diff --git "a/checkpoint-2000/trainer_state.json" "b/checkpoint-2000/trainer_state.json" --- "a/checkpoint-2000/trainer_state.json" +++ "b/checkpoint-2000/trainer_state.json" @@ -10,12002 +10,12002 @@ "log_history": [ { "epoch": 0.0, - "learning_rate": 8.333333333333334e-06, + "learning_rate": 3.3333333333333333e-06, "loss": 3.1281, "step": 1 }, { "epoch": 0.0, - "learning_rate": 1.6666666666666667e-05, + "learning_rate": 6.666666666666667e-06, "loss": 3.2134, "step": 2 }, { "epoch": 0.0, - "learning_rate": 2.5e-05, - "loss": 3.1325, + "learning_rate": 1e-05, + "loss": 3.1329, "step": 3 }, { "epoch": 0.0, - "learning_rate": 3.3333333333333335e-05, - "loss": 3.1263, + "learning_rate": 1.3333333333333333e-05, + "loss": 3.1299, "step": 4 }, { "epoch": 0.01, - "learning_rate": 4.1666666666666665e-05, - "loss": 3.1569, + "learning_rate": 1.6666666666666667e-05, + "loss": 3.1673, "step": 5 }, { "epoch": 0.01, - "learning_rate": 5e-05, - "loss": 3.213, + "learning_rate": 2e-05, + "loss": 3.2316, "step": 6 }, { "epoch": 0.01, - "learning_rate": 5.833333333333333e-05, - "loss": 3.2105, + "learning_rate": 2.3333333333333336e-05, + "loss": 3.2393, "step": 7 }, { "epoch": 0.01, - "learning_rate": 6.666666666666667e-05, - "loss": 3.1681, + "learning_rate": 2.6666666666666667e-05, + "loss": 3.2113, "step": 8 }, { "epoch": 0.01, - "learning_rate": 7.5e-05, - "loss": 3.1277, + "learning_rate": 3e-05, + "loss": 3.19, "step": 9 }, { "epoch": 0.01, - "learning_rate": 8.333333333333333e-05, - "loss": 3.1917, + "learning_rate": 3.3333333333333335e-05, + "loss": 3.2793, "step": 10 }, { "epoch": 0.01, - "learning_rate": 9.166666666666667e-05, - "loss": 3.0592, + "learning_rate": 3.6666666666666666e-05, + "loss": 3.1682, "step": 11 }, { "epoch": 0.01, - "learning_rate": 0.0001, - "loss": 3.0844, + "learning_rate": 4e-05, + "loss": 3.2348, "step": 12 }, { "epoch": 0.01, - "learning_rate": 0.00010833333333333334, - "loss": 2.8544, + "learning_rate": 4.3333333333333334e-05, + "loss": 3.0313, "step": 13 }, { "epoch": 0.02, - "learning_rate": 0.00011666666666666667, - "loss": 2.8214, + "learning_rate": 4.666666666666667e-05, + "loss": 3.0315, "step": 14 }, { "epoch": 0.02, - "learning_rate": 0.000125, - "loss": 2.829, + "learning_rate": 5e-05, + "loss": 3.0926, "step": 15 }, { "epoch": 0.02, - "learning_rate": 0.00013333333333333334, - "loss": 2.7026, + "learning_rate": 5.333333333333333e-05, + "loss": 2.9897, "step": 16 }, { "epoch": 0.02, - "learning_rate": 0.00014166666666666668, - "loss": 2.6199, + "learning_rate": 5.666666666666667e-05, + "loss": 2.95, "step": 17 }, { "epoch": 0.02, - "learning_rate": 0.00015, - "loss": 2.5594, + "learning_rate": 6e-05, + "loss": 2.9196, "step": 18 }, { "epoch": 0.02, - "learning_rate": 0.00015833333333333332, - "loss": 2.4738, + "learning_rate": 6.333333333333333e-05, + "loss": 2.8583, "step": 19 }, { "epoch": 0.02, - "learning_rate": 0.00016666666666666666, - "loss": 2.4792, + "learning_rate": 6.666666666666667e-05, + "loss": 2.8926, "step": 20 }, { "epoch": 0.02, - "learning_rate": 0.000175, - "loss": 2.4097, + "learning_rate": 7e-05, + "loss": 2.8734, "step": 21 }, { "epoch": 0.02, - "learning_rate": 0.00018333333333333334, - "loss": 2.3469, + "learning_rate": 7.333333333333333e-05, + "loss": 2.8405, "step": 22 }, { "epoch": 0.03, - "learning_rate": 0.00019166666666666667, - "loss": 2.2218, + "learning_rate": 7.666666666666667e-05, + "loss": 2.7254, "step": 23 }, { "epoch": 0.03, - "learning_rate": 0.0002, - "loss": 2.1477, + "learning_rate": 8e-05, + "loss": 2.6502, "step": 24 }, { "epoch": 0.03, - "learning_rate": 0.00020833333333333335, - "loss": 2.0773, + "learning_rate": 8.333333333333334e-05, + "loss": 2.6134, "step": 25 }, { "epoch": 0.03, - "learning_rate": 0.00021666666666666668, - "loss": 2.0249, + "learning_rate": 8.666666666666667e-05, + "loss": 2.5713, "step": 26 }, { "epoch": 0.03, - "learning_rate": 0.00022500000000000002, - "loss": 1.913, + "learning_rate": 9e-05, + "loss": 2.5316, "step": 27 }, { "epoch": 0.03, - "learning_rate": 0.00023333333333333333, - "loss": 1.8384, + "learning_rate": 9.333333333333334e-05, + "loss": 2.4645, "step": 28 }, { "epoch": 0.03, - "learning_rate": 0.00024166666666666667, - "loss": 1.7988, + "learning_rate": 9.666666666666667e-05, + "loss": 2.4353, "step": 29 }, { "epoch": 0.03, - "learning_rate": 0.00025, - "loss": 1.8328, + "learning_rate": 0.0001, + "loss": 2.4353, "step": 30 }, { "epoch": 0.03, - "learning_rate": 0.00025833333333333334, - "loss": 1.7341, + "learning_rate": 0.00010333333333333334, + "loss": 2.3508, "step": 31 }, { "epoch": 0.03, - "learning_rate": 0.0002666666666666667, - "loss": 1.7436, + "learning_rate": 0.00010666666666666667, + "loss": 2.324, "step": 32 }, { "epoch": 0.04, - "learning_rate": 0.000275, - "loss": 1.6803, + "learning_rate": 0.00011000000000000002, + "loss": 2.2213, "step": 33 }, { "epoch": 0.04, - "learning_rate": 0.00028333333333333335, - "loss": 1.6908, + "learning_rate": 0.00011333333333333334, + "loss": 2.2284, "step": 34 }, { "epoch": 0.04, - "learning_rate": 0.0002916666666666667, - "loss": 1.672, + "learning_rate": 0.00011666666666666668, + "loss": 2.1722, "step": 35 }, { "epoch": 0.04, - "learning_rate": 0.0003, - "loss": 