+      "epoch": 1.03,
+      "learning_rate": 1.0051165780879503e-05,
+      "loss": 0.0929,
       "step": 1300
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.237434283907213e-10,
-      "loss": 0.1114,
+      "epoch": 1.03,
+      "learning_rate": 1.0038374408913684e-05,
+      "loss": 0.0989,
       "step": 1301
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 3.09359049488478e-11,
-      "loss": 0.106,
+      "epoch": 1.03,
+      "learning_rate": 1.002558297415871e-05,
+      "loss": 0.1298,
       "step": 1302
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 0.0,
-      "loss": 0.1397,
+      "epoch": 1.03,
+      "learning_rate": 1.0012791497544238e-05,
+      "loss": 0.1526,
       "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
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.1058,
+      "step": 1304
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.987208502455767e-06,
+      "loss": 0.1386,
+      "step": 1305
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.974417025841293e-06,
+      "loss": 0.1363,
+      "step": 1306
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.961625591086321e-06,
+      "loss": 0.1063,
+      "step": 1307
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.9488342191205e-06,
+      "loss": 0.1177,
+      "step": 1308
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.936042930873403e-06,
+      "loss": 0.1196,
+      "step": 1309
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.923251747274441e-06,
+      "loss": 0.1472,
+      "step": 1310
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.910460689252876e-06,
+      "loss": 0.0934,
+      "step": 1311
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.897669777737745e-06,
+      "loss": 0.1014,
+      "step": 1312
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.884879033657859e-06,
+      "loss": 0.0866,
+      "step": 1313
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.872088477941748e-06,
+      "loss": 0.1465,
+      "step": 1314
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.859298131517639e-06,
+      "loss": 0.1132,
+      "step": 1315
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.846508015313407e-06,
+      "loss": 0.1006,
+      "step": 1316
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.833718150256567e-06,
+      "loss": 0.1733,
+      "step": 1317
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.820928557274202e-06,
+      "loss": 0.1131,
+      "step": 1318
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.808139257292971e-06,
+      "loss": 0.1271,
+      "step": 1319
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.795350271239034e-06,
+      "loss": 0.1135,
+      "step": 1320
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.782561620038055e-06,
+      "loss": 0.0894,
+      "step": 1321
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.769773324615133e-06,
+      "loss": 0.1249,
+      "step": 1322
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.756985405894802e-06,
+      "loss": 0.1321,
+      "step": 1323
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.744197884800968e-06,
+      "loss": 0.1151,
+      "step": 1324
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.731410782256889e-06,
+      "loss": 0.1065,
+      "step": 1325
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.718624119185138e-06,
+      "loss": 0.1257,
+      "step": 1326
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.705837916507575e-06,
+      "loss": 0.1026,
+      "step": 1327
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.693052195145292e-06,
+      "loss": 0.1129,
+      "step": 1328
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.680266976018613e-06,
+      "loss": 0.1324,
+      "step": 1329
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.66748228004702e-06,
+      "loss": 0.1367,
+      "step": 1330
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.654698128149162e-06,
+      "loss": 0.0964,
+      "step": 1331
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.64191454124277e-06,
+      "loss": 0.1393,
+      "step": 1332
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.62913154024468e-06,
+      "loss": 0.1442,
+      "step": 1333
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.61634914607074e-06,
+      "loss": 0.1195,
+      "step": 1334
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.603567379635836e-06,
+      "loss": 0.1049,
+      "step": 1335
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.590786261853798e-06,
+      "loss": 0.0886,
+      "step": 1336
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.578005813637414e-06,
+      "loss": 0.1292,
+      "step": 1337
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.565226055898366e-06,
+      "loss": 0.1288,
+      "step": 1338
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.552447009547214e-06,
+      "loss": 0.0917,
+      "step": 1339
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.539668695493344e-06,
+      "loss": 0.1371,
+      "step": 1340
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.52689113464496e-06,
+      "loss": 0.1031,
+      "step": 1341
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.514114347909011e-06,
+      "loss": 0.116,
+      "step": 1342
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.501338356191204e-06,
+      "loss": 0.1036,
+      "step": 1343
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.488563180395922e-06,
+      "loss": 0.1125,
+      "step": 1344
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.475788841426232e-06,
+      "loss": 0.1235,
+      "step": 1345
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.463015360183819e-06,
+      "loss": 0.1076,
+      "step": 1346
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.450242757568975e-06,
+      "loss": 0.1176,
+      "step": 1347
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.43747105448054e-06,
+      "loss": 0.1097,
+      "step": 1348
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.424700271815901e-06,
+      "loss": 0.1003,
+      "step": 1349
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.41193043047092e-06,
+      "loss": 0.1074,
+      "step": 1350
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.399161551339933e-06,
+      "loss": 0.1317,
+      "step": 1351
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.386393655315696e-06,
+      "loss": 0.1223,
+      "step": 1352
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.373626763289352e-06,
+      "loss": 0.1052,
+      "step": 1353
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.36086089615041e-06,
+      "loss": 0.1279,
+      "step": 1354
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.3480960747867e-06,
+      "loss": 0.1226,
+      "step": 1355
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.335332320084331e-06,
+      "loss": 0.1329,
+      "step": 1356
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.322569652927685e-06,
+      "loss": 0.1088,
+      "step": 1357
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.309808094199343e-06,
+      "loss": 0.0987,
+      "step": 1358
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.297047664780093e-06,
+      "loss": 0.1162,
+      "step": 1359
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.284288385548858e-06,
+      "loss": 0.1067,
+      "step": 1360
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.271530277382695e-06,
+      "loss": 0.1264,
+      "step": 1361
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.258773361156725e-06,
+      "loss": 0.125,
+      "step": 1362
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.246017657744142e-06,
+      "loss": 0.0833,
+      "step": 1363
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.233263188016138e-06,
+      "loss": 0.1018,
+      "step": 1364
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.220509972841893e-06,
+      "loss": 0.1202,
+      "step": 1365
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.207758033088533e-06,
+      "loss": 0.088,
+      "step": 1366
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.195007389621098e-06,
+      "loss": 0.1188,
+      "step": 1367
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.182258063302504e-06,
+      "loss": 0.0954,
+      "step": 1368
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.16951007499352e-06,
+      "loss": 0.1119,
+      "step": 1369
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.156763445552714e-06,
+      "loss": 0.1262,
+      "step": 1370
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.144018195836445e-06,
+      "loss": 0.1054,
+      "step": 1371
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.131274346698797e-06,
+      "loss": 0.0927,
+      "step": 1372
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.118531918991578e-06,
+      "loss": 0.1281,
+      "step": 1373
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.105790933564259e-06,
+      "loss": 0.0911,
+      "step": 1374
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.09305141126396e-06,
+      "loss": 0.1024,
+      "step": 1375
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.080313372935399e-06,
+      "loss": 0.1323,
+      "step": 1376
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.067576839420876e-06,
+      "loss": 0.1189,
+      "step": 1377
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.054841831560216e-06,
+      "loss": 0.1198,
+      "step": 1378
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.042108370190757e-06,
+      "loss": 0.1011,
+      "step": 1379
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.029376476147303e-06,
+      "loss": 0.08,
+      "step": 1380
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.016646170262096e-06,
+      "loss": 0.1158,
+      "step": 1381
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.003917473364774e-06,
+      "loss": 0.1215,
+      "step": 1382
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.991190406282352e-06,
+      "loss": 0.1092,
+      "step": 1383
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.978464989839165e-06,
+      "loss": 0.1125,
+      "step": 1384
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.965741244856864e-06,
+      "loss": 0.1083,
+      "step": 1385
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.953019192154344e-06,
+      "loss": 0.1226,
+      "step": 1386
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.940298852547753e-06,
+      "loss": 0.1212,
+      "step": 1387
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.927580246850418e-06,
+      "loss": 0.1281,
+      "step": 1388
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.914863395872844e-06,
+      "loss": 0.118,
+      "step": 1389
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.90214832042265e-06,
+      "loss": 0.1238,
+      "step": 1390
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.889435041304565e-06,
+      "loss": 0.0967,
+      "step": 1391
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.876723579320363e-06,
+      "loss": 0.1374,
+      "step": 1392
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.86401395526886e-06,
+      "loss": 0.1206,
+      "step": 1393
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.85130618994585e-06,
+      "loss": 0.1271,
+      "step": 1394
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.838600304144102e-06,
+      "loss": 0.1315,
+      "step": 1395
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.825896318653294e-06,
+      "loss": 0.1094,
+      "step": 1396
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.813194254260006e-06,
+      "loss": 0.0943,
+      "step": 1397
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.800494131747667e-06,
+      "loss": 0.1308,
+      "step": 1398
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.787795971896536e-06,
+      "loss": 0.1221,
+      "step": 1399
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.775099795483651e-06,
+      "loss": 0.1048,
+      "step": 1400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.762405623282817e-06,
+      "loss": 0.0948,
+      "step": 1401
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.749713476064547e-06,
+      "loss": 0.1193,
+      "step": 1402
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.737023374596051e-06,
+      "loss": 0.1216,
+      "step": 1403
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.724335339641185e-06,
+      "loss": 0.0941,
+      "step": 1404
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.711649391960424e-06,
+      "loss": 0.1353,
+      "step": 1405
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.698965552310834e-06,
+      "loss": 0.1336,
+      "step": 1406
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.686283841446027e-06,
+      "loss": 0.0952,
+      "step": 1407
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.673604280116127e-06,
+      "loss": 0.1077,
+      "step": 1408
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.660926889067753e-06,
+      "loss": 0.1183,
+      "step": 1409
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.648251689043961e-06,
+      "loss": 0.1169,
+      "step": 1410
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.635578700784232e-06,
+      "loss": 0.1028,
+      "step": 1411
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.622907945024418e-06,
+      "loss": 0.0891,
+      "step": 1412
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.61023944249673e-06,
+      "loss": 0.1246,
+      "step": 1413
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.597573213929677e-06,
+      "loss": 0.1144,
+      "step": 1414
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.584909280048064e-06,
+      "loss": 0.1301,
+      "step": 1415
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.572247661572926e-06,
+      "loss": 0.1291,
+      "step": 1416
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.559588379221525e-06,
+      "loss": 0.1057,
+      "step": 1417
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.546931453707285e-06,
+      "loss": 0.1033,
+      "step": 1418
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.534276905739783e-06,
+      "loss": 0.1374,
+      "step": 1419
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.521624756024706e-06,
+      "loss": 0.1385,
+      "step": 1420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.508975025263814e-06,
+      "loss": 0.1025,
+      "step": 1421
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.496327734154905e-06,
+      "loss": 0.1136,
+      "step": 1422
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.483682903391796e-06,
+      "loss": 0.11,
+      "step": 1423
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.471040553664269e-06,
+      "loss": 0.1225,
+      "step": 1424
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.458400705658051e-06,
+      "loss": 0.1168,
+      "step": 1425
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.445763380054773e-06,
+      "loss": 0.1194,
+      "step": 1426
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.433128597531943e-06,
+      "loss": 0.0973,
+      "step": 1427
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.420496378762901e-06,
+      "loss": 0.1379,
+      "step": 1428
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.407866744416801e-06,
+      "loss": 0.1208,
+      "step": 1429
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.395239715158558e-06,
+      "loss": 0.0772,
+      "step": 1430
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.382615311648833e-06,
+      "loss": 0.1213,
+      "step": 1431
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.369993554543987e-06,
+      "loss": 0.0975,
+      "step": 1432
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.357374464496056e-06,
+      "loss": 0.1396,
+      "step": 1433
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.344758062152696e-06,
+      "loss": 0.1149,
+      "step": 1434
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.332144368157192e-06,
+      "loss": 0.1043,
+      "step": 1435
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.319533403148368e-06,
+      "loss": 0.0946,
+      "step": 1436
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.306925187760608e-06,
+      "loss": 0.1142,
+      "step": 1437
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.29431974262378e-06,
+      "loss": 0.1141,
+      "step": 1438
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.28171708836323e-06,
+      "loss": 0.1318,
+      "step": 1439
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.269117245599729e-06,
+      "loss": 0.1111,
+      "step": 1440
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.256520234949456e-06,
+      "loss": 0.1019,
+      "step": 1441
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.243926077023945e-06,
+      "loss": 0.1022,
+      "step": 1442
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.23133479243008e-06,
+      "loss": 0.1031,
+      "step": 1443
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.218746401770021e-06,
+      "loss": 0.0907,
+      "step": 1444
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.206160925641211e-06,
+      "loss": 0.1296,
+      "step": 1445
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.193578384636317e-06,
+      "loss": 0.1234,
+      "step": 1446
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.180998799343203e-06,
+      "loss": 0.1246,
+      "step": 1447
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.168422190344896e-06,
+      "loss": 0.121,
+      "step": 1448
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.155848578219563e-06,
+      "loss": 0.1431,
+      "step": 1449
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.14327798354045e-06,
+      "loss": 0.1319,
+      "step": 1450
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.130710426875881e-06,
+      "loss": 0.1188,
+      "step": 1451
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.118145928789198e-06,