1.6964, + "learning_rate": 0.00012, + "loss": 2.1838, "step": 36 }, { "epoch": 0.04, - "learning_rate": 0.00030833333333333337, - "loss": 1.6025, + "learning_rate": 0.00012333333333333334, + "loss": 2.0307, "step": 37 }, { "epoch": 0.04, - "learning_rate": 0.00031666666666666665, - "loss": 1.6053, + "learning_rate": 0.00012666666666666666, + "loss": 1.9839, "step": 38 }, { "epoch": 0.04, - "learning_rate": 0.00032500000000000004, - "loss": 1.5432, + "learning_rate": 0.00013000000000000002, + "loss": 1.8751, "step": 39 }, { "epoch": 0.04, - "learning_rate": 0.0003333333333333333, - "loss": 1.5595, + "learning_rate": 0.00013333333333333334, + "loss": 1.8539, "step": 40 }, { "epoch": 0.04, - "learning_rate": 0.00034166666666666666, - "loss": 1.5724, + "learning_rate": 0.00013666666666666666, + "loss": 1.8668, "step": 41 }, { "epoch": 0.05, - "learning_rate": 0.00035, - "loss": 1.5366, + "learning_rate": 0.00014, + "loss": 1.8033, "step": 42 }, { "epoch": 0.05, - "learning_rate": 0.00035833333333333333, - "loss": 1.5278, + "learning_rate": 0.00014333333333333334, + "loss": 1.7899, "step": 43 }, { "epoch": 0.05, - "learning_rate": 0.00036666666666666667, - "loss": 1.5732, + "learning_rate": 0.00014666666666666666, + "loss": 1.813, "step": 44 }, { "epoch": 0.05, - "learning_rate": 0.000375, - "loss": 1.5424, + "learning_rate": 0.00015000000000000001, + "loss": 1.7906, "step": 45 }, { "epoch": 0.05, - "learning_rate": 0.00038333333333333334, - "loss": 1.496, + "learning_rate": 0.00015333333333333334, + "loss": 1.7245, "step": 46 }, { "epoch": 0.05, - "learning_rate": 0.0003916666666666667, - "loss": 1.4984, + "learning_rate": 0.00015666666666666666, + "loss": 1.6904, "step": 47 }, { "epoch": 0.05, - "learning_rate": 0.0004, - "loss": 1.4642, + "learning_rate": 0.00016, + "loss": 1.6549, "step": 48 }, { "epoch": 0.05, - "learning_rate": 0.00040833333333333336, - "loss": 1.4562, + "learning_rate": 0.00016333333333333334, + "loss": 1.632, "step": 49 }, { "epoch": 0.05, - "learning_rate": 0.0004166666666666667, - "loss": 1.5446, + "learning_rate": 0.0001666666666666667, + "loss": 1.7171, "step": 50 }, { "epoch": 0.06, - "learning_rate": 0.000425, - "loss": 1.4786, + "learning_rate": 0.00017, + "loss": 1.6282, "step": 51 }, { "epoch": 0.06, - "learning_rate": 0.00043333333333333337, - "loss": 1.4949, + "learning_rate": 0.00017333333333333334, + "loss": 1.6302, "step": 52 }, { "epoch": 0.06, - "learning_rate": 0.00044166666666666665, - "loss": 1.4503, + "learning_rate": 0.00017666666666666666, + "loss": 1.5987, "step": 53 }, { "epoch": 0.06, - "learning_rate": 0.00045000000000000004, - "loss": 1.501, + "learning_rate": 0.00018, + "loss": 1.6402, "step": 54 }, { "epoch": 0.06, - "learning_rate": 0.0004583333333333333, - "loss": 1.4322, + "learning_rate": 0.00018333333333333334, + "loss": 1.567, "step": 55 }, { "epoch": 0.06, - "learning_rate": 0.00046666666666666666, - "loss": 1.3982, + "learning_rate": 0.0001866666666666667, + "loss": 1.5274, "step": 56 }, { "epoch": 0.06, - "learning_rate": 0.000475, - "loss": 1.4274, + "learning_rate": 0.00019, + "loss": 1.5425, "step": 57 }, { "epoch": 0.06, - "learning_rate": 0.00048333333333333334, - "loss": 1.4058, + "learning_rate": 0.00019333333333333333, + "loss": 1.5181, "step": 58 }, { "epoch": 0.06, - "learning_rate": 0.0004916666666666666, - "loss": 1.422, + "learning_rate": 0.00019666666666666666, + "loss": 1.5587, "step": 59 }, { "epoch": 0.07, - "learning_rate": 0.0005, - "loss": 1.4153, + "learning_rate": 0.0002, + "loss": 1.5245, "step": 60 }, { "epoch": 0.07, - "learning_rate": 0.0005, - "loss": 1.4575, + "learning_rate": 0.0002, + "loss": 1.5707, "step": 61 }, { "epoch": 0.07, - "learning_rate": 0.0005, - "loss": 1.4206, + "learning_rate": 0.0002, + "loss": 1.5388, "step": 62 }, { "epoch": 0.07, - "learning_rate": 0.0005, - "loss": 1.4459, + "learning_rate": 0.0002, + "loss": 1.5679, "step": 63 }, { "epoch": 0.07, - "learning_rate": 0.0005, - "loss": 1.4473, + "learning_rate": 0.0002, + "loss": 1.5507, "step": 64 }, { "epoch": 0.07, - "learning_rate": 0.0005, - "loss": 1.4369, + "learning_rate": 0.0002, + "loss": 1.554, "step": 65 }, { "epoch": 0.07, - "learning_rate": 0.0005, - "loss": 1.3791, + "learning_rate": 0.0002, + "loss": 1.4835, "step": 66 }, { "epoch": 0.07, - "learning_rate": 0.0005, - "loss": 1.4655, + "learning_rate": 0.0002, + "loss": 1.5693, "step": 67 }, { "epoch": 0.07, - "learning_rate": 0.0005, - "loss": 1.3951, + "learning_rate": 0.0002, + "loss": 1.5108, "step": 68 }, { "epoch": 0.08, - "learning_rate": 0.0005, - "loss": 1.3741, + "learning_rate": 0.0002, + "loss": 1.4769, "step": 69 }, { "epoch": 0.08, - "learning_rate": 0.0005, - "loss": 1.3824, + "learning_rate": 0.0002, + "loss": 1.5075, "step": 70 }, { "epoch": 0.08, - "learning_rate": 0.0005, - "loss": 1.3948, + "learning_rate": 0.0002, + "loss": 1.5132, "step": 71 }, { "epoch": 0.08, - "learning_rate": 0.0005, - "loss": 1.383, + "learning_rate": 0.0002, + "loss": 1.4959, "step": 72 }, { "epoch": 0.08, - "learning_rate": 0.0005, - "loss": 1.3417, + "learning_rate": 0.0002, + "loss": 1.4469, "step": 73 }, { "epoch": 0.08, - "learning_rate": 0.0005, - "loss": 1.3329, + "learning_rate": 0.0002, + "loss": 1.4231, "step": 74 }, { "epoch": 0.08, - "learning_rate": 0.0005, - "loss": 1.3993, + "learning_rate": 0.0002, + "loss": 1.5133, "step": 75 }, { "epoch": 0.08, - "learning_rate": 0.0005, - "loss": 1.3095, + "learning_rate": 0.0002, + "loss": 1.4163, "step": 76 }, { "epoch": 0.08, - "learning_rate": 0.0005, - "loss": 1.3457, + "learning_rate": 0.0002, + "loss": 1.4366, "step": 77 }, { "epoch": 0.08, - "learning_rate": 0.0005, - "loss": 1.3462, + "learning_rate": 0.0002, + "loss": 1.4637, "step": 78 }, { "epoch": 0.09, - "learning_rate": 0.0005, - "loss": 1.3713, + "learning_rate": 0.0002, + "loss": 1.4901, "step": 79 }, { "epoch": 0.09, - "learning_rate": 0.0005, - "loss": 1.4215, + "learning_rate": 0.0002, + "loss": 1.5124, "step": 80 }, { "epoch": 0.09, - "learning_rate": 0.0005, - "loss": 1.3378, + "learning_rate": 0.0002, + "loss": 1.4442, "step": 81 }, { "epoch": 0.09, - "learning_rate": 0.0005, - "loss": 1.3221, + "learning_rate": 0.0002, + "loss": 1.43, "step": 82 }, { "epoch": 0.09, - "learning_rate": 0.0005, - "loss": 1.4079, + "learning_rate": 0.0002, + "loss": 1.5116, "step": 83 }, { "epoch": 0.09, - "learning_rate": 0.0005, - "loss": 1.3724, + "learning_rate": 0.0002, + "loss": 1.4918, "step": 84 }, { "epoch": 0.09, - "learning_rate": 0.0005, - "loss": 1.3094, + "learning_rate": 0.0002, + "loss": 1.4134, "step": 85 }, { "epoch": 0.09, - "learning_rate": 0.0005, - "loss": 1.2731, + "learning_rate": 0.0002, + "loss": 1.3838, "step": 86 }, { "epoch": 0.09, - "learning_rate": 0.0005, - "loss": 1.2718, + "learning_rate": 0.0002, + "loss": 1.3964, "step": 87 }, { "epoch": 0.1, - "learning_rate": 0.0005, - "loss": 1.2988, + "learning_rate": 0.0002, + "loss": 1.4231, "step": 88 }, { "epoch": 0.1, - "learning_rate": 0.0005, - "loss": 1.316, + "learning_rate": 0.0002, + "loss": 1.4364, "step": 89 }, { "epoch": 0.1, - "learning_rate": 0.0005, - "loss": 1.2531, + "learning_rate": 0.0002, + "loss": 1.3788, "step": 90 }, { "epoch": 0.1, - "learning_rate": 0.0005, - "loss": 1.3109, + "learning_rate": 0.0002, + "loss": 1.4249, "step": 91 }, { "epoch": 0.1, - "learning_rate": 0.0005, - "loss": 1.3246, + "learning_rate": 0.0002, + "loss": 1.4522, "step": 92 }, { "epoch": 0.1, - "learning_rate": 0.0005, - "loss": 1.3568, + "learning_rate": 0.0002, + "loss": 1.483, "step": 93 }, { "epoch": 0.1, - "learning_rate": 0.0005, - "loss": 1.2643, + "learning_rate": 0.0002, + "loss": 1.3924, "step": 94 }, { "epoch": 0.1, - "learning_rate": 0.0005, - "loss": 1.292, + "learning_rate": 0.0002, + "loss": 1.4024, "step": 95 }, { "epoch": 0.1, - "learning_rate": 0.0005, - "loss": 1.2677, + "learning_rate": 0.0002, + "loss": 1.3849, "step": 96 }, { "epoch": 0.11, - "learning_rate": 0.0005, - "loss": 1.2676, + "learning_rate": 0.0002, + "loss": 1.3796, "step": 97 }, { "epoch": 0.11, - "learning_rate": 0.0005, - "loss": 1.2829, + "learning_rate": 0.0002, + "loss": 1.4111, "step": 98 }, { "epoch": 0.11, - "learning_rate": 0.0005, - "loss": 1.2874, + "learning_rate": 0.0002, + "loss": 1.4173, "step": 99 }, { "epoch": 0.11, - "learning_rate": 0.0005, - "loss": 1.233, + "learning_rate": 0.0002, + "loss": 1.3514, "step": 100 }, { "epoch": 0.11, - "learning_rate": 0.0005, - "loss": 1.3235, + "learning_rate": 0.0002, + "loss": 1.4473, "step": 101 }, { "epoch": 0.11, - "learning_rate": 0.0005, - "loss": 1.2773, + "learning_rate": 0.0002, + "loss": 1.3952, "step": 102 }, { "epoch": 0.11, - "learning_rate": 0.0005, - "loss": 1.2528, + "learning_rate": 0.0002, + "loss": 1.3597, "step": 103 }, { "epoch": 0.11, - "learning_rate": 0.0005, - "loss": 1.2758, + "learning_rate": 0.0002, + "loss": 1.3837, "step": 104 }, { "epoch": 0.11, - "learning_rate": 0.0005, - "loss": 1.241, + "learning_rate": 0.0002, + "loss": 1.3653, "step": 105 }, { "epoch": 0.12, - "learning_rate": 0.0005, - "loss": 1.2566, + "learning_rate": 0.0002, + "loss": 1.3871, "step": 106 }, { "epoch": 0.12, - "learning_rate": 0.0005, - "loss": 1.2811, + "learning_rate": 0.0002, + "loss": 1.4003, "step": 107 }, { "epoch": 0.12, - "learning_rate": 0.0005, - "loss": 1.2691, + "learning_rate": 0.0002, + "loss": 1.397, "step": 108 }, { "epoch": 0.12, - "learning_rate": 0.0005, - "loss": 1.2337, + "learning_rate": 0.0002, + "loss": 1.3492, "step": 109 }, { "epoch": 0.12, - "learning_rate": 0.0005, - "loss": 1.2259, + "learning_rate": 0.0002, + "loss": 1.3507, "step": 110 }, { "epoch": 0.12, - "learning_rate": 0.0005, - "loss": 1.3066, + "learning_rate": 0.0002, + "loss": 1.4189, "step": 111 }, { "epoch": 0.12, - "learning_rate": 0.0005, - "loss": 1.297, + "learning_rate": 0.0002, + "loss": 1.4228, "step": 112 }, { "epoch": 0.12, - "learning_rate": 0.0005, - "loss": 1.239, + "learning_rate": 0.0002, + "loss": 1.3476, "step": 113 }, { "epoch": 0.12, - "learning_rate": 0.0005, - "loss": 1.2429, + "learning_rate": 0.0002, + "loss": 1.361, "step": 114 }, { "epoch": 0.13, - "learning_rate": 0.0005, - "loss": 1.2376, + "learning_rate": 0.0002, + "loss": 1.3573, "step": 115 }, { "epoch": 0.13, - "learning_rate": 0.0005, - "loss": 1.2336, + "learning_rate": 0.0002, + "loss": 1.3665, "step": 116 }, { "epoch": 0.13, - "learning_rate": 0.0005, - "loss": 1.2281, + "learning_rate": 0.0002, + "loss": 1.3573, "step": 117 }, { "epoch": 