+      "loss": 0.1265,
+      "step": 1452
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.105584509838754e-06,
+      "loss": 0.0994,
+      "step": 1453
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.093026190577839e-06,
+      "loss": 0.1344,
+      "step": 1454
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.080470991554703e-06,
+      "loss": 0.1015,
+      "step": 1455
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.067918933312459e-06,
+      "loss": 0.1094,
+      "step": 1456
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.055370036389105e-06,
+      "loss": 0.0986,
+      "step": 1457
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.042824321317453e-06,
+      "loss": 0.1158,
+      "step": 1458
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.030281808625114e-06,
+      "loss": 0.1063,
+      "step": 1459
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.017742518834454e-06,
+      "loss": 0.0912,
+      "step": 1460
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.005206472462576e-06,
+      "loss": 0.0928,
+      "step": 1461
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.99267369002126e-06,
+      "loss": 0.1161,
+      "step": 1462
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.980144192016967e-06,
+      "loss": 0.1266,
+      "step": 1463
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.967617998950762e-06,
+      "loss": 0.1181,
+      "step": 1464
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.955095131318319e-06,
+      "loss": 0.1492,
+      "step": 1465
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.942575609609857e-06,
+      "loss": 0.1143,
+      "step": 1466
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.930059454310138e-06,
+      "loss": 0.1045,
+      "step": 1467
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.917546685898393e-06,
+      "loss": 0.0968,
+      "step": 1468
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.905037324848334e-06,
+      "loss": 0.1104,
+      "step": 1469
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.89253139162808e-06,
+      "loss": 0.1227,
+      "step": 1470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.880028906700153e-06,
+      "loss": 0.1268,
+      "step": 1471
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.867529890521424e-06,
+      "loss": 0.114,
+      "step": 1472
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.8550343635431e-06,
+      "loss": 0.113,
+      "step": 1473
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.842542346210663e-06,
+      "loss": 0.1047,
+      "step": 1474
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.83005385896387e-06,
+      "loss": 0.099,
+      "step": 1475
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.817568922236683e-06,
+      "loss": 0.1469,
+      "step": 1476
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.805087556457275e-06,
+      "loss": 0.1182,
+      "step": 1477
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.792609782047958e-06,
+      "loss": 0.1171,
+      "step": 1478
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.78013561942518e-06,
+      "loss": 0.1067,
+      "step": 1479
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.76766508899947e-06,
+      "loss": 0.1714,
+      "step": 1480
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.755198211175428e-06,
+      "loss": 0.1167,
+      "step": 1481
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.742735006351656e-06,
+      "loss": 0.1134,
+      "step": 1482
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.73027549492077e-06,
+      "loss": 0.1042,
+      "step": 1483
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.717819697269322e-06,
+      "loss": 0.1207,
+      "step": 1484
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.7053676337778e-06,
+      "loss": 0.1107,
+      "step": 1485
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.69291932482058e-06,
+      "loss": 0.1024,
+      "step": 1486
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.680474790765895e-06,
+      "loss": 0.1205,
+      "step": 1487
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.668034051975793e-06,
+      "loss": 0.1429,
+      "step": 1488
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.655597128806125e-06,
+      "loss": 0.1113,
+      "step": 1489
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.643164041606489e-06,
+      "loss": 0.1049,
+      "step": 1490
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.630734810720212e-06,
+      "loss": 0.113,
+      "step": 1491
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.618309456484309e-06,
+      "loss": 0.119,
+      "step": 1492
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.605887999229454e-06,
+      "loss": 0.0907,
+      "step": 1493
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.593470459279939e-06,
+      "loss": 0.1008,
+      "step": 1494
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.581056856953656e-06,
+      "loss": 0.0959,
+      "step": 1495
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.568647212562043e-06,
+      "loss": 0.0927,
+      "step": 1496
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.55624154641007e-06,
+      "loss": 0.1263,
+      "step": 1497
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.543839878796195e-06,
+      "loss": 0.1307,
+      "step": 1498
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.531442230012336e-06,
+      "loss": 0.0914,
+      "step": 1499
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.519048620343825e-06,
+      "loss": 0.1039,
+      "step": 1500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.506659070069404e-06,
+      "loss": 0.1297,
+      "step": 1501
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.494273599461153e-06,
+      "loss": 0.1018,
+      "step": 1502
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.481892228784491e-06,
+      "loss": 0.1352,
+      "step": 1503
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.469514978298119e-06,
+      "loss": 0.1014,
+      "step": 1504
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.457141868254007e-06,
+      "loss": 0.1374,
+      "step": 1505
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.444772918897336e-06,
+      "loss": 0.1168,
+      "step": 1506
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.432408150466497e-06,
+      "loss": 0.0963,
+      "step": 1507
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.42004758319302e-06,
+      "loss": 0.0786,
+      "step": 1508
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.40769123730158e-06,
+      "loss": 0.1115,
+      "step": 1509
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.395339133009931e-06,
+      "loss": 0.1039,
+      "step": 1510
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.382991290528892e-06,
+      "loss": 0.0908,
+      "step": 1511
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.370647730062311e-06,
+      "loss": 0.1023,
+      "step": 1512
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.358308471807028e-06,
+      "loss": 0.1085,
+      "step": 1513
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3459735359528366e-06,
+      "loss": 0.1108,
+      "step": 1514
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.333642942682473e-06,
+      "loss": 0.112,
+      "step": 1515
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3213167121715514e-06,
+      "loss": 0.1291,
+      "step": 1516
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.308994864588562e-06,
+      "loss": 0.0679,
+      "step": 1517
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.296677420094811e-06,
+      "loss": 0.1075,
+      "step": 1518
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.284364398844412e-06,
+      "loss": 0.0886,
+      "step": 1519
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.27205582098423e-06,
+      "loss": 0.098,
+      "step": 1520
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.25975170665387e-06,
+      "loss": 0.1066,
+      "step": 1521
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.247452075985622e-06,
+      "loss": 0.1211,
+      "step": 1522
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.235156949104455e-06,
+      "loss": 0.1437,
+      "step": 1523
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.222866346127952e-06,
+      "loss": 0.1037,
+      "step": 1524
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.210580287166307e-06,
+      "loss": 0.1028,
+      "step": 1525
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.198298792322271e-06,
+      "loss": 0.1126,
+      "step": 1526
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.186021881691132e-06,
+      "loss": 0.1193,
+      "step": 1527
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.173749575360671e-06,
+      "loss": 0.1026,
+      "step": 1528
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1614818934111475e-06,
+      "loss": 0.0995,
+      "step": 1529
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1492188559152364e-06,
+      "loss": 0.0954,
+      "step": 1530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.136960482938035e-06,
+      "loss": 0.1101,
+      "step": 1531
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.124706794536984e-06,
+      "loss": 0.0863,
+      "step": 1532
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.112457810761883e-06,
+      "loss": 0.1286,
+      "step": 1533
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.100213551654816e-06,
+      "loss": 0.1199,
+      "step": 1534
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.087974037250146e-06,
+      "loss": 0.0854,
+      "step": 1535
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.075739287574467e-06,
+      "loss": 0.1062,
+      "step": 1536
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.063509322646581e-06,
+      "loss": 0.1118,
+      "step": 1537
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.051284162477459e-06,
+      "loss": 0.0999,
+      "step": 1538
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.039063827070214e-06,
+      "loss": 0.1337,
+      "step": 1539
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.026848336420053e-06,
+      "loss": 0.1162,
+      "step": 1540
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.014637710514274e-06,
+      "loss": 0.1248,
+      "step": 1541
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.002431969332197e-06,
+      "loss": 0.1089,
+      "step": 1542
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.990231132845169e-06,
+      "loss": 0.1096,
+      "step": 1543
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.978035221016487e-06,
+      "loss": 0.1177,
+      "step": 1544
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.965844253801416e-06,
+      "loss": 0.1243,
+      "step": 1545
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.953658251147109e-06,
+      "loss": 0.1437,
+      "step": 1546
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.941477232992614e-06,
+      "loss": 0.1254,
+      "step": 1547
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.929301219268806e-06,
+      "loss": 0.1088,
+      "step": 1548
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.917130229898387e-06,
+      "loss": 0.1089,
+      "step": 1549
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.90496428479583e-06,
+      "loss": 0.1133,
+      "step": 1550
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.892803403867352e-06,
+      "loss": 0.1484,
+      "step": 1551
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8806476070108905e-06,
+      "loss": 0.1118,
+      "step": 1552
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.868496914116063e-06,
+      "loss": 0.0985,
+      "step": 1553
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.856351345064127e-06,
+      "loss": 0.1158,
+      "step": 1554
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.844210919727971e-06,
+      "loss": 0.094,
+      "step": 1555
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8320756579720545e-06,
+      "loss": 0.103,
+      "step": 1556
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.819945579652401e-06,
+      "loss": 0.1048,
+      "step": 1557
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.807820704616532e-06,
+      "loss": 0.1028,
+      "step": 1558
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.795701052703482e-06,
+      "loss": 0.1279,
+      "step": 1559
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.783586643743714e-06,
+      "loss": 0.1251,
+      "step": 1560
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7714774975591335e-06,
+      "loss": 0.0842,
+      "step": 1561
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.75937363396302e-06,
+      "loss": 0.1297,
+      "step": 1562
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7472750727600155e-06,
+      "loss": 0.1036,
+      "step": 1563
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.735181833746087e-06,
+      "loss": 0.0971,
+      "step": 1564
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.7230939367084915e-06,
+      "loss": 0.1054,
+      "step": 1565
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.711011401425741e-06,
+      "loss": 0.1098,
+      "step": 1566
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.698934247667587e-06,
+      "loss": 0.1138,
+      "step": 1567
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.686862495194958e-06,
+      "loss": 0.1178,
+      "step": 1568
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6747961637599645e-06,
+      "loss": 0.1363,
+      "step": 1569
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.662735273105827e-06,
+      "loss": 0.1278,
+      "step": 1570
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.650679842966881e-06,
+      "loss": 0.0872,
+      "step": 1571
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.638629893068516e-06,
+      "loss": 0.1117,
+      "step": 1572
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.62658544312716e-06,
+      "loss": 0.1217,
+      "step": 1573
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.614546512850237e-06,
+      "loss": 0.1036,
+      "step": 1574
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6025131219361505e-06,
+      "loss": 0.102,
+      "step": 1575
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.590485290074224e-06,
+      "loss": 0.1044,
+      "step": 1576
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5784630369447e-06,
+      "loss": 0.1024,
+      "step": 1577
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.566446382218683e-06,
+      "loss": 0.1012,
+      "step": 1578
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5544353455581245e-06,
+      "loss": 0.1415,
+      "step": 1579
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.542429946615774e-06,
+      "loss": 0.1237,
+      "step": 1580
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.53043020503517e-06,
+      "loss": 0.1089,
+      "step": 1581
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5184361404505795e-06,
+      "loss": 0.127,
+      "step": 1582
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.506447772486997e-06,
+      "loss": 0.1379,
+      "step": 1583
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4944651207600765e-06,
+      "loss": 0.1214,
+      "step": 1584
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4824882048761406e-06,
+      "loss": 0.1317,
+      "step": 1585
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.470517044432104e-06,
+      "loss": 0.107,
+      "step": 1586
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.458551659015486e-06,
+      "loss": 0.0815,
+      "step": 1587
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.446592068204341e-06,
+      "loss": 0.1072,
+      "step": 1588
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.43463829156725e-06,
+      "loss": 0.1001,
+      "step": 1589
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.422690348663276e-06,
+      "loss": 0.0909,
+      "step": 1590
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.410748259041941e-06,
+      "loss": 0.0905,
+      "step": 1591
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.398812042243187e-06,
+      "loss": 0.1001,
+      "step": 1592
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3868817177973505e-06,
+      "loss": 0.1324,
+      "step": 1593
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3749573052251155e-06,
+      "loss": 0.076,
+      "step": 1594
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.363038824037511e-06,
+      "loss": 0.1181,
+      "step": 1595
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.351126293735843e-06,
+      "loss": 0.1502,
+      "step": 1596
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.339219733811697e-06,
+      "loss": 0.1324,
+      "step": 1597
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.327319163746871e-06,
+      "loss": 0.1182,
+      "step": 1598
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.315424603013382e-06,
+      "loss": 0.1083,
+      "step": 1599
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.303536071073397e-06,
+      "loss": 0.1491,
+      "step": 1600
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.291653587379236e-06,
+      "loss": 0.1053,
+      "step": 1601
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2797771713733025e-06,
+      "loss": 0.0877,
+      "step": 1602
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.267906842488088e-06,
+      "loss": 0.1218,
+      "step": 1603
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.256042620146119e-06,
+      "loss": 0.1157,
+      "step": 1604
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2441845237599285e-06,
+      "loss": 0.0705,
+      "step": 1605
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.232332572732025e-06,
+      "loss": 0.0987,
+      "step": 1606
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.22048678645487e-06,