0.13, - "learning_rate": 0.0005, - "loss": 1.2077, + "learning_rate": 0.0002, + "loss": 1.3186, "step": 118 }, { "epoch": 0.13, - "learning_rate": 0.0005, - "loss": 1.2819, + "learning_rate": 0.0002, + "loss": 1.3983, "step": 119 }, { "epoch": 0.13, - "learning_rate": 0.0005, - "loss": 1.2342, + "learning_rate": 0.0002, + "loss": 1.3565, "step": 120 }, { "epoch": 0.13, - "learning_rate": 0.0005, - "loss": 1.2229, + "learning_rate": 0.0002, + "loss": 1.3707, "step": 121 }, { "epoch": 0.13, - "learning_rate": 0.0005, - "loss": 1.2834, + "learning_rate": 0.0002, + "loss": 1.4164, "step": 122 }, { "epoch": 0.13, - "learning_rate": 0.0005, - "loss": 1.2088, + "learning_rate": 0.0002, + "loss": 1.3117, "step": 123 }, { "epoch": 0.13, - "learning_rate": 0.0005, - "loss": 1.2204, + "learning_rate": 0.0002, + "loss": 1.3496, "step": 124 }, { "epoch": 0.14, - "learning_rate": 0.0005, - "loss": 1.182, + "learning_rate": 0.0002, + "loss": 1.2973, "step": 125 }, { "epoch": 0.14, - "learning_rate": 0.0005, - "loss": 1.1772, + "learning_rate": 0.0002, + "loss": 1.3079, "step": 126 }, { "epoch": 0.14, - "learning_rate": 0.0005, - "loss": 1.181, + "learning_rate": 0.0002, + "loss": 1.301, "step": 127 }, { "epoch": 0.14, - "learning_rate": 0.0005, - "loss": 1.2731, + "learning_rate": 0.0002, + "loss": 1.3964, "step": 128 }, { "epoch": 0.14, - "learning_rate": 0.0005, - "loss": 1.2672, + "learning_rate": 0.0002, + "loss": 1.38, "step": 129 }, { "epoch": 0.14, - "learning_rate": 0.0005, - "loss": 1.238, + "learning_rate": 0.0002, + "loss": 1.3465, "step": 130 }, { "epoch": 0.14, - "learning_rate": 0.0005, - "loss": 1.2286, + "learning_rate": 0.0002, + "loss": 1.3436, "step": 131 }, { "epoch": 0.14, - "learning_rate": 0.0005, - "loss": 1.1899, + "learning_rate": 0.0002, + "loss": 1.2999, "step": 132 }, { "epoch": 0.14, - "learning_rate": 0.0005, - "loss": 1.237, + "learning_rate": 0.0002, + "loss": 1.3506, "step": 133 }, { "epoch": 0.15, - "learning_rate": 0.0005, - "loss": 1.1862, + "learning_rate": 0.0002, + "loss": 1.2978, "step": 134 }, { "epoch": 0.15, - "learning_rate": 0.0005, - "loss": 1.2598, + "learning_rate": 0.0002, + "loss": 1.3489, "step": 135 }, { "epoch": 0.15, - "learning_rate": 0.0005, - "loss": 1.301, + "learning_rate": 0.0002, + "loss": 1.412, "step": 136 }, { "epoch": 0.15, - "learning_rate": 0.0005, - "loss": 1.2209, + "learning_rate": 0.0002, + "loss": 1.3365, "step": 137 }, { "epoch": 0.15, - "learning_rate": 0.0005, - "loss": 1.2473, + "learning_rate": 0.0002, + "loss": 1.3499, "step": 138 }, { "epoch": 0.15, - "learning_rate": 0.0005, - "loss": 1.1937, + "learning_rate": 0.0002, + "loss": 1.2941, "step": 139 }, { "epoch": 0.15, - "learning_rate": 0.0005, - "loss": 1.1837, + "learning_rate": 0.0002, + "loss": 1.3032, "step": 140 }, { "epoch": 0.15, - "learning_rate": 0.0005, - "loss": 1.2567, + "learning_rate": 0.0002, + "loss": 1.3647, "step": 141 }, { "epoch": 0.15, - "learning_rate": 0.0005, - "loss": 1.2951, + "learning_rate": 0.0002, + "loss": 1.3943, "step": 142 }, { "epoch": 0.16, - "learning_rate": 0.0005, - "loss": 1.2081, + "learning_rate": 0.0002, + "loss": 1.3021, "step": 143 }, { "epoch": 0.16, - "learning_rate": 0.0005, - "loss": 1.24, + "learning_rate": 0.0002, + "loss": 1.3362, "step": 144 }, { "epoch": 0.16, - "learning_rate": 0.0005, - "loss": 1.2182, + "learning_rate": 0.0002, + "loss": 1.3304, "step": 145 }, { "epoch": 0.16, - "learning_rate": 0.0005, - "loss": 1.1947, + "learning_rate": 0.0002, + "loss": 1.2877, "step": 146 }, { "epoch": 0.16, - "learning_rate": 0.0005, - "loss": 1.2134, + "learning_rate": 0.0002, + "loss": 1.3201, "step": 147 }, { "epoch": 0.16, - "learning_rate": 0.0005, - "loss": 1.1861, + "learning_rate": 0.0002, + "loss": 1.2843, "step": 148 }, { "epoch": 0.16, - "learning_rate": 0.0005, - "loss": 1.2662, + "learning_rate": 0.0002, + "loss": 1.3648, "step": 149 }, { "epoch": 0.16, - "learning_rate": 0.0005, - "loss": 1.1699, + "learning_rate": 0.0002, + "loss": 1.2658, "step": 150 }, { "epoch": 0.16, - "learning_rate": 0.0005, - "loss": 1.1974, + "learning_rate": 0.0002, + "loss": 1.2921, "step": 151 }, { "epoch": 0.17, - "learning_rate": 0.0005, - "loss": 1.2301, + "learning_rate": 0.0002, + "loss": 1.3213, "step": 152 }, { "epoch": 0.17, - "learning_rate": 0.0005, - "loss": 1.1921, + "learning_rate": 0.0002, + "loss": 1.2659, "step": 153 }, { "epoch": 0.17, - "learning_rate": 0.0005, - "loss": 1.1531, + "learning_rate": 0.0002, + "loss": 1.2435, "step": 154 }, { "epoch": 0.17, - "learning_rate": 0.0005, - "loss": 1.1716, + "learning_rate": 0.0002, + "loss": 1.2622, "step": 155 }, { "epoch": 0.17, - "learning_rate": 0.0005, - "loss": 1.2362, + "learning_rate": 0.0002, + "loss": 1.3431, "step": 156 }, { "epoch": 0.17, - "learning_rate": 0.0005, - "loss": 1.1793, + "learning_rate": 0.0002, + "loss": 1.276, "step": 157 }, { "epoch": 0.17, - "learning_rate": 0.0005, - "loss": 1.2342, + "learning_rate": 0.0002, + "loss": 1.3261, "step": 158 }, { "epoch": 0.17, - "learning_rate": 0.0005, - "loss": 1.2101, + "learning_rate": 0.0002, + "loss": 1.3117, "step": 159 }, { "epoch": 0.17, - "learning_rate": 0.0005, - "loss": 1.1724, + "learning_rate": 0.0002, + "loss": 1.2541, "step": 160 }, { "epoch": 0.18, - "learning_rate": 0.0005, - "loss": 1.1928, + "learning_rate": 0.0002, + "loss": 1.2908, "step": 161 }, { "epoch": 0.18, - "learning_rate": 0.0005, - "loss": 1.2171, + "learning_rate": 0.0002, + "loss": 1.3138, "step": 162 }, { "epoch": 0.18, - "learning_rate": 0.0005, - "loss": 1.2294, + "learning_rate": 0.0002, + "loss": 1.3336, "step": 163 }, { "epoch": 0.18, - "learning_rate": 0.0005, - "loss": 1.2505, + "learning_rate": 0.0002, + "loss": 1.34, "step": 164 }, { "epoch": 0.18, - "learning_rate": 0.0005, - "loss": 1.2111, + "learning_rate": 0.0002, + "loss": 1.2834, "step": 165 }, { "epoch": 0.18, - "learning_rate": 0.0005, - "loss": 1.2237, + "learning_rate": 0.0002, + "loss": 1.3009, "step": 166 }, { "epoch": 0.18, - "learning_rate": 0.0005, - "loss": 1.1797, + "learning_rate": 0.0002, + "loss": 1.2749, "step": 167 }, { "epoch": 0.18, - "learning_rate": 0.0005, - "loss": 1.1526, + "learning_rate": 0.0002, + "loss": 1.2358, "step": 168 }, { "epoch": 0.18, - "learning_rate": 0.0005, - "loss": 1.1983, + "learning_rate": 0.0002, + "loss": 1.2746, "step": 169 }, { "epoch": 0.18, - "learning_rate": 0.0005, - "loss": 1.1987, + "learning_rate": 0.0002, + "loss": 1.2692, "step": 170 }, { "epoch": 0.19, - "learning_rate": 0.0005, - "loss": 1.2492, + "learning_rate": 0.0002, + "loss": 1.3431, "step": 171 }, { "epoch": 0.19, - "learning_rate": 0.0005, - "loss": 1.15, + "learning_rate": 0.0002, + "loss": 1.2373, "step": 172 }, { "epoch": 0.19, - "learning_rate": 0.0005, - "loss": 1.1369, + "learning_rate": 0.0002, + "loss": 1.2097, "step": 173 }, { "epoch": 0.19, - "learning_rate": 0.0005, - "loss": 1.203, + "learning_rate": 0.0002, + "loss": 1.2712, "step": 174 }, { "epoch": 0.19, - "learning_rate": 0.0005, - "loss": 1.2171, + "learning_rate": 0.0002, + "loss": 1.2907, "step": 175 }, { "epoch": 0.19, - "learning_rate": 0.0005, - "loss": 1.2159, + "learning_rate": 0.0002, + "loss": 1.2868, "step": 176 }, { "epoch": 0.19, - "learning_rate": 0.0005, - "loss": 1.2028, + "learning_rate": 0.0002, + "loss": 1.2788, "step": 177 }, { "epoch": 0.19, - "learning_rate": 0.0005, - "loss": 1.1498, + "learning_rate": 0.0002, + "loss": 1.2158, "step": 178 }, { "epoch": 0.19, - "learning_rate": 0.0005, - "loss": 1.1855, + "learning_rate": 0.0002, + "loss": 1.2474, "step": 179 }, { "epoch": 0.2, - "learning_rate": 0.0005, - "loss": 1.2133, + "learning_rate": 0.0002, + "loss": 1.2899, "step": 180 }, { "epoch": 0.2, - "learning_rate": 0.0005, - "loss": 1.1798, + "learning_rate": 0.0002, + "loss": 1.2421, "step": 181 }, { "epoch": 0.2, - "learning_rate": 0.0005, - "loss": 1.1597, + "learning_rate": 0.0002, + "loss": 1.232, "step": 182 }, { "epoch": 0.2, - "learning_rate": 0.0005, - "loss": 1.1383, + "learning_rate": 0.0002, + "loss": 1.2007, "step": 183 }, { "epoch": 0.2, - "learning_rate": 0.0005, - "loss": 1.2074, + "learning_rate": 0.0002, + "loss": 1.2867, "step": 184 }, { "epoch": 0.2, - "learning_rate": 0.0005, - "loss": 1.1202, + "learning_rate": 0.0002, + "loss": 1.1981, "step": 185 }, { "epoch": 0.2, - "learning_rate": 0.0005, - "loss": 1.1667, + "learning_rate": 0.0002, + "loss": 1.2314, "step": 186 }, { "epoch": 0.2, - "learning_rate": 0.0005, - "loss": 1.2343, + "learning_rate": 0.0002, + "loss": 1.3119, "step": 187 }, { "epoch": 0.2, - "learning_rate": 0.0005, - "loss": 1.1639, + "learning_rate": 0.0002, + "loss": 1.2293, "step": 188 }, { "epoch": 0.21, - "learning_rate": 0.0005, - "loss": 1.2114, + "learning_rate": 0.0002, + "loss": 1.2781, "step": 189 }, { "epoch": 0.21, - "learning_rate": 0.0005, - "loss": 1.1465, + "learning_rate": 0.0002, + "loss": 1.218, "step": 190 }, { "epoch": 0.21, - "learning_rate": 0.0005, - "loss": 1.1341, + "learning_rate": 0.0002, + "loss": 1.2044, "step": 191 }, { "epoch": 0.21, - "learning_rate": 0.0005, - "loss": 1.2378, + "learning_rate": 0.0002, + "loss": 1.3069, "step": 192 }, { "epoch": 0.21, - "learning_rate": 0.0005, - "loss": 1.1704, + "learning_rate": 0.0002, + "loss": 1.2219, "step": 193 }, { "epoch": 0.21, - "learning_rate": 0.0005, - "loss": 1.1667, + "learning_rate": 0.0002, + "loss": 1.2309, "step": 194 }, { "epoch": 0.21, - "learning_rate": 0.0005, - "loss": 1.1565, + "learning_rate": 0.0002, + "loss": 1.2016, "step": 195 }, { "epoch": 0.21, - "learning_rate": 0.0005, - "loss": 1.1623, + "learning_rate": 0.0002, + "loss": 1.2298, "step": 196 }, { "epoch": 0.21, - "learning_rate": 0.0005, - "loss": 1.1766, + "learning_rate": 0.0002, + "loss": 1.2214, "step": 197 }, { "epoch": 0.22, - "learning_rate": 0.0005, - "loss": 1.2176, + "learning_rate": 0.0002, + "loss": 1.2799, "step": 198 }, { "epoch": 0.22, - "learning_rate": 0.0005, - "loss": 1.1777, + "learning_rate": 0.0002, + "loss": 1.2336, "step": 199 }, { "epoch": 0.22, - "learning_rate": 0.0005, - "loss": 1.2673, + "learning_rate": 0.0002, + "loss": 1.3338, "step": 200 }, { "epoch": 0.22, - "learning_rate": 0.0005, - "loss": 1.2108, + "learning_rate": 0.0002, + "loss": 1.2936, "step": 201 }, { "epoch": 0.22, - "learning_rate": 0.0005, - "loss": 1.1591, + "learning_rate": 0.0002, + "loss": 1.2256, "step": 202 }, { "epoch": 0.22, - "learning_rate": 0.0005, - "loss": 1.1441, + "learning_rate": 0.0002, + "loss": 1.1905, "step": 203 }, { "epoch": 