+      "loss": 0.1096,
+      "step": 1607
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.208647184310826e-06,
+      "loss": 0.1108,
+      "step": 1608
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.196813785672149e-06,
+      "loss": 0.1364,
+      "step": 1609
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.184986609900934e-06,
+      "loss": 0.1254,
+      "step": 1610
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.1091,
+      "step": 1611
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.16135100435836e-06,
+      "loss": 0.1031,
+      "step": 1612
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.149542613260157e-06,
+      "loss": 0.1045,
+      "step": 1613
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.137740522375687e-06,
+      "loss": 0.0953,
+      "step": 1614
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.1259447510158136e-06,
+      "loss": 0.0791,
+      "step": 1615
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.114155318481076e-06,
+      "loss": 0.0871,
+      "step": 1616
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.102372244061631e-06,
+      "loss": 0.113,
+      "step": 1617
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.090595547037242e-06,
+      "loss": 0.0952,
+      "step": 1618
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.078825246677229e-06,
+      "loss": 0.0994,
+      "step": 1619
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.06706136224045e-06,
+      "loss": 0.121,
+      "step": 1620
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.055303912975261e-06,
+      "loss": 0.0863,
+      "step": 1621
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0435529181195e-06,
+      "loss": 0.1151,
+      "step": 1622
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.031808396900422e-06,
+      "loss": 0.1118,
+      "step": 1623
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.020070368534719e-06,
+      "loss": 0.1064,
+      "step": 1624
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.00833885222843e-06,
+      "loss": 0.1304,
+      "step": 1625
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.996613867176964e-06,
+      "loss": 0.1031,
+      "step": 1626
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.984895432565022e-06,
+      "loss": 0.1021,
+      "step": 1627
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.973183567566605e-06,
+      "loss": 0.1032,
+      "step": 1628
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.96147829134495e-06,
+      "loss": 0.0908,
+      "step": 1629
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.949779623052526e-06,
+      "loss": 0.1146,
+      "step": 1630
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9380875818309805e-06,
+      "loss": 0.0962,
+      "step": 1631
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.926402186811118e-06,
+      "loss": 0.1371,
+      "step": 1632
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.914723457112877e-06,
+      "loss": 0.1128,
+      "step": 1633
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.903051411845282e-06,
+      "loss": 0.1186,
+      "step": 1634
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8913860701064175e-06,
+      "loss": 0.1058,
+      "step": 1635
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.879727450983412e-06,
+      "loss": 0.1199,
+      "step": 1636
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.868075573552383e-06,
+      "loss": 0.1133,
+      "step": 1637
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.856430456878424e-06,
+      "loss": 0.1246,
+      "step": 1638
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.844792120015556e-06,
+      "loss": 0.075,
+      "step": 1639
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.833160582006722e-06,
+      "loss": 0.0945,
+      "step": 1640
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.821535861883729e-06,
+      "loss": 0.0992,
+      "step": 1641
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8099179786672365e-06,
+      "loss": 0.1475,
+      "step": 1642
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.798306951366701e-06,
+      "loss": 0.1541,
+      "step": 1643
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.786702798980388e-06,
+      "loss": 0.1134,
+      "step": 1644
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.775105540495284e-06,
+      "loss": 0.1214,
+      "step": 1645
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.763515194887126e-06,
+      "loss": 0.1316,
+      "step": 1646
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.751931781120308e-06,
+      "loss": 0.0929,
+      "step": 1647
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.740355318147916e-06,
+      "loss": 0.1221,
+      "step": 1648
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.728785824911627e-06,
+      "loss": 0.0956,
+      "step": 1649
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.717223320341751e-06,
+      "loss": 0.1013,
+      "step": 1650
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.70566782335713e-06,
+      "loss": 0.1545,
+      "step": 1651
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.69411935286516e-06,
+      "loss": 0.1329,
+      "step": 1652
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.682577927761732e-06,
+      "loss": 0.0914,
+      "step": 1653
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.671043566931216e-06,
+      "loss": 0.0976,
+      "step": 1654
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.659516289246414e-06,
+      "loss": 0.1177,
+      "step": 1655
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.647996113568547e-06,
+      "loss": 0.1068,
+      "step": 1656
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.636483058747209e-06,
+      "loss": 0.0964,
+      "step": 1657
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.624977143620347e-06,
+      "loss": 0.1194,
+      "step": 1658
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.613478387014223e-06,
+      "loss": 0.0936,
+      "step": 1659
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.601986807743388e-06,
+      "loss": 0.1042,
+      "step": 1660
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5905024246106485e-06,
+      "loss": 0.1586,
+      "step": 1661
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.579025256407038e-06,
+      "loss": 0.086,
+      "step": 1662
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.567555321911782e-06,
+      "loss": 0.0759,
+      "step": 1663
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.55609263989227e-06,
+      "loss": 0.1117,
+      "step": 1664
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.544637229104027e-06,
+      "loss": 0.1033,
+      "step": 1665
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.533189108290682e-06,
+      "loss": 0.1112,
+      "step": 1666
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5217482961839305e-06,
+      "loss": 0.0974,
+      "step": 1667
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.51031481150352e-06,
+      "loss": 0.1115,
+      "step": 1668
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.498888672957187e-06,
+      "loss": 0.0941,
+      "step": 1669
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.487469899240678e-06,
+      "loss": 0.1419,
+      "step": 1670
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.476058509037658e-06,
+      "loss": 0.09,
+      "step": 1671
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4646545210197435e-06,
+      "loss": 0.1115,
+      "step": 1672
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.453257953846405e-06,
+      "loss": 0.1057,
+      "step": 1673
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.441868826165002e-06,
+      "loss": 0.1156,
+      "step": 1674
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.430487156610695e-06,
+      "loss": 0.1216,
+      "step": 1675
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.419112963806468e-06,
+      "loss": 0.1304,
+      "step": 1676
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.407746266363039e-06,
+      "loss": 0.1052,
+      "step": 1677
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3963870828788975e-06,
+      "loss": 0.0969,
+      "step": 1678
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3850354319402095e-06,
+      "loss": 0.109,
+      "step": 1679
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.373691332120832e-06,
+      "loss": 0.1193,
+      "step": 1680
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.362354801982259e-06,
+      "loss": 0.1284,
+      "step": 1681
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.351025860073604e-06,
+      "loss": 0.0956,
+      "step": 1682
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3397045249315615e-06,
+      "loss": 0.1294,
+      "step": 1683
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.328390815080381e-06,
+      "loss": 0.1096,
+      "step": 1684
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.317084749031835e-06,
+      "loss": 0.0992,
+      "step": 1685
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3057863452851875e-06,
+      "loss": 0.0826,
+      "step": 1686
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.294495622327167e-06,
+      "loss": 0.1263,
+      "step": 1687
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.283212598631935e-06,
+      "loss": 0.1362,
+      "step": 1688
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.271937292661054e-06,
+      "loss": 0.1096,
+      "step": 1689
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.260669722863457e-06,
+      "loss": 0.0997,
+      "step": 1690
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.249409907675422e-06,
+      "loss": 0.0926,
+      "step": 1691
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.238157865520539e-06,
+      "loss": 0.1166,
+      "step": 1692
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.226913614809677e-06,
+      "loss": 0.1146,
+      "step": 1693
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.215677173940959e-06,
+      "loss": 0.1338,
+      "step": 1694
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.204448561299718e-06,
+      "loss": 0.1144,
+      "step": 1695
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.193227795258505e-06,
+      "loss": 0.0982,
+      "step": 1696
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.182014894176999e-06,
+      "loss": 0.1004,
+      "step": 1697
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.170809876402039e-06,
+      "loss": 0.0932,
+      "step": 1698
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.159612760267541e-06,
+      "loss": 0.1063,
+      "step": 1699
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.148423564094517e-06,
+      "loss": 0.1205,
+      "step": 1700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.137242306190991e-06,
+      "loss": 0.1042,
+      "step": 1701
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.126069004852033e-06,
+      "loss": 0.1116,
+      "step": 1702
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.114903678359655e-06,
+      "loss": 0.121,
+      "step": 1703
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.103746344982859e-06,
+      "loss": 0.1357,
+      "step": 1704
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.092597022977539e-06,
+      "loss": 0.1099,
+      "step": 1705
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.081455730586495e-06,
+      "loss": 0.1085,
+      "step": 1706
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.070322486039383e-06,
+      "loss": 0.1445,
+      "step": 1707
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.059197307552698e-06,
+      "loss": 0.0771,
+      "step": 1708
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.048080213329729e-06,
+      "loss": 0.0956,
+      "step": 1709
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.036971221560543e-06,
+      "loss": 0.0966,
+      "step": 1710
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.025870350421945e-06,
+      "loss": 0.1124,
+      "step": 1711
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0147776180774575e-06,
+      "loss": 0.142,
+      "step": 1712
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.003693042677283e-06,
+      "loss": 0.0967,
+      "step": 1713
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.992616642358279e-06,
+      "loss": 0.1094,
+      "step": 1714
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9815484352439244e-06,
+      "loss": 0.1332,
+      "step": 1715
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.970488439444296e-06,
+      "loss": 0.1089,
+      "step": 1716
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.95943667305603e-06,
+      "loss": 0.0885,
+      "step": 1717
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.948393154162303e-06,
+      "loss": 0.0999,
+      "step": 1718
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.937357900832793e-06,
+      "loss": 0.1002,
+      "step": 1719
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.926330931123659e-06,
+      "loss": 0.1047,
+      "step": 1720
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.915312263077488e-06,
+      "loss": 0.1538,
+      "step": 1721
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.904301914723315e-06,
+      "loss": 0.1362,
+      "step": 1722
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.89329990407653e-06,
+      "loss": 0.1118,
+      "step": 1723
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.882306249138909e-06,
+      "loss": 0.0806,
+      "step": 1724
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.871320967898528e-06,
+      "loss": 0.1385,
+      "step": 1725
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.860344078329791e-06,
+      "loss": 0.1173,
+      "step": 1726
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.849375598393342e-06,
+      "loss": 0.1229,
+      "step": 1727
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.838415546036095e-06,
+      "loss": 0.108,
+      "step": 1728
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.827463939191141e-06,
+      "loss": 0.095,
+      "step": 1729
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.816520795777789e-06,
+      "loss": 0.0899,
+      "step": 1730
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.805586133701468e-06,
+      "loss": 0.0863,
+      "step": 1731
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7946599708537485e-06,
+      "loss": 0.1126,
+      "step": 1732
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.783742325112286e-06,
+      "loss": 0.0931,
+      "step": 1733
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.772833214340805e-06,
+      "loss": 0.1059,
+      "step": 1734
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.761932656389061e-06,
+      "loss": 0.1187,
+      "step": 1735
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.751040669092819e-06,
+      "loss": 0.1284,
+      "step": 1736
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.740157270273816e-06,
+      "loss": 0.0901,
+      "step": 1737
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.729282477739741e-06,
+      "loss": 0.1147,
+      "step": 1738
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.718416309284196e-06,
+      "loss": 0.0965,
+      "step": 1739
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.707558782686677e-06,
+      "loss": 0.1096,
+      "step": 1740
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6967099157125384e-06,
+      "loss": 0.1155,
+      "step": 1741
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.685869726112963e-06,
+      "loss": 0.104,
+      "step": 1742
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.675038231624939e-06,
+      "loss": 0.0837,
+      "step": 1743
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.664215449971225e-06,
+      "loss": 0.1276,
+      "step": 1744
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.653401398860324e-06,
+      "loss": 0.0988,
+      "step": 1745
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6425960959864556e-06,
+      "loss": 0.1089,
+      "step": 1746
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.631799559029524e-06,
+      "loss": 0.1105,
+      "step": 1747
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.621011805655093e-06,
+      "loss": 0.1199,
+      "step": 1748
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.61023285351434e-06,
+      "loss": 0.0996,
+      "step": 1749
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.599462720244071e-06,
+      "loss": 0.0936,
+      "step": 1750
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5887014234666275e-06,
+      "loss": 0.1174,
+      "step": 1751
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.577948980789924e-06,
+      "loss": 0.0956,
+      "step": 1752
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.567205409807362e-06,
+      "loss": 0.0928,
+      "step": 1753
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.556470728097849e-06,
+      "loss": 0.1124,
+      "step": 1754
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.545744953225726e-06,
+      "loss": 0.1182,
+      "step": 1755
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.535028102740785e-06,
+      "loss": 0.1145,
+      "step": 1756
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.524320194178189e-06,
+      "loss": 0.0852,
+      "step": 1757
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5136212450584895e-06,
+      "loss": 0.0776,
+      "step": 1758
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.502931272887572e-06,
+      "loss": 0.0721,
+      "step": 1759
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.492250295156632e-06,
+      "loss": 0.106,
+      "step": 1760
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.481578329342149e-06,
+      "loss": 0.1134,
+      "step": 1761
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.470915392905862e-06,
+      "loss": 0.0875,
+      "step": 1762
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.460261503294726e-06,
+      "loss": 0.1205,
+      "step": 1763
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.449616677940904e-06,
+      "loss": 0.128,
+      "step": 1764
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4389809342617195e-06,
+      "loss": 0.1334,
+      "step": 1765
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.428354289659641e-06,
+      "loss": 0.1457,
+      "step": 1766
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.417736761522249e-06,
+      "loss": 0.108,
+      "step": 1767
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.407128367222203e-06,
+      "loss": 0.0988,
+      "step": 1768
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.396529124117223e-06,
+      "loss": 0.0855,
+      "step": 1769
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.38593904955005e-06,
+      "loss": 0.0679,
+      "step": 1770
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.37535816084843e-06,
+      "loss": 0.105,
+      "step": 1771
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.364786475325072e-06,
+      "loss": 0.1068,
+      "step": 1772
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.354224010277632e-06,
+      "loss": 0.1066,
+      "step": 1773
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.343670782988679e-06,
+      "loss": 0.0895,
+      "step": 1774
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.333126810725655e-06,
+      "loss": 0.085,
+      "step": 1775
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.322592110740882e-06,
+      "loss": 0.0972,
+      "step": 1776
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.312066700271483e-06,
+      "loss": 0.1205,
+      "step": 1777
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.30155059653941e-06,
+      "loss": 0.1224,
+      "step": 1778
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.291043816751357e-06,
+      "loss": 0.0989,
+      "step": 1779
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.280546378098792e-06,
+      "loss": 0.1213,
+      "step": 1780
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.270058297757871e-06,
+      "loss": 0.0892,
+      "step": 1781
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.259579592889464e-06,
+      "loss": 0.1166,
+      "step": 1782
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.249110280639076e-06,
+      "loss": 0.1055,
+      "step": 1783
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.238650378136859e-06,
+      "loss": 0.1145,
+      "step": 1784
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.228199902497565e-06,
+      "loss": 0.1312,
+      "step": 1785
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.217758870820522e-06,
+      "loss": 0.1145,
+      "step": 1786
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.207327300189602e-06,
+      "loss": 0.1071,
+      "step": 1787
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.196905207673201e-06,
+      "loss": 0.1199,
+      "step": 1788
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.186492610324204e-06,
+      "loss": 0.0898,
+      "step": 1789
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.176089525179961e-06,
+      "loss": 0.1044,
+      "step": 1790
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.165695969262259e-06,
+      "loss": 0.1165,
+      "step": 1791
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.155311959577289e-06,
+      "loss": 0.101,
+      "step": 1792
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.144937513115627e-06,
+      "loss": 0.108,
+      "step": 1793
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.134572646852196e-06,
+      "loss": 0.0797,
+      "step": 1794
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.124217377746251e-06,
+      "loss": 0.1042,
+      "step": 1795
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.113871722741337e-06,
+      "loss": 0.1257,
+      "step": 1796
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.103535698765272e-06,
+      "loss": 0.0837,
+      "step": 1797
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.093209322730114e-06,
+      "loss": 0.1339,
+      "step": 1798
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.082892611532136e-06,
+      "loss": 0.1001,
+      "step": 1799
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.072585582051798e-06,
+      "loss": 0.0933,
+      "step": 1800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0622882511537076e-06,
+      "loss": 0.1021,
+      "step": 1801
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.052000635686627e-06,
+      "loss": 0.1205,
+      "step": 1802
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0417227524833925e-06,
+      "loss": 0.1362,
+      "step": 1803
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.031454618360945e-06,
+      "loss": 0.1127,
+      "step": 1804
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.021196250120248e-06,
+      "loss": 0.1093,
+      "step": 1805
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.0109476645463076e-06,
+      "loss": 0.0836,
+      "step": 1806
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.000708878408103e-06,
+      "loss": 0.091,
+      "step": 1807
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.990479908458602e-06,
+      "loss": 0.1052,
+      "step": 1808
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.980260771434685e-06,
+      "loss": 0.1153,
+      "step": 1809
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.970051484057171e-06,
+      "loss": 0.0851,
+      "step": 1810
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.959852063030738e-06,
+      "loss": 0.0973,
+      "step": 1811
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.949662525043935e-06,
+      "loss": 0.1131,
+      "step": 1812
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.939482886769136e-06,
+      "loss": 0.1079,
+      "step": 1813
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.929313164862518e-06,
+      "loss": 0.1047,
+      "step": 1814
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.919153375964032e-06,
+      "loss": 0.0898,
+      "step": 1815
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.909003536697374e-06,
+      "loss": 0.1052,
+      "step": 1816
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.898863663669965e-06,
+      "loss": 0.0675,
+      "step": 1817
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.888733773472916e-06,
+      "loss": 0.0986,
+      "step": 1818
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.878613882681002e-06,
+      "loss": 0.1084,
+      "step": 1819
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.868504007852641e-06,
+      "loss": 0.126,
+      "step": 1820
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8584041655298606e-06,
+      "loss": 0.0946,
+      "step": 1821
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.848314372238272e-06,
+      "loss": 0.1079,
+      "step": 1822
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.838234644487045e-06,
+      "loss": 0.1122,
+      "step": 1823
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.828164998768879e-06,
+      "loss": 0.0923,
+      "step": 1824
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8181054515599806e-06,
+      "loss": 0.1212,
+      "step": 1825
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8080560193200288e-06,
+      "loss": 0.1066,
+      "step": 1826
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.798016718492148e-06,
+      "loss": 0.1168,
+      "step": 1827
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7879875655029018e-06,
+      "loss": 0.1356,
+      "step": 1828
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7779685767622255e-06,
+      "loss": 0.0709,
+      "step": 1829
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7679597686634495e-06,
+      "loss": 0.0979,
+      "step": 1830
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.757961157583221e-06,
+      "loss": 0.1018,
+      "step": 1831
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7479727598815287e-06,
+      "loss": 0.0893,
+      "step": 1832
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7379945919016225e-06,
+      "loss": 0.1252,
+      "step": 1833
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7280266699700406e-06,
+      "loss": 0.0769,
+      "step": 1834
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7180690103965313e-06,
+      "loss": 0.1218,
+      "step": 1835
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7081216294740773e-06,
+      "loss": 0.1113,
+      "step": 1836
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6981845434788188e-06,
+      "loss": 0.1016,
+      "step": 1837
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.688257768670065e-06,
+      "loss": 0.0956,
+      "step": 1838
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.678341321290252e-06,
+      "loss": 0.117,
+      "step": 1839
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.668435217564915e-06,
+      "loss": 0.116,
+      "step": 1840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.658539473702667e-06,
+      "loss": 0.1169,
+      "step": 1841
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6486541058951696e-06,
+      "loss": 0.1091,
+      "step": 1842
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.638779130317106e-06,
+      "loss": 0.1016,
+      "step": 1843
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.628914563126156e-06,
+      "loss": 0.0984,
+      "step": 1844
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6190604204629685e-06,
+      "loss": 0.1249,
+      "step": 1845
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6092167184511352e-06,
+      "loss": 0.0934,
+      "step": 1846
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5993834731971654e-06,
+      "loss": 0.0938,
+      "step": 1847
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5895607007904597e-06,
+      "loss": 0.1131,
+      "step": 1848
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5797484173032806e-06,
+      "loss": 0.1056,
+      "step": 1849
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.569946638790729e-06,
+      "loss": 0.1337,
+      "step": 1850
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5601553812907174e-06,
+      "loss": 0.1021,
+      "step": 1851
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5503746608239487e-06,
+      "loss": 0.0991,
+      "step": 1852
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5406044933938688e-06,
+      "loss": 0.1284,
+      "step": 1853
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5308448949866805e-06,
+      "loss": 0.1154,
+      "step": 1854
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5210958815712672e-06,
+      "loss": 0.084,
+      "step": 1855
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5113574690992203e-06,
+      "loss": 0.0977,
+      "step": 1856
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5016296735047584e-06,
+      "loss": 0.1054,
+      "step": 1857
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4919125107047537e-06,
+      "loss": 0.1159,
+      "step": 1858
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.482205996598654e-06,
+      "loss": 0.1116,
+      "step": 1859
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.472510147068515e-06,
+      "loss": 0.09,
+      "step": 1860
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4628249779789105e-06,
+      "loss": 0.1182,
+      "step": 1861
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4531505051769665e-06,
+      "loss": 0.1057,
+      "step": 1862
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4434867444922857e-06,
+      "loss": 0.0962,
+      "step": 1863
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.433833711736957e-06,
+      "loss": 0.1252,
+      "step": 1864
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4241914227055096e-06,
+      "loss": 0.0949,
+      "step": 1865
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.414559893174898e-06,
+      "loss": 0.0944,
+      "step": 1866
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4049391389044674e-06,
+      "loss": 0.0805,
+      "step": 1867
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3953291756359354e-06,
+      "loss": 0.1193,
+      "step": 1868
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3857300190933606e-06,
+      "loss": 0.0933,
+      "step": 1869
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.376141684983121e-06,
+      "loss": 0.1129,
+      "step": 1870
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.366564188993887e-06,
+      "loss": 0.118,
+      "step": 1871
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3569975467965955e-06,
+      "loss": 0.1418,
+      "step": 1872
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.347441774044421e-06,
+      "loss": 0.0979,
+      "step": 1873
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.337896886372757e-06,
+      "loss": 0.1007,
+      "step": 1874
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3283628993991846e-06,
+      "loss": 0.1293,
+      "step": 1875
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3188398287234504e-06,
+      "loss": 0.0911,
+      "step": 1876
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3093276899274373e-06,
+      "loss": 0.1238,
+      "step": 1877
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2998264985751425e-06,
+      "loss": 0.0944,
+      "step": 1878
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2903362702126516e-06,
+      "loss": 0.0966,
+      "step": 1879
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2808570203681135e-06,
+      "loss": 0.1076,
+      "step": 1880
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.271388764551702e-06,
+      "loss": 0.1036,
+      "step": 1881
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2619315182556234e-06,
+      "loss": 0.0905,
+      "step": 1882
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2524852969540477e-06,
+      "loss": 0.1076,
+      "step": 1883
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.243050116103128e-06,
+      "loss": 0.1134,
+      "step": 1884
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2336259911409283e-06,
+      "loss": 0.0868,
+      "step": 1885
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2242129374874478e-06,
+      "loss": 0.1237,
+      "step": 1886
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2148109705445442e-06,
+      "loss": 0.0807,
+      "step": 1887
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.205420105695963e-06,
+      "loss": 0.1403,
+      "step": 1888
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1960403583072596e-06,
+      "loss": 0.1089,
+      "step": 1889
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.186671743725812e-06,
+      "loss": 0.11,
+      "step": 1890
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1773142772807796e-06,
+      "loss": 0.1127,
+      "step": 1891
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1679679742830806e-06,
+      "loss": 0.0851,
+      "step": 1892
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.158632850025367e-06,
+      "loss": 0.1307,
+      "step": 1893
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1493089197820015e-06,
+      "loss": 0.1142,
+      "step": 1894
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.139996198809028e-06,
+      "loss": 0.1198,
+      "step": 1895
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1306947023441524e-06,
+      "loss": 0.0957,
+      "step": 1896
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.121404445606714e-06,
+      "loss": 0.0869,
+      "step": 1897
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.112125443797659e-06,
+      "loss": 0.0782,
+      "step": 1898
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1028577120995216e-06,
+      "loss": 0.0894,
+      "step": 1899
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0936012656763937e-06,
+      "loss": 0.0966,
+      "step": 1900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0843561196739013e-06,
+      "loss": 0.1087,
+      "step": 1901
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.075122289219181e-06,
+      "loss": 0.1168,
+      "step": 1902
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0658997894208573e-06,
+      "loss": 0.0904,
+      "step": 1903
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0566886353690106e-06,
+      "loss": 0.1073,
+      "step": 1904
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.047488842135159e-06,
+      "loss": 0.1021,
+      "step": 1905
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.038300424772237e-06,
+      "loss": 0.0925,
+      "step": 1906
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0291233983145494e-06,
+      "loss": 0.1018,
+      "step": 1907
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.019957777777788e-06,
+      "loss": 0.1134,
+      "step": 1908
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.010803578158954e-06,
+      "loss": 0.0972,
+      "step": 1909
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.00166081443639e-06,
+      "loss": 0.0985,
+      "step": 1910
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9925295015696978e-06,
+      "loss": 0.1194,
+      "step": 1911
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9834096544997725e-06,
+      "loss": 0.084,
+      "step": 1912
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9743012881487187e-06,
+      "loss": 0.0986,
+      "step": 1913
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.965204417419886e-06,
+      "loss": 0.0777,
+      "step": 1914
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.956119057197785e-06,
+      "loss": 0.1202,
+      "step": 1915
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9470452223481206e-06,
+      "loss": 0.1105,
+      "step": 1916
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9379829277177152e-06,