0.22, - "learning_rate": 0.0005, - "loss": 1.2305, + "learning_rate": 0.0002, + "loss": 1.2922, "step": 204 }, { "epoch": 0.22, - "learning_rate": 0.0005, - "loss": 1.1417, + "learning_rate": 0.0002, + "loss": 1.1935, "step": 205 }, { "epoch": 0.22, - "learning_rate": 0.0005, - "loss": 1.1638, + "learning_rate": 0.0002, + "loss": 1.2179, "step": 206 }, { "epoch": 0.23, - "learning_rate": 0.0005, - "loss": 1.2092, + "learning_rate": 0.0002, + "loss": 1.2739, "step": 207 }, { "epoch": 0.23, - "learning_rate": 0.0005, - "loss": 1.1597, + "learning_rate": 0.0002, + "loss": 1.2188, "step": 208 }, { "epoch": 0.23, - "learning_rate": 0.0005, - "loss": 1.1423, + "learning_rate": 0.0002, + "loss": 1.1991, "step": 209 }, { "epoch": 0.23, - "learning_rate": 0.0005, - "loss": 1.1304, + "learning_rate": 0.0002, + "loss": 1.1834, "step": 210 }, { "epoch": 0.23, - "learning_rate": 0.0005, - "loss": 1.1535, + "learning_rate": 0.0002, + "loss": 1.2065, "step": 211 }, { "epoch": 0.23, - "learning_rate": 0.0005, - "loss": 1.1886, + "learning_rate": 0.0002, + "loss": 1.2591, "step": 212 }, { "epoch": 0.23, - "learning_rate": 0.0005, - "loss": 1.1739, + "learning_rate": 0.0002, + "loss": 1.2224, "step": 213 }, { "epoch": 0.23, - "learning_rate": 0.0005, - "loss": 1.1597, + "learning_rate": 0.0002, + "loss": 1.2152, "step": 214 }, { "epoch": 0.23, - "learning_rate": 0.0005, - "loss": 1.2508, + "learning_rate": 0.0002, + "loss": 1.3144, "step": 215 }, { "epoch": 0.24, - "learning_rate": 0.0005, - "loss": 1.2121, + "learning_rate": 0.0002, + "loss": 1.2636, "step": 216 }, { "epoch": 0.24, - "learning_rate": 0.0005, - "loss": 1.1698, + "learning_rate": 0.0002, + "loss": 1.2209, "step": 217 }, { "epoch": 0.24, - "learning_rate": 0.0005, - "loss": 1.1392, + "learning_rate": 0.0002, + "loss": 1.1843, "step": 218 }, { "epoch": 0.24, - "learning_rate": 0.0005, - "loss": 1.1906, + "learning_rate": 0.0002, + "loss": 1.2511, "step": 219 }, { "epoch": 0.24, - "learning_rate": 0.0005, - "loss": 1.1481, + "learning_rate": 0.0002, + "loss": 1.1989, "step": 220 }, { "epoch": 0.24, - "learning_rate": 0.0005, - "loss": 1.1912, + "learning_rate": 0.0002, + "loss": 1.2512, "step": 221 }, { "epoch": 0.24, - "learning_rate": 0.0005, - "loss": 1.1583, + "learning_rate": 0.0002, + "loss": 1.2077, "step": 222 }, { "epoch": 0.24, - "learning_rate": 0.0005, - "loss": 1.1819, + "learning_rate": 0.0002, + "loss": 1.2353, "step": 223 }, { "epoch": 0.24, - "learning_rate": 0.0005, - "loss": 1.1733, + "learning_rate": 0.0002, + "loss": 1.2118, "step": 224 }, { "epoch": 0.24, - "learning_rate": 0.0005, - "loss": 1.208, + "learning_rate": 0.0002, + "loss": 1.2557, "step": 225 }, { "epoch": 0.25, - "learning_rate": 0.0005, - "loss": 1.1297, + "learning_rate": 0.0002, + "loss": 1.1948, "step": 226 }, { "epoch": 0.25, - "learning_rate": 0.0005, - "loss": 1.1523, + "learning_rate": 0.0002, + "loss": 1.2082, "step": 227 }, { "epoch": 0.25, - "learning_rate": 0.0005, - "loss": 1.1804, + "learning_rate": 0.0002, + "loss": 1.2288, "step": 228 }, { "epoch": 0.25, - "learning_rate": 0.0005, - "loss": 1.1335, + "learning_rate": 0.0002, + "loss": 1.1733, "step": 229 }, { "epoch": 0.25, - "learning_rate": 0.0005, - "loss": 1.1825, + "learning_rate": 0.0002, + "loss": 1.2184, "step": 230 }, { "epoch": 0.25, - "learning_rate": 0.0005, - "loss": 1.1292, + "learning_rate": 0.0002, + "loss": 1.178, "step": 231 }, { "epoch": 0.25, - "learning_rate": 0.0005, - "loss": 1.1735, + "learning_rate": 0.0002, + "loss": 1.2214, "step": 232 }, { "epoch": 0.25, - "learning_rate": 0.0005, - "loss": 1.2123, + "learning_rate": 0.0002, + "loss": 1.2549, "step": 233 }, { "epoch": 0.25, - "learning_rate": 0.0005, - "loss": 1.1286, + "learning_rate": 0.0002, + "loss": 1.1806, "step": 234 }, { "epoch": 0.26, - "learning_rate": 0.0005, - "loss": 1.2323, + "learning_rate": 0.0002, + "loss": 1.2866, "step": 235 }, { "epoch": 0.26, - "learning_rate": 0.0005, - "loss": 1.1497, + "learning_rate": 0.0002, + "loss": 1.1899, "step": 236 }, { "epoch": 0.26, - "learning_rate": 0.0005, - "loss": 1.1598, + "learning_rate": 0.0002, + "loss": 1.2065, "step": 237 }, { "epoch": 0.26, - "learning_rate": 0.0005, - "loss": 1.2146, + "learning_rate": 0.0002, + "loss": 1.2593, "step": 238 }, { "epoch": 0.26, - "learning_rate": 0.0005, - "loss": 1.1907, + "learning_rate": 0.0002, + "loss": 1.2308, "step": 239 }, { "epoch": 0.26, - "learning_rate": 0.0005, - "loss": 1.1524, + "learning_rate": 0.0002, + "loss": 1.1992, "step": 240 }, { "epoch": 0.26, - "learning_rate": 0.0005, - "loss": 1.155, + "learning_rate": 0.0002, + "loss": 1.1966, "step": 241 }, { "epoch": 0.26, - "learning_rate": 0.0005, - "loss": 1.1665, + "learning_rate": 0.0002, + "loss": 1.2079, "step": 242 }, { "epoch": 0.26, - "learning_rate": 0.0005, - "loss": 1.2103, + "learning_rate": 0.0002, + "loss": 1.2653, "step": 243 }, { "epoch": 0.27, - "learning_rate": 0.0005, - "loss": 1.1455, + "learning_rate": 0.0002, + "loss": 1.1859, "step": 244 }, { "epoch": 0.27, - "learning_rate": 0.0005, - "loss": 1.1145, + "learning_rate": 0.0002, + "loss": 1.1464, "step": 245 }, { "epoch": 0.27, - "learning_rate": 0.0005, - "loss": 1.1516, + "learning_rate": 0.0002, + "loss": 1.1883, "step": 246 }, { "epoch": 0.27, - "learning_rate": 0.0005, - "loss": 1.1471, + "learning_rate": 0.0002, + "loss": 1.1843, "step": 247 }, { "epoch": 0.27, - "learning_rate": 0.0005, - "loss": 1.1938, + "learning_rate": 0.0002, + "loss": 1.2355, "step": 248 }, { "epoch": 0.27, - "learning_rate": 0.0005, - "loss": 1.1558, + "learning_rate": 0.0002, + "loss": 1.1958, "step": 249 }, { "epoch": 0.27, - "learning_rate": 0.0005, - "loss": 1.1639, + "learning_rate": 0.0002, + "loss": 1.2049, "step": 250 }, { "epoch": 0.27, - "learning_rate": 0.0005, - "loss": 1.1371, + "learning_rate": 0.0002, + "loss": 1.1816, "step": 251 }, { "epoch": 0.27, - "learning_rate": 0.0005, - "loss": 1.1195, + "learning_rate": 0.0002, + "loss": 1.1636, "step": 252 }, { "epoch": 0.28, - "learning_rate": 0.0005, - "loss": 1.1999, + "learning_rate": 0.0002, + "loss": 1.2519, "step": 253 }, { "epoch": 0.28, - "learning_rate": 0.0005, - "loss": 1.1647, + "learning_rate": 0.0002, + "loss": 1.2056, "step": 254 }, { "epoch": 0.28, - "learning_rate": 0.0005, - "loss": 1.1431, + "learning_rate": 0.0002, + "loss": 1.1782, "step": 255 }, { "epoch": 0.28, - "learning_rate": 0.0005, - "loss": 1.1219, + "learning_rate": 0.0002, + "loss": 1.1684, "step": 256 }, { "epoch": 0.28, - "learning_rate": 0.0005, - "loss": 1.1378, + "learning_rate": 0.0002, + "loss": 1.1818, "step": 257 }, { "epoch": 0.28, - "learning_rate": 0.0005, - "loss": 1.1689, + "learning_rate": 0.0002, + "loss": 1.2023, "step": 258 }, { "epoch": 0.28, - "learning_rate": 0.0005, - "loss": 1.1478, + "learning_rate": 0.0002, + "loss": 1.1809, "step": 259 }, { "epoch": 0.28, - "learning_rate": 0.0005, - "loss": 1.1591, + "learning_rate": 0.0002, + "loss": 1.2001, "step": 260 }, { "epoch": 0.28, - "learning_rate": 0.0005, - "loss": 1.1389, + "learning_rate": 0.0002, + "loss": 1.1647, "step": 261 }, { "epoch": 0.29, - "learning_rate": 0.0005, - "loss": 1.1746, + "learning_rate": 0.0002, + "loss": 1.2125, "step": 262 }, { "epoch": 0.29, - "learning_rate": 0.0005, - "loss": 1.1305, + "learning_rate": 0.0002, + "loss": 1.1664, "step": 263 }, { "epoch": 0.29, - "learning_rate": 0.0005, - "loss": 1.0773, + "learning_rate": 0.0002, + "loss": 1.1149, "step": 264 }, { "epoch": 0.29, - "learning_rate": 0.0005, - "loss": 1.1679, + "learning_rate": 0.0002, + "loss": 1.2082, "step": 265 }, { "epoch": 0.29, - "learning_rate": 0.0005, - "loss": 1.1133, + "learning_rate": 0.0002, + "loss": 1.1443, "step": 266 }, { "epoch": 0.29, - "learning_rate": 0.0005, - "loss": 1.1807, + "learning_rate": 0.0002, + "loss": 1.2177, "step": 267 }, { "epoch": 0.29, - "learning_rate": 0.0005, - "loss": 1.0736, + "learning_rate": 0.0002, + "loss": 1.1145, "step": 268 }, { "epoch": 0.29, - "learning_rate": 0.0005, - "loss": 1.1416, + "learning_rate": 0.0002, + "loss": 1.1897, "step": 269 }, { "epoch": 0.29, - "learning_rate": 0.0005, - "loss": 1.0843, + "learning_rate": 0.0002, + "loss": 1.1249, "step": 270 }, { "epoch": 0.29, - "learning_rate": 0.0005, - "loss": 1.2055, + "learning_rate": 0.0002, + "loss": 1.2332, "step": 271 }, { "epoch": 0.3, - "learning_rate": 0.0005, - "loss": 1.1401, + "learning_rate": 0.0002, + "loss": 1.1861, "step": 272 }, { "epoch": 0.3, - "learning_rate": 0.0005, - "loss": 1.1247, + "learning_rate": 0.0002, + "loss": 1.1661, "step": 273 }, { "epoch": 0.3, - "learning_rate": 0.0005, - "loss": 1.1633, + "learning_rate": 0.0002, + "loss": 1.2134, "step": 274 }, { "epoch": 0.3, - "learning_rate": 0.0005, - "loss": 1.1592, + "learning_rate": 0.0002, + "loss": 1.1959, "step": 275 }, { "epoch": 0.3, - "learning_rate": 0.0005, - "loss": 1.1481, + "learning_rate": 0.0002, + "loss": 1.1896, "step": 276 }, { "epoch": 0.3, - "learning_rate": 0.0005, - "loss": 1.1493, + "learning_rate": 0.0002, + "loss": 1.1781, "step": 277 }, { "epoch": 0.3, - "learning_rate": 0.0005, - "loss": 1.18, + "learning_rate": 0.0002, + "loss": 1.2184, "step": 278 }, { "epoch": 0.3, - "learning_rate": 0.0005, - "loss": 1.152, + "learning_rate": 0.0002, + "loss": 1.1735, "step": 279 }, { "epoch": 0.3, - "learning_rate": 0.0005, - "loss": 1.1175, + "learning_rate": 0.0002, + "loss": 1.1483, "step": 280 }, { "epoch": 0.31, - "learning_rate": 0.0005, - "loss": 1.1736, + "learning_rate": 0.0002, + "loss": 1.2048, "step": 281 }, { "epoch": 0.31, - "learning_rate": 0.0005, - "loss": 1.1598, + "learning_rate": 0.0002, + "loss": 1.1827, "step": 282 }, { "epoch": 0.31, - "learning_rate": 0.0005, - "loss": 1.1614, + "learning_rate": 0.0002, + "loss": 1.1962, "step": 283 }, { "epoch": 0.31, - "learning_rate": 0.0005, - "loss": 1.1112, + "learning_rate": 0.0002, + "loss": 1.135, "step": 284 }, { "epoch": 0.31, - "learning_rate": 0.0005, - "loss": 1.174, + "learning_rate": 0.0002, + "loss": 1.2087, "step": 285 }, { "epoch": 0.31, - "learning_rate": 0.0005, - "loss": 1.1545, + "learning_rate": 0.0002, + "loss": 1.1939, "step": 286 }, { "epoch": 0.31, - "learning_rate": 0.0005, - "loss": 1.1704, + "learning_rate": 0.0002, + "loss": 1.2154, "step": 287 }, { "epoch": 0.31, - "learning_rate": 0.0005, - "loss": 1.1427, + "learning_rate": 0.0002, + "loss": 1.1739, "step": 288 }, { "epoch": 0.31, - "learning_rate": 0.0005, - "loss": 1.0912, + "learning_rate": 0.0002, + "loss": 1.1183, "step": 