+      "loss": 0.0983,
+      "step": 1917
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.1083,
+      "step": 1918
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9198930184075944e-06,
+      "loss": 0.1047,
+      "step": 1919
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9108654333270346e-06,
+      "loss": 0.0736,
+      "step": 1920
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.901849447664008e-06,
+      "loss": 0.1058,
+      "step": 1921
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.892845076170685e-06,
+      "loss": 0.1132,
+      "step": 1922
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8838523335802525e-06,
+      "loss": 0.1205,
+      "step": 1923
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8748712346068464e-06,
+      "loss": 0.1159,
+      "step": 1924
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.865901793945576e-06,
+      "loss": 0.1326,
+      "step": 1925
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8569440262724502e-06,
+      "loss": 0.1125,
+      "step": 1926
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8479979462444017e-06,
+      "loss": 0.1086,
+      "step": 1927
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8390635684992163e-06,
+      "loss": 0.0721,
+      "step": 1928
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8301409076555574e-06,
+      "loss": 0.1097,
+      "step": 1929
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.821229978312889e-06,
+      "loss": 0.12,
+      "step": 1930
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8123307950515087e-06,
+      "loss": 0.091,
+      "step": 1931
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8034433724324716e-06,
+      "loss": 0.1118,
+      "step": 1932
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7945677249976e-06,
+      "loss": 0.0936,
+      "step": 1933
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7857038672694492e-06,
+      "loss": 0.1068,
+      "step": 1934
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.776851813751281e-06,
+      "loss": 0.0761,
+      "step": 1935
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7680115789270478e-06,
+      "loss": 0.0655,
+      "step": 1936
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7591831772613576e-06,
+      "loss": 0.0932,
+      "step": 1937
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.750366623199462e-06,
+      "loss": 0.11,
+      "step": 1938
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7415619311672236e-06,
+      "loss": 0.1219,
+      "step": 1939
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7327691155710978e-06,
+      "loss": 0.093,
+      "step": 1940
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.723988190798108e-06,
+      "loss": 0.1233,
+      "step": 1941
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7152191712158207e-06,
+      "loss": 0.1006,
+      "step": 1942
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.706462071172322e-06,
+      "loss": 0.0919,
+      "step": 1943
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.697716904996196e-06,
+      "loss": 0.1026,
+      "step": 1944
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6889836869965016e-06,
+      "loss": 0.1219,
+      "step": 1945
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6802624314627436e-06,
+      "loss": 0.1268,
+      "step": 1946
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6715531526648585e-06,
+      "loss": 0.0865,
+      "step": 1947
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6628558648531845e-06,
+      "loss": 0.0916,
+      "step": 1948
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.654170582258441e-06,
+      "loss": 0.0869,
+      "step": 1949
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.645497319091692e-06,
+      "loss": 0.0958,
+      "step": 1950
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.63683608954436e-06,
+      "loss": 0.1339,
+      "step": 1951
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6281869077881507e-06,
+      "loss": 0.1393,
+      "step": 1952
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.619549787975081e-06,
+      "loss": 0.0931,
+      "step": 1953
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6109247442374088e-06,
+      "loss": 0.1093,
+      "step": 1954
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.602311790687655e-06,
+      "loss": 0.1113,
+      "step": 1955
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.593710941418537e-06,
+      "loss": 0.1224,
+      "step": 1956
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.585122210502987e-06,
+      "loss": 0.1092,
+      "step": 1957
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5765456119940933e-06,
+      "loss": 0.1079,
+      "step": 1958
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5679811599251003e-06,
+      "loss": 0.0941,
+      "step": 1959
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.559428868309377e-06,
+      "loss": 0.1187,
+      "step": 1960
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5508887511403936e-06,
+      "loss": 0.0993,
+      "step": 1961
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.542360822391702e-06,
+      "loss": 0.0952,
+      "step": 1962
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5338450960169105e-06,
+      "loss": 0.1057,
+      "step": 1963
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.525341585949662e-06,
+      "loss": 0.1234,
+      "step": 1964
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5168503061036086e-06,
+      "loss": 0.1204,
+      "step": 1965
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5083712703723952e-06,
+      "loss": 0.1284,
+      "step": 1966
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.499904492629627e-06,
+      "loss": 0.0926,
+      "step": 1967
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4914499867288577e-06,
+      "loss": 0.1216,
+      "step": 1968
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.483007766503558e-06,
+      "loss": 0.1151,
+      "step": 1969
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.474577845767099e-06,
+      "loss": 0.1034,
+      "step": 1970
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4661602383127235e-06,
+      "loss": 0.0822,
+      "step": 1971
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4577549579135318e-06,
+      "loss": 0.1206,
+      "step": 1972
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.449362018322451e-06,
+      "loss": 0.1247,
+      "step": 1973
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.440981433272216e-06,
+      "loss": 0.1017,
+      "step": 1974
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.43261321647535e-06,
+      "loss": 0.125,
+      "step": 1975
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.42425738162413e-06,
+      "loss": 0.1208,
+      "step": 1976
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4159139423905898e-06,
+      "loss": 0.1183,
+      "step": 1977
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4075829124264606e-06,
+      "loss": 0.1036,
+      "step": 1978
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3992643053631904e-06,
+      "loss": 0.0852,
+      "step": 1979
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3909581348118803e-06,
+      "loss": 0.0932,
+      "step": 1980
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3826644143633017e-06,
+      "loss": 0.0948,
+      "step": 1981
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3743831575878352e-06,
+      "loss": 0.1076,
+      "step": 1982
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.366114378035489e-06,
+      "loss": 0.117,
+      "step": 1983
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3578580892358337e-06,
+      "loss": 0.0993,
+      "step": 1984
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3496143046980256e-06,
+      "loss": 0.0975,
+      "step": 1985
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3413830379107395e-06,
+      "loss": 0.1153,
+      "step": 1986
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3331643023421813e-06,
+      "loss": 0.1702,
+      "step": 1987
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.324958111440051e-06,
+      "loss": 0.1171,
+      "step": 1988
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.316764478631518e-06,
+      "loss": 0.1119,
+      "step": 1989
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.30858341732321e-06,
+      "loss": 0.1026,
+      "step": 1990
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.300414940901182e-06,
+      "loss": 0.0939,
+      "step": 1991
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.292259062730897e-06,
+      "loss": 0.096,
+      "step": 1992
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2841157961572034e-06,
+      "loss": 0.1148,
+      "step": 1993
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2759851545043175e-06,
+      "loss": 0.1021,
+      "step": 1994
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2678671510757953e-06,
+      "loss": 0.0939,
+      "step": 1995
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.259761799154516e-06,
+      "loss": 0.1007,
+      "step": 1996
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.251669112002657e-06,
+      "loss": 0.1186,
+      "step": 1997
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.243589102861673e-06,
+      "loss": 0.0998,
+      "step": 1998
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.235521784952275e-06,
+      "loss": 0.1091,
+      "step": 1999
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.227467171474409e-06,
+      "loss": 0.1029,
+      "step": 2000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2194252756072343e-06,
+      "loss": 0.1175,
+      "step": 2001
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2113961105090933e-06,
+      "loss": 0.1293,
+      "step": 2002
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2033796893175152e-06,
+      "loss": 0.127,
+      "step": 2003
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.195376025149156e-06,
+      "loss": 0.1107,
+      "step": 2004
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1873851310998194e-06,
+      "loss": 0.1296,
+      "step": 2005
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.179407020244395e-06,
+      "loss": 0.1104,
+      "step": 2006
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1714417056368752e-06,
+      "loss": 0.0797,
+      "step": 2007
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1634892003102935e-06,
+      "loss": 0.1292,
+      "step": 2008
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.155549517276747e-06,
+      "loss": 0.1229,
+      "step": 2009
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1476226695273326e-06,
+      "loss": 0.0852,
+      "step": 2010
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1397086700321635e-06,
+      "loss": 0.1026,
+      "step": 2011
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1318075317403152e-06,
+      "loss": 0.1287,
+      "step": 2012
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.123919267579828e-06,
+      "loss": 0.123,
+      "step": 2013
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1160438904576743e-06,
+      "loss": 0.0912,
+      "step": 2014
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.108181413259741e-06,
+      "loss": 0.1008,
+      "step": 2015
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1003318488508107e-06,
+      "loss": 0.1106,
+      "step": 2016
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.092495210074532e-06,
+      "loss": 0.0858,
+      "step": 2017
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0846715097534087e-06,
+      "loss": 0.0867,
+      "step": 2018
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0768607606887724e-06,
+      "loss": 0.0914,
+      "step": 2019
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.069062975660765e-06,
+      "loss": 0.1193,
+      "step": 2020
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0612781674283142e-06,
+      "loss": 0.1053,
+      "step": 2021
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0535063487291176e-06,
+      "loss": 0.0846,
+      "step": 2022
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.045747532279616e-06,
+      "loss": 0.0811,
+      "step": 2023
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.038001730774978e-06,
+      "loss": 0.0969,
+      "step": 2024
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0302689568890753e-06,
+      "loss": 0.1034,
+      "step": 2025
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.022549223274465e-06,
+      "loss": 0.0862,
+      "step": 2026
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0148425425623673e-06,
+      "loss": 0.125,
+      "step": 2027
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0071489273626376e-06,
+      "loss": 0.1233,
+      "step": 2028
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.999468390263769e-06,
+      "loss": 0.087,
+      "step": 2029
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9918009438328365e-06,
+      "loss": 0.1066,
+      "step": 2030
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9841466006155162e-06,
+      "loss": 0.1011,
+      "step": 2031
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.976505373136025e-06,
+      "loss": 0.1167,
+      "step": 2032
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.968877273897136e-06,
+      "loss": 0.1227,
+      "step": 2033
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9612623153801267e-06,
+      "loss": 0.0989,
+      "step": 2034
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.953660510044789e-06,
+      "loss": 0.1138,
+      "step": 2035
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.946071870329377e-06,
+      "loss": 0.1013,
+      "step": 2036
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9384964086506185e-06,
+      "loss": 0.1002,
+      "step": 2037
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.930934137403665e-06,
+      "loss": 0.0981,
+      "step": 2038
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.923385068962095e-06,
+      "loss": 0.0852,
+      "step": 2039
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9158492156778807e-06,
+      "loss": 0.1117,
+      "step": 2040
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.908326589881372e-06,
+      "loss": 0.1172,
+      "step": 2041
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9008172038812744e-06,
+      "loss": 0.0892,
+      "step": 2042
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8933210699646342e-06,
+      "loss": 0.0998,
+      "step": 2043
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.885838200396808e-06,
+      "loss": 0.0732,
+      "step": 2044
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8783686074214546e-06,
+      "loss": 0.1003,
+      "step": 2045
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8709123032605058e-06,
+      "loss": 0.1159,
+      "step": 2046
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8634693001141513e-06,
+      "loss": 0.1147,
+      "step": 2047
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.856039610160818e-06,
+      "loss": 0.0852,
+      "step": 2048
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8486232455571473e-06,
+      "loss": 0.1415,
+      "step": 2049
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8412202184379801e-06,
+      "loss": 0.1121,
+      "step": 2050
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8338305409163314e-06,
+      "loss": 0.1079,
+      "step": 2051
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.826454225083375e-06,
+      "loss": 0.1166,
+      "step": 2052
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8190912830084207e-06,
+      "loss": 0.1307,
+      "step": 2053
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.811741726738898e-06,
+      "loss": 0.1295,
+      "step": 2054
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8044055683003358e-06,
+      "loss": 0.1511,
+      "step": 2055
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7970828196963286e-06,
+      "loss": 0.1095,
+      "step": 2056
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7897734929085508e-06,
+      "loss": 0.1019,
+      "step": 2057
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7824775998966926e-06,
+      "loss": 0.0995,
+      "step": 2058
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7751951525984857e-06,
+      "loss": 0.0803,
+      "step": 2059
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7679261629296408e-06,
+      "loss": 0.0895,
+      "step": 2060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7606706427838682e-06,
+      "loss": 0.0759,
+      "step": 2061
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7534286040328208e-06,
+      "loss": 0.1221,
+      "step": 2062
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7462000585261096e-06,
+      "loss": 0.0981,
+      "step": 2063
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7389850180912537e-06,
+      "loss": 0.1047,
+      "step": 2064
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7317834945336843e-06,
+      "loss": 0.106,
+      "step": 2065
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.724595499636711e-06,
+      "loss": 0.1012,
+      "step": 2066
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7174210451615091e-06,
+      "loss": 0.0927,
+      "step": 2067
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7102601428470988e-06,
+      "loss": 0.0968,
+      "step": 2068
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7031128044103272e-06,
+      "loss": 0.0896,
+      "step": 2069
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6959790415458454e-06,
+      "loss": 0.0997,
+      "step": 2070
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6888588659260929e-06,
+      "loss": 0.0948,
+      "step": 2071
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6817522892012762e-06,
+      "loss": 0.0968,
+      "step": 2072
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6746593229993545e-06,
+      "loss": 0.0989,
+      "step": 2073
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6675799789260128e-06,
+      "loss": 0.104,
+      "step": 2074
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6605142685646503e-06,
+      "loss": 0.0785,
+      "step": 2075
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6534622034763558e-06,
+      "loss": 0.1016,
+      "step": 2076
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6464237951998952e-06,
+      "loss": 0.0853,
+      "step": 2077
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6393990552516848e-06,
+      "loss": 0.1185,
+      "step": 2078
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6323879951257783e-06,
+      "loss": 0.0835,
+      "step": 2079
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6253906262938457e-06,
+      "loss": 0.1046,
+      "step": 2080
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6184069602051578e-06,
+      "loss": 0.0972,
+      "step": 2081
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.611437008286555e-06,
+      "loss": 0.0986,
+      "step": 2082
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6044807819424545e-06,
+      "loss": 0.0947,
+      "step": 2083
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5975382925547966e-06,
+      "loss": 0.115,
+      "step": 2084
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5906095514830645e-06,
+      "loss": 0.0683,
+      "step": 2085
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5836945700642248e-06,
+      "loss": 0.0959,
+      "step": 2086
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5767933596127528e-06,
+      "loss": 0.1139,
+      "step": 2087
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.56990593142057e-06,
+      "loss": 0.0873,
+      "step": 2088
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5630322967570655e-06,
+      "loss": 0.1096,
+      "step": 2089
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5561724668690436e-06,
+      "loss": 0.1147,
+      "step": 2090
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5493264529807305e-06,
+      "loss": 0.0993,
+      "step": 2091
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5424942662937436e-06,
+      "loss": 0.1052,
+      "step": 2092
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5356759179870762e-06,
+      "loss": 0.0907,
+      "step": 2093
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5288714192170796e-06,
+      "loss": 0.099,
+      "step": 2094
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.522080781117441e-06,
+      "loss": 0.113,
+      "step": 2095
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5153040147991716e-06,
+      "loss": 0.0837,
+      "step": 2096
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5085411313505849e-06,
+      "loss": 0.1132,
+      "step": 2097
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5017921418372772e-06,
+      "loss": 0.1019,
+      "step": 2098
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4950570573021138e-06,
+      "loss": 0.0984,
+      "step": 2099
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4883358887652044e-06,
+      "loss": 0.0972,
+      "step": 2100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4816286472238939e-06,
+      "loss": 0.1187,
+      "step": 2101
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.474935343652736e-06,
+      "loss": 0.0954,
+      "step": 2102
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4682559890034787e-06,
+      "loss": 0.1101,
+      "step": 2103
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.461590594205049e-06,
+      "loss": 0.124,
+      "step": 2104
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4549391701635308e-06,
+      "loss": 0.0929,
+      "step": 2105
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4483017277621482e-06,
+      "loss": 0.0896,
+      "step": 2106
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4416782778612514e-06,
+      "loss": 0.0902,
+      "step": 2107
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4350688312982864e-06,
+      "loss": 0.088,
+      "step": 2108
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.428473398887802e-06,
+      "loss": 0.1083,
+      "step": 2109
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.421891991421399e-06,
+      "loss": 0.0947,
+      "step": 2110
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4153246196677483e-06,
+      "loss": 0.0913,
+      "step": 2111
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4087712943725384e-06,
+      "loss": 0.1063,
+      "step": 2112
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.40223202625849e-06,
+      "loss": 0.1119,
+      "step": 2113
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.395706826025306e-06,
+      "loss": 0.0922,
+      "step": 2114
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3891957043496917e-06,
+      "loss": 0.1083,
+      "step": 2115
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3826986718852952e-06,
+      "loss": 0.0917,
+      "step": 2116
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3762157392627317e-06,
+      "loss": 0.092,
+      "step": 2117
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3697469170895282e-06,
+      "loss": 0.074,
+      "step": 2118
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.363292215950135e-06,
+      "loss": 0.0932,
+      "step": 2119
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3568516464058946e-06,
+      "loss": 0.0926,
+      "step": 2120
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.350425218995024e-06,
+      "loss": 0.1147,
+      "step": 2121
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3440129442326045e-06,
+      "loss": 0.1162,
+      "step": 2122
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3376148326105586e-06,
+      "loss": 0.1169,
+      "step": 2123
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3312308945976348e-06,
+      "loss": 0.0731,
+      "step": 2124
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3248611406393918e-06,
+      "loss": 0.1046,
+      "step": 2125
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.318505581158177e-06,
+      "loss": 0.1431,
+      "step": 2126
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3121642265531154e-06,
+      "loss": 0.1076,
+      "step": 2127
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.305837087200087e-06,
+      "loss": 0.0955,
+      "step": 2128
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.299524173451715e-06,
+      "loss": 0.0734,
+      "step": 2129
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2932254956373457e-06,
+      "loss": 0.1308,
+      "step": 2130
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.286941064063031e-06,
+      "loss": 0.0961,
+      "step": 2131
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2806708890115138e-06,
+      "loss": 0.0994,
+      "step": 2132
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2744149807422113e-06,
+      "loss": 0.1203,
+      "step": 2133
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2681733494911897e-06,
+      "loss": 0.1043,
+      "step": 2134
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2619460054711685e-06,
+      "loss": 0.1023,
+      "step": 2135
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2557329588714739e-06,
+      "loss": 0.1067,
+      "step": 2136
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2495342198580562e-06,
+      "loss": 0.1105,
+      "step": 2137
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2433497985734356e-06,
+      "loss": 0.1109,
+      "step": 2138
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.237179705136725e-06,
+      "loss": 0.1237,
+      "step": 2139
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2310239496435749e-06,
+      "loss": 0.1037,
+      "step": 2140
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2248825421661937e-06,
+      "loss": 0.0939,
+      "step": 2141
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2187554927532963e-06,
+      "loss": 0.1103,
+      "step": 2142
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2126428114301204e-06,
+      "loss": 0.085,
+      "step": 2143
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2065445081983795e-06,
+      "loss": 0.0743,
+      "step": 2144
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2004605930362724e-06,
+      "loss": 0.1251,
+      "step": 2145
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.194391075898451e-06,
+      "loss": 0.1022,
+      "step": 2146
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1883359667160087e-06,
+      "loss": 0.0874,
+      "step": 2147
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1822952753964667e-06,
+      "loss": 0.0915,
+      "step": 2148
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1762690118237518e-06,
+      "loss": 0.1314,
+      "step": 2149
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1702571858581867e-06,
+      "loss": 0.1144,
+      "step": 2150
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1642598073364707e-06,
+      "loss": 0.0811,
+      "step": 2151
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.158276886071662e-06,
+      "loss": 0.1055,
+      "step": 2152
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1523084318531641e-06,
+      "loss": 0.0935,
+      "step": 2153
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1463544544467109e-06,
+      "loss": 0.0897,
+      "step": 2154
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1404149635943462e-06,
+      "loss": 0.0855,
+      "step": 2155
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.134489969014414e-06,
+      "loss": 0.119,
+      "step": 2156
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1285794804015349e-06,
+      "loss": 0.0902,
+      "step": 2157
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1226835074265985e-06,
+      "loss": 0.1167,
+      "step": 2158
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1168020597367435e-06,
+      "loss": 0.1005,
+      "step": 2159
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1109351469553331e-06,
+      "loss": 0.0844,
+      "step": 2160
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.105082778681964e-06,
+      "loss": 0.0784,
+      "step": 2161
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0992449644924186e-06,
+      "loss": 0.1007,
+      "step": 2162
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0934217139386805e-06,
+      "loss": 0.1119,
+      "step": 2163
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.087613036548888e-06,
+      "loss": 0.0931,
+      "step": 2164
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0818189418273527e-06,
+      "loss": 0.0895,
+      "step": 2165
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0760394392545058e-06,
+      "loss": 0.1105,
+      "step": 2166
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0702745382869207e-06,
+      "loss": 0.0786,
+      "step": 2167
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.064524248357265e-06,
+      "loss": 0.0833,
+      "step": 2168
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0587885788743112e-06,
+      "loss": 0.0958,
+      "step": 2169
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0530675392228995e-06,
+      "loss": 0.1067,
+      "step": 2170
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.047361138763937e-06,
+      "loss": 0.1068,
+      "step": 2171
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0416693868343796e-06,
+      "loss": 0.1244,
+      "step": 2172
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.035992292747211e-06,
+      "loss": 0.0838,
+      "step": 2173
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.030329865791434e-06,
+      "loss": 0.1138,
+      "step": 2174
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0246821152320507e-06,
+      "loss": 0.0855,
+      "step": 2175
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0190490503100515e-06,
+      "loss": 0.0889,
+      "step": 2176
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0134306802423965e-06,
+      "loss": 0.0766,
+      "step": 2177
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0078270142220015e-06,
+      "loss": 0.1117,
+      "step": 2178
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0022380614177251e-06,
+      "loss": 0.1197,
+      "step": 2179
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.966638309743481e-07,
+      "loss": 0.0921,
+      "step": 2180
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.911043320125657e-07,
+      "loss": 0.1137,
+      "step": 2181
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.85559573628967e-07,
+      "loss": 0.0942,
+      "step": 2182
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.800295648960245e-07,
+      "loss": 0.0994,
+      "step": 2183
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.74514314862074e-07,
+      "loss": 0.1119,
+      "step": 2184
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.690138325513043e-07,
+      "loss": 0.1077,
+      "step": 2185
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.635281269637409e-07,
+      "loss": 0.0962,
+      "step": 2186
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.580572070752335e-07,
+      "loss": 0.1031,
+      "step": 2187
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.52601081837431e-07,
+      "loss": 0.0844,
+      "step": 2188
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.471597601777871e-07,
+      "loss": 0.0991,
+      "step": 2189
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.417332509995203e-07,
+      "loss": 0.1286,
+      "step": 2190
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.36321563181628e-07,
+      "loss": 0.1021,
+      "step": 2191
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.309247055788384e-07,
+      "loss": 0.0727,
+      "step": 2192
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.255426870216311e-07,
+      "loss": 0.1175,
+      "step": 2193
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.201755163161918e-07,
+      "loss": 0.1144,
+      "step": 2194
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.148232022444259e-07,
+      "loss": 0.0812,
+      "step": 2195
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.094857535639157e-07,
+      "loss": 0.103,
+      "step": 2196
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.041631790079275e-07,
+      "loss": 0.0816,
+      "step": 2197
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.988554872853927e-07,
+      "loss": 0.1036,
+      "step": 2198
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.935626870808856e-07,
+      "loss": 0.0742,
+      "step": 2199
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.882847870546174e-07,
+      "loss": 0.1024,
+      "step": 2200
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.830217958424192e-07,
+      "loss": 0.1329,
+      "step": 2201
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.77773722055727e-07,
+      "loss": 0.0707,
+      "step": 2202
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.725405742815695e-07,
+      "loss": 0.094,
+      "step": 2203
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.673223610825532e-07,
+      "loss": 0.1018,
+      "step": 2204
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.62119090996848e-07,
+      "loss": 0.0858,
+      "step": 2205
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.569307725381715e-07,
+      "loss": 0.128,
+      "step": 2206
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.517574141957796e-07,
+      "loss": 0.1041,
+      "step": 2207
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.46599024434449e-07,
+      "loss": 0.1041,
+      "step": 2208
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.414556116944672e-07,
+      "loss": 0.0922,
+      "step": 2209
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.363271843916099e-07,
+      "loss": 0.1097,
+      "step": 2210
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.312137509171392e-07,
+      "loss": 0.0856,
+      "step": 2211
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.261153196377814e-07,
+      "loss": 0.0861,
+      "step": 2212
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.210318988957166e-07,
+      "loss": 0.112,
+      "step": 2213
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.159634970085595e-07,
+      "loss": 0.1006,
+      "step": 2214
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.109101222693616e-07,
+      "loss": 0.0876,
+      "step": 2215
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.058717829465723e-07,
+      "loss": 0.1351,
+      "step": 2216
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.008484872840538e-07,
+      "loss": 0.1204,
+      "step": 2217
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.958402435010415e-07,
+      "loss": 0.1224,
+      "step": 2218
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.908470597921547e-07,
+      "loss": 0.0896,
+      "step": 2219
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.858689443273548e-07,
+      "loss": 0.0815,
+      "step": 2220
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.809059052519674e-07,
+      "loss": 0.1093,
+      "step": 2221
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.759579506866311e-07,
+      "loss": 0.0938,
+      "step": 2222
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.710250887273196e-07,
+      "loss": 0.0851,
+      "step": 2223
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.66107327445299e-07,
+      "loss": 0.108,
+      "step": 2224
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 0.0616,
+      "step": 2225
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.563171390746627e-07,
+      "loss": 0.1059,
+      "step": 2226
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.514447280049964e-07,
+      "loss": 0.0962,
+      "step": 2227
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.465874496504944e-07,
+      "loss": 0.0948,
+      "step": 2228
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.417453119587525e-07,
+      "loss": 0.117,
+      "step": 2229
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.369183228526e-07,
+      "loss": 0.0999,
+      "step": 2230
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.321064902300723e-07,
+      "loss": 0.0896,
+      "step": 2231
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.273098219644137e-07,
+      "loss": 0.0989,
+      "step": 2232
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.225283259040472e-07,
+      "loss": 0.1103,
+      "step": 2233
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.17762009872579e-07,
+      "loss": 0.1049,
+      "step": 2234
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.130108816687687e-07,
+      "loss": 0.0822,
+      "step": 2235
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.082749490665353e-07,
+      "loss": 0.0901,
+      "step": 2236
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.035542198149237e-07,
+      "loss": 0.0718,
+      "step": 2237
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.98848701638114e-07,
+      "loss": 0.1173,
+      "step": 2238
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.941584022353865e-07,
+      "loss": 0.0784,
+      "step": 2239
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.894833292811265e-07,
+      "loss": 0.1205,
+      "step": 2240
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.848234904248041e-07,
+      "loss": 0.1219,
+      "step": 2241
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.801788932909648e-07,
+      "loss": 0.1046,
+      "step": 2242
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.755495454792116e-07,
+      "loss": 0.0728,
+      "step": 2243
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.709354545641989e-07,
+      "loss": 0.0938,
+      "step": 2244
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.663366280956152e-07,
+      "loss": 0.1286,
+      "step": 2245
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.617530735981758e-07,
+      "loss": 0.0884,
+      "step": 2246
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.571847985716063e-07,
+      "loss": 0.1028,
+      "step": 2247
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.526318104906293e-07,
+      "loss": 0.0963,
+      "step": 2248
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.480941168049593e-07,
+      "loss": 0.0757,
+      "step": 2249
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.435717249392803e-07,
+      "loss": 0.1049,
+      "step": 2250
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.390646422932445e-07,
+      "loss": 0.1034,
+      "step": 2251
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.345728762414504e-07,
+      "loss": 0.096,
+      "step": 2252
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.300964341334382e-07,
+      "loss": 0.0752,
+      "step": 2253
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.256353232936718e-07,
+      "loss": 0.0803,
+      "step": 2254
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.211895510215316e-07,
+      "loss": 0.1036,
+      "step": 2255
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.167591245913029e-07,
+      "loss": 0.1109,
+      "step": 2256
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.123440512521539e-07,
+      "loss": 0.0772,
+      "step": 2257
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.079443382281424e-07,
+      "loss": 0.1204,
+      "step": 2258
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.035599927181834e-07,
+      "loss": 0.0903,
+      "step": 2259
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.99191021896055e-07,
+      "loss": 0.0984,
+      "step": 2260
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.948374329103723e-07,
+      "loss": 0.0949,
+      "step": 2261
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.904992328845893e-07,
+      "loss": 0.1001,
+      "step": 2262
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.861764289169713e-07,
+      "loss": 0.0909,
+      "step": 2263
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.818690280806038e-07,
+      "loss": 0.1025,
+      "step": 2264
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.775770374233558e-07,
+      "loss": 0.0997,
+      "step": 2265
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.73300463967893e-07,
+      "loss": 0.0833,
+      "step": 2266
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.690393147116491e-07,
+      "loss": 0.0912,
+      "step": 2267
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.647935966268225e-07,
+      "loss": 0.0788,
+      "step": 2268
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.60563316660363e-07,
+      "loss": 0.1073,
+      "step": 2269
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.563484817339581e-07,
+      "loss": 0.1038,
+      "step": 2270
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.521490987440259e-07,
+      "loss": 0.1089,
+      "step": 2271
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.47965174561701e-07,
+      "loss": 0.0866,
+      "step": 2272
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.437967160328228e-07,
+      "loss": 0.0994,
+      "step": 2273
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.396437299779278e-07,
+      "loss": 0.0781,
+      "step": 2274
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.355062231922326e-07,
+      "loss": 0.1184,
+      "step": 2275
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.313842024456306e-07,
+      "loss": 0.1128,
+      "step": 2276
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.272776744826724e-07,
+      "loss": 0.0903,
+      "step": 2277
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.231866460225621e-07,
+      "loss": 0.087,
+      "step": 2278
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.191111237591406e-07,
+      "loss": 0.0942,
+      "step": 2279
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.150511143608782e-07,
+      "loss": 0.132,
+      "step": 2280
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.110066244708645e-07,
+      "loss": 0.1114,
+      "step": 2281
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.069776607067944e-07,
+      "loss": 0.1287,
+      "step": 2282
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.029642296609538e-07,
+      "loss": 0.0852,
+      "step": 2283
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.98966337900224e-07,
+      "loss": 0.0937,
+      "step": 2284
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.949839919660481e-07,
+      "loss": 0.1149,
+      "step": 2285
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.91017198374445e-07,
+      "loss": 0.0541,
+      "step": 2286
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.870659636159758e-07,
+      "loss": 0.1147,
+      "step": 2287
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.831302941557537e-07,
+      "loss": 0.1211,
+      "step": 2288
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.79210196433414e-07,
+      "loss": 0.0908,
+      "step": 2289
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7530567686312035e-07,
+      "loss": 0.0813,
+      "step": 2290
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7141674183354247e-07,
+      "loss": 0.0894,
+      "step": 2291
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6754339770785474e-07,
+      "loss": 0.092,
+      "step": 2292
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.636856508237164e-07,
+      "loss": 0.0701,
+      "step": 2293
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5984350749326835e-07,
+      "loss": 0.1288,
+      "step": 2294
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5601697400312175e-07,
+      "loss": 0.115,
+      "step": 2295
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5220605661434605e-07,
+      "loss": 0.0927,
+      "step": 2296
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4841076156245665e-07,
+      "loss": 0.0889,
+      "step": 2297
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.4463109505741177e-07,
+      "loss": 0.0703,
+      "step": 2298
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.408670632835932e-07,
+      "loss": 0.1006,
+      "step": 2299
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3711867239980335e-07,
+      "loss": 0.1015,
+      "step": 2300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3338592853925277e-07,
+      "loss": 0.0883,
+      "step": 2301
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.296688378095493e-07,
+      "loss": 0.0901,
+      "step": 2302
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2596740629268997e-07,
+      "loss": 0.0907,
+      "step": 2303
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.222816400450458e-07,
+      "loss": 0.1006,
+      "step": 2304
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.186115450973616e-07,
+      "loss": 0.0752,
+      "step": 2305
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1495712745473595e-07,
+      "loss": 0.082,
+      "step": 2306
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1131839309661803e-07,
+      "loss": 0.0911,
+      "step": 2307
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.076953479767964e-07,
+      "loss": 0.0941,
+      "step": 2308
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.040879980233836e-07,
+      "loss": 0.1101,
+      "step": 2309
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.004963491388203e-07,
+      "loss": 0.1013,
+      "step": 2310
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.969204071998445e-07,
+      "loss": 0.0915,
+      "step": 2311
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.93360178057508e-07,
+      "loss": 0.1022,
+      "step": 2312
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8981566753714116e-07,
+      "loss": 0.1056,
+      "step": 2313
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8628688143836244e-07,
+      "loss": 0.0979,
+      "step": 2314
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.827738255350555e-07,
+      "loss": 0.1064,
+      "step": 2315
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.792765055753755e-07,
+      "loss": 0.0824,
+      "step": 2316
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.757949272817174e-07,
+      "loss": 0.0994,
+      "step": 2317
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.723290963507309e-07,
+      "loss": 0.0646,
+      "step": 2318
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.688790184532909e-07,
+      "loss": 0.0821,
+      "step": 2319
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.654446992345018e-07,
+      "loss": 0.1042,
+      "step": 2320
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.620261443136819e-07,
+      "loss": 0.1142,
+      "step": 2321
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5862335928435465e-07,
+      "loss": 0.0867,
+      "step": 2322
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5523634971424194e-07,
+      "loss": 0.0906,
+      "step": 2323
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.5186512114525283e-07,
+      "loss": 0.0699,
+      "step": 2324
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.485096790934739e-07,
+      "loss": 0.095,
+      "step": 2325
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.451700290491633e-07,
+      "loss": 0.1068,
+      "step": 2326
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4184617647674e-07,
+      "loss": 0.1068,
+      "step": 2327
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3853812681477136e-07,
+      "loss": 0.1246,
+      "step": 2328
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3524588547597327e-07,
+      "loss": 0.0876,
+      "step": 2329
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3196945784718993e-07,
+      "loss": 0.0911,
+      "step": 2330
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.287088492893942e-07,
+      "loss": 0.0945,
+      "step": 2331
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.25464065137675e-07,
+      "loss": 0.0957,
+      "step": 2332
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2223511070122893e-07,
+      "loss": 0.1094,
+      "step": 2333
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.190219912633519e-07,
+      "loss": 0.0998,
+      "step": 2334
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.158247120814251e-07,
+      "loss": 0.0953,
+      "step": 2335
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1264327838692153e-07,
+      "loss": 0.1058,
+      "step": 2336
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.094776953853762e-07,
+      "loss": 0.0914,
+      "step": 2337
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.063279682564002e-07,
+      "loss": 0.1128,
+      "step": 2338
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0319410215365e-07,
+      "loss": 0.0817,
+      "step": 2339
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0007610220483927e-07,
+      "loss": 0.1036,
+      "step": 2340
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.969739735117128e-07,
+      "loss": 0.1169,
+      "step": 2341
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9388772115005457e-07,
+      "loss": 0.1043,
+      "step": 2342
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9081735016966205e-07,
+      "loss": 0.0976,
+      "step": 2343
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.877628655943576e-07,
+      "loss": 0.0884,
+      "step": 2344
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.847242724219612e-07,
+      "loss": 0.0796,
+      "step": 2345
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8170157562429466e-07,
+      "loss": 0.0781,
+      "step": 2346
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7869478014716953e-07,
+      "loss": 0.0797,
+      "step": 2347
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.757038909103793e-07,
+      "loss": 0.0938,
+      "step": 2348
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7272891280769044e-07,
+      "loss": 0.1047,
+      "step": 2349
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.697698507068358e-07,
+      "loss": 0.0803,
+      "step": 2350
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6682670944950804e-07,
+      "loss": 0.1028,
+      "step": 2351
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.638994938513451e-07,
+      "loss": 0.0959,
+      "step": 2352
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.609882087019311e-07,
+      "loss": 0.0971,
+      "step": 2353
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.580928587647824e-07,
+      "loss": 0.14,
+      "step": 2354
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5521344877734165e-07,
+      "loss": 0.0884,
+      "step": 2355
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.523499834509724e-07,
+      "loss": 0.1259,
+      "step": 2356
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.495024674709468e-07,
+      "loss": 0.1064,
+      "step": 2357
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4667090549644e-07,
+      "loss": 0.1196,
+      "step": 2358
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.43855302160525e-07,
+      "loss": 0.1262,
+      "step": 2359
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4105566207016207e-07,
+      "loss": 0.0922,
+      "step": 2360
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3827198980619025e-07,
+      "loss": 0.1024,
+      "step": 2361
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3550428992332508e-07,
+      "loss": 0.0752,
+      "step": 2362
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.327525669501418e-07,
+      "loss": 0.1002,
+      "step": 2363
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3001682538908333e-07,
+      "loss": 0.1056,
+      "step": 2364
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2729706971643117e-07,
+      "loss": 0.1298,
+      "step": 2365
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.245933043823234e-07,
+      "loss": 0.1013,
+      "step": 2366
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2190553381072234e-07,
+      "loss": 0.087,
+      "step": 2367
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1923376239942895e-07,
+      "loss": 0.0836,
+      "step": 2368
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1657799452005856e-07,
+      "loss": 0.1233,
+      "step": 2369
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.139382345180474e-07,
+      "loss": 0.093,
+      "step": 2370
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1131448671263378e-07,
+      "loss": 0.1034,
+      "step": 2371
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0870675539686024e-07,
+      "loss": 0.094,
+      "step": 2372
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0611504483756038e-07,
+      "loss": 0.0853,
+      "step": 2373
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0353935927535428e-07,
+      "loss": 0.0767,
+      "step": 2374
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.009797029246452e-07,
+      "loss": 0.1102,
+      "step": 2375
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9843607997360403e-07,
+      "loss": 0.0966,
+      "step": 2376
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.959084945841705e-07,
+      "loss": 0.0655,
+      "step": 2377
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9339695089204192e-07,
+      "loss": 0.0849,
+      "step": 2378
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9090145300666885e-07,
+      "loss": 0.0969,
+      "step": 2379
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.884220050112462e-07,
+      "loss": 0.0739,
+      "step": 2380
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8595861096270874e-07,
+      "loss": 0.0774,
+      "step": 2381
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8351127489172227e-07,
+      "loss": 0.104,
+      "step": 2382
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8108000080267918e-07,
+      "loss": 0.0755,
+      "step": 2383
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7866479267369062e-07,
+      "loss": 0.1073,
+      "step": 2384
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7626565445657883e-07,
+      "loss": 0.1075,
+      "step": 2385
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7388259007687368e-07,
+      "loss": 0.1037,
+      "step": 2386
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.715156034338039e-07,
+      "loss": 0.105,
+      "step": 2387
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.691646984002937e-07,
+      "loss": 0.0742,
+      "step": 2388
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6682987882294722e-07,
+      "loss": 0.0929,
+      "step": 2389
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6451114852206073e-07,
+      "loss": 0.1,
+      "step": 2390
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6220851129159164e-07,
+      "loss": 0.0807,
+      "step": 2391
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5992197089917727e-07,
+      "loss": 0.0696,
+      "step": 2392
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.576515310861071e-07,
+      "loss": 0.113,
+      "step": 2393
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.55397195567335e-07,
+      "loss": 0.1101,
+      "step": 2394
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5315896803145824e-07,
+      "loss": 0.1001,
+      "step": 2395
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5093685214072173e-07,
+      "loss": 0.0885,
+      "step": 2396
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4873085153100485e-07,
+      "loss": 0.1009,
+      "step": 2397
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4654096981182031e-07,
+      "loss": 0.0987,
+      "step": 2398
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4436721056630853e-07,
+      "loss": 0.0953,
+      "step": 2399
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4220957735122663e-07,
+      "loss": 0.1018,
+      "step": 2400
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.400680736969484e-07,
+      "loss": 0.1134,
+      "step": 2401
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3794270310745538e-07,
+      "loss": 0.1038,
+      "step": 2402
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3583346906033024e-07,
+      "loss": 0.1019,
+      "step": 2403
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3374037500675452e-07,
+      "loss": 0.0931,
+      "step": 2404
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3166342437150204e-07,
+      "loss": 0.0977,
+      "step": 2405
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2960262055292884e-07,
+      "loss": 0.0936,
+      "step": 2406
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.275579669229743e-07,
+      "loss": 0.1158,
+      "step": 2407
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2552946682715116e-07,
+      "loss": 0.1138,
+      "step": 2408
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2351712358454115e-07,
+      "loss": 0.1286,
+      "step": 2409
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.215209404877904e-07,
+      "loss": 0.1031,
+      "step": 2410
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1954092080310288e-07,
+      "loss": 0.0922,
+      "step": 2411
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1757706777023592e-07,
+      "loss": 0.0974,
+      "step": 2412
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1562938460249473e-07,
+      "loss": 0.0996,
+      "step": 2413
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1369787448672675e-07,
+      "loss": 0.0719,
+      "step": 2414
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1178254058331616e-07,
+      "loss": 0.1146,
+      "step": 2415
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0988338602618053e-07,
+      "loss": 0.0908,
+      "step": 2416
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0800041392276194e-07,
+      "loss": 0.1051,
+      "step": 2417
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.061336273540281e-07,
+      "loss": 0.0946,
+      "step": 2418
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0428302937445899e-07,
+      "loss": 0.098,
+      "step": 2419
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.024486230120525e-07,
+      "loss": 0.0644,
+      "step": 2420
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0063041126830542e-07,
+      "loss": 0.1095,
+      "step": 2421
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.882839711822468e-08,
+      "loss": 0.0811,
+      "step": 2422
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.704258351030838e-08,
+      "loss": 0.1116,
+      "step": 2423
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.527297336654917e-08,
+      "loss": 0.1015,
+      "step": 2424
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.351956958242648e-08,
+      "loss": 0.1056,
+      "step": 2425
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.178237502690423e-08,
+      "loss": 0.1237,
+      "step": 2426
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.006139254242319e-08,
+      "loss": 0.0856,
+      "step": 2427
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.835662494489638e-08,
+      "loss": 0.0803,
+      "step": 2428
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.6668075023707e-08,
+      "loss": 0.0869,
+      "step": 2429
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.499574554170276e-08,
+      "loss": 0.0859,
+      "step": 2430
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.333963923519039e-08,
+      "loss": 0.1169,
+      "step": 2431
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.169975881393122e-08,
+      "loss": 0.0719,
+      "step": 2432
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.007610696114e-08,
+      "loss": 0.095,
+      "step": 2433
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.846868633347492e-08,
+      "loss": 0.0792,
+      "step": 2434
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.687749956103774e-08,
+      "loss": 0.0871,
+      "step": 2435
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.530254924736691e-08,
+      "loss": 0.1289,
+      "step": 2436
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.374383796943663e-08,
+      "loss": 0.0956,
+      "step": 2437
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.22013682776479e-08,
+      "loss": 0.1094,
+      "step": 2438
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.067514269582743e-08,
+      "loss": 0.1248,
+      "step": 2439
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.916516372122429e-08,
+      "loss": 0.1058,
+      "step": 2440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.767143382450214e-08,
+      "loss": 0.1251,
+      "step": 2441
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.619395544974039e-08,
+      "loss": 0.0831,
+      "step": 2442
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.473273101442412e-08,
+      "loss": 0.1151,
+      "step": 2443
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.32877629094475e-08,
+      "loss": 0.0939,
+      "step": 2444
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.185905349910038e-08,
+      "loss": 0.1188,
+      "step": 2445
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.044660512107392e-08,
+      "loss": 0.0817,
+      "step": 2446
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.905042008645057e-08,
+      "loss": 0.1092,
+      "step": 2447
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.7670500679702925e-08,
+      "loss": 0.1127,
+      "step": 2448
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.630684915868934e-08,
+      "loss": 0.0899,
+      "step": 2449
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.4959467754651665e-08,
+      "loss": 0.1035,
+      "step": 2450
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3628358672205285e-08,
+      "loss": 0.0841,
+      "step": 2451
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.231352408934687e-08,
+      "loss": 0.1184,
+      "step": 2452
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.101496615744106e-08,
+      "loss": 0.0993,
+      "step": 2453
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.973268700121936e-08,
+      "loss": 0.1074,
+      "step": 2454
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.846668871877902e-08,
+      "loss": 0.1041,
+      "step": 2455
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.721697338157749e-08,
+      "loss": 0.098,
+      "step": 2456
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.59835430344302e-08,
+      "loss": 0.1006,
+      "step": 2457
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.476639969550722e-08,
+      "loss": 0.1189,
+      "step": 2458
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.3565545356327734e-08,
+      "loss": 0.1057,
+      "step": 2459
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2380981981759994e-08,
+      "loss": 0.0892,
+      "step": 2460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.1212711510015826e-08,
+      "loss": 0.0948,
+      "step": 2461
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.006073585264725e-08,
+      "loss": 0.1016,
+      "step": 2462
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.89250568945454e-08,
+      "loss": 0.0968,
+      "step": 2463
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7805676493938294e-08,
+      "loss": 0.0885,
+      "step": 2464
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.6702596482381946e-08,
+      "loss": 0.1009,
+      "step": 2465
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.5615818664764825e-08,
+      "loss": 0.1102,
+      "step": 2466
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.454534481929783e-08,
+      "loss": 0.1006,
+      "step": 2467
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.349117669751767e-08,
+      "loss": 0.0883,
+      "step": 2468
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.245331602428126e-08,
+      "loss": 0.0933,
+      "step": 2469
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.14317644977602e-08,
+      "loss": 0.1268,
+      "step": 2470
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0426523789442994e-08,
+      "loss": 0.1061,
+      "step": 2471
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9437595544130615e-08,
+      "loss": 0.0721,
+      "step": 2472
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8464981379929814e-08,
+      "loss": 0.0766,
+      "step": 2473
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7508682888257587e-08,
+      "loss": 0.109,
+      "step": 2474
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.6568701633832295e-08,
+      "loss": 0.1099,
+      "step": 2475
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5645039154675867e-08,
+      "loss": 0.0725,
+      "step": 2476
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4737696962106038e-08,
+      "loss": 0.1212,
+      "step": 2477
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3846676540739687e-08,
+      "loss": 0.0939,
+      "step": 2478
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2971979348485053e-08,
+      "loss": 0.0941,
+      "step": 2479
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2113606816546172e-08,
+      "loss": 0.1126,
+      "step": 2480
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.127156034941069e-08,
+      "loss": 0.0881,
+      "step": 2481
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0445841324856497e-08,
+      "loss": 0.0916,
+      "step": 2482
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9636451093947296e-08,
+      "loss": 0.0631,
+      "step": 2483
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8843390981024835e-08,
+      "loss": 0.0836,
+      "step": 2484
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8066662283715562e-08,
+      "loss": 0.113,
+      "step": 2485
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7306266272921756e-08,
+      "loss": 0.1282,
+      "step": 2486
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6562204192821507e-08,
+      "loss": 0.0882,
+      "step": 2487
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.583447726086762e-08,
+      "loss": 0.0831,
+      "step": 2488
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5123086667786502e-08,
+      "loss": 0.1008,
+      "step": 2489
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4428033577571498e-08,
+      "loss": 0.0951,
+      "step": 2490
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3749319127486228e-08,
+      "loss": 0.1088,
+      "step": 2491
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3086944428060132e-08,
+      "loss": 0.1057,
+      "step": 2492
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2440910563086273e-08,
+      "loss": 0.092,
+      "step": 2493
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.181121858962353e-08,
+      "loss": 0.1051,
+      "step": 2494
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1197869537986627e-08,
+      "loss": 0.0694,
+      "step": 2495
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0600864411753897e-08,
+      "loss": 0.101,
+      "step": 2496
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0020204187759507e-08,
+      "loss": 0.0943,
+      "step": 2497
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.455889816095687e-09,
+      "loss": 0.0721,
+      "step": 2498
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.907922220104947e-09,
+      "loss": 0.0903,
+      "step": 2499
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.376302296387862e-09,
+      "loss": 0.1176,
+      "step": 2500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.861030914791956e-09,
+      "loss": 0.1007,
+      "step": 2501
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.362108918418376e-09,
+      "loss": 0.1127,
+      "step": 2502
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.8795371236163315e-09,
+      "loss": 0.0891,
+      "step": 2503
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.413316319979768e-09,
+      "loss": 0.0974,
+      "step": 2504
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.9634472703518075e-09,
+      "loss": 0.1013,
+      "step": 2505
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.529930710820308e-09,
+      "loss": 0.0932,
+      "step": 2506
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.112767350713421e-09,
+      "loss": 0.096,
+      "step": 2507
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.711957872606254e-09,
+      "loss": 0.1019,
+      "step": 2508
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.327502932311989e-09,
+      "loss": 0.1083,
+      "step": 2509
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.959403158885211e-09,
+      "loss": 0.1349,
+      "step": 2510
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.607659154621912e-09,
+      "loss": 0.1229,
+      "step": 2511
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.2722714950517154e-09,
+      "loss": 0.0949,
+      "step": 2512
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.95324072894565e-09,
+      "loss": 0.1158,
+      "step": 2513
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6505673783094875e-09,
+      "loss": 0.1115,
+      "step": 2514
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.364251938384854e-09,
+      "loss": 0.1134,
+      "step": 2515
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.0942948776481175e-09,
+      "loss": 0.0909,
+      "step": 2516
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8406966378103909e-09,
+      "loss": 0.1012,
+      "step": 2517
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6034576338141982e-09,
+      "loss": 0.0986,
+      "step": 2518
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3825782538368083e-09,
+      "loss": 0.1018,
+      "step": 2519
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.178058859285791e-09,
+      "loss": 0.0954,
+      "step": 2520
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.898997848001302e-10,
+      "loss": 0.1006,
+      "step": 2521
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.181013382524416e-10,
+      "loss": 0.1085,
+      "step": 2522
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.626638007434239e-10,
+      "loss": 0.0824,
+      "step": 2523
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.23587426601857e-10,
+      "loss": 0.1167,
+      "step": 2524
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.008724433890443e-10,
+      "loss": 0.1106,
+      "step": 2525
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.94519051895481e-10,
+      "loss": 0.0982,
+      "step": 2526
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0452742613641386e-10,
+      "loss": 0.0894,
+      "step": 2527
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3089771336072256e-10,
+      "loss": 0.1176,
+      "step": 2528
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.363003404314839e-11,
+      "loss": 0.0861,
+      "step": 2529
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.272448188429422e-11,
+      "loss": 0.1179,
+      "step": 2530
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.181123817285752e-12,
+      "loss": 0.0809,
+      "step": 2531
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.0594,
+      "step": 2532
+    },
+    {
+      "epoch": 2.0,
+      "step": 2532,
+      "total_flos": 436338855837696.0,
+      "train_loss": 0.13578517516077412,
+      "train_runtime": 10680.9408,
+      "train_samples_per_second": 30.326,
+      "train_steps_per_second": 0.237
     }
   ],
   "logging_steps": 1.0,
-  "max_steps": 1303,
+  "max_steps": 2532,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 1,
-  "save_steps": 800,
-  "total_flos": 223839164989440.0,
+  "num_train_epochs": 2,
+  "save_steps": 1304,
+  "total_flos": 436338855837696.0,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null