289 }, { "epoch": 0.32, - "learning_rate": 0.0005, - "loss": 1.172, + "learning_rate": 0.0002, + "loss": 1.21, "step": 290 }, { "epoch": 0.32, - "learning_rate": 0.0005, - "loss": 1.1939, + "learning_rate": 0.0002, + "loss": 1.2178, "step": 291 }, { "epoch": 0.32, - "learning_rate": 0.0005, - "loss": 1.0911, + "learning_rate": 0.0002, + "loss": 1.1316, "step": 292 }, { "epoch": 0.32, - "learning_rate": 0.0005, - "loss": 1.106, + "learning_rate": 0.0002, + "loss": 1.1333, "step": 293 }, { "epoch": 0.32, - "learning_rate": 0.0005, - "loss": 1.1358, + "learning_rate": 0.0002, + "loss": 1.1642, "step": 294 }, { "epoch": 0.32, - "learning_rate": 0.0005, - "loss": 1.0833, + "learning_rate": 0.0002, + "loss": 1.1222, "step": 295 }, { "epoch": 0.32, - "learning_rate": 0.0005, - "loss": 1.1556, + "learning_rate": 0.0002, + "loss": 1.1885, "step": 296 }, { "epoch": 0.32, - "learning_rate": 0.0005, - "loss": 1.1171, + "learning_rate": 0.0002, + "loss": 1.1408, "step": 297 }, { "epoch": 0.32, - "learning_rate": 0.0005, - "loss": 1.1522, + "learning_rate": 0.0002, + "loss": 1.1883, "step": 298 }, { "epoch": 0.33, - "learning_rate": 0.0005, - "loss": 1.1619, + "learning_rate": 0.0002, + "loss": 1.1877, "step": 299 }, { "epoch": 0.33, - "learning_rate": 0.0005, - "loss": 1.1661, + "learning_rate": 0.0002, + "loss": 1.1871, "step": 300 }, { "epoch": 0.33, - "learning_rate": 0.0005, - "loss": 1.1426, + "learning_rate": 0.0002, + "loss": 1.1683, "step": 301 }, { "epoch": 0.33, - "learning_rate": 0.0005, - "loss": 1.1481, + "learning_rate": 0.0002, + "loss": 1.1613, "step": 302 }, { "epoch": 0.33, - "learning_rate": 0.0005, - "loss": 1.1153, + "learning_rate": 0.0002, + "loss": 1.1386, "step": 303 }, { "epoch": 0.33, - "learning_rate": 0.0005, - "loss": 1.1626, + "learning_rate": 0.0002, + "loss": 1.1946, "step": 304 }, { "epoch": 0.33, - "learning_rate": 0.0005, - "loss": 1.1958, + "learning_rate": 0.0002, + "loss": 1.2382, "step": 305 }, { "epoch": 0.33, - "learning_rate": 0.0005, - "loss": 1.1717, + "learning_rate": 0.0002, + "loss": 1.2042, "step": 306 }, { "epoch": 0.33, - "learning_rate": 0.0005, - "loss": 1.1187, + "learning_rate": 0.0002, + "loss": 1.1518, "step": 307 }, { "epoch": 0.34, - "learning_rate": 0.0005, - "loss": 1.0895, + "learning_rate": 0.0002, + "loss": 1.1244, "step": 308 }, { "epoch": 0.34, - "learning_rate": 0.0005, - "loss": 1.1251, + "learning_rate": 0.0002, + "loss": 1.155, "step": 309 }, { "epoch": 0.34, - "learning_rate": 0.0005, - "loss": 1.1967, + "learning_rate": 0.0002, + "loss": 1.2259, "step": 310 }, { "epoch": 0.34, - "learning_rate": 0.0005, - "loss": 1.1811, + "learning_rate": 0.0002, + "loss": 1.2078, "step": 311 }, { "epoch": 0.34, - "learning_rate": 0.0005, - "loss": 1.1392, + "learning_rate": 0.0002, + "loss": 1.1552, "step": 312 }, { "epoch": 0.34, - "learning_rate": 0.0005, - "loss": 1.1373, + "learning_rate": 0.0002, + "loss": 1.1575, "step": 313 }, { "epoch": 0.34, - "learning_rate": 0.0005, - "loss": 1.1145, + "learning_rate": 0.0002, + "loss": 1.1519, "step": 314 }, { "epoch": 0.34, - "learning_rate": 0.0005, - "loss": 1.1429, + "learning_rate": 0.0002, + "loss": 1.1716, "step": 315 }, { "epoch": 0.34, - "learning_rate": 0.0005, - "loss": 1.075, + "learning_rate": 0.0002, + "loss": 1.0983, "step": 316 }, { "epoch": 0.34, - "learning_rate": 0.0005, - "loss": 1.2638, + "learning_rate": 0.0002, + "loss": 1.298, "step": 317 }, { "epoch": 0.35, - "learning_rate": 0.0005, - "loss": 1.1076, + "learning_rate": 0.0002, + "loss": 1.1319, "step": 318 }, { "epoch": 0.35, - "learning_rate": 0.0005, - "loss": 1.0853, + "learning_rate": 0.0002, + "loss": 1.1091, "step": 319 }, { "epoch": 0.35, - "learning_rate": 0.0005, - "loss": 1.0948, + "learning_rate": 0.0002, + "loss": 1.1256, "step": 320 }, { "epoch": 0.35, - "learning_rate": 0.0005, - "loss": 1.1421, + "learning_rate": 0.0002, + "loss": 1.1845, "step": 321 }, { "epoch": 0.35, - "learning_rate": 0.0005, - "loss": 1.1067, + "learning_rate": 0.0002, + "loss": 1.134, "step": 322 }, { "epoch": 0.35, - "learning_rate": 0.0005, - "loss": 1.148, + "learning_rate": 0.0002, + "loss": 1.164, "step": 323 }, { "epoch": 0.35, - "learning_rate": 0.0005, - "loss": 1.1525, + "learning_rate": 0.0002, + "loss": 1.1835, "step": 324 }, { "epoch": 0.35, - "learning_rate": 0.0005, - "loss": 1.1467, + "learning_rate": 0.0002, + "loss": 1.1669, "step": 325 }, { "epoch": 0.35, - "learning_rate": 0.0005, - "loss": 1.1286, + "learning_rate": 0.0002, + "loss": 1.1618, "step": 326 }, { "epoch": 0.36, - "learning_rate": 0.0005, - "loss": 1.1707, + "learning_rate": 0.0002, + "loss": 1.1971, "step": 327 }, { "epoch": 0.36, - "learning_rate": 0.0005, - "loss": 1.1688, + "learning_rate": 0.0002, + "loss": 1.1995, "step": 328 }, { "epoch": 0.36, - "learning_rate": 0.0005, - "loss": 1.1618, + "learning_rate": 0.0002, + "loss": 1.2016, "step": 329 }, { "epoch": 0.36, - "learning_rate": 0.0005, - "loss": 1.1654, + "learning_rate": 0.0002, + "loss": 1.1848, "step": 330 }, { "epoch": 0.36, - "learning_rate": 0.0005, - "loss": 1.109, + "learning_rate": 0.0002, + "loss": 1.1407, "step": 331 }, { "epoch": 0.36, - "learning_rate": 0.0005, - "loss": 1.1761, + "learning_rate": 0.0002, + "loss": 1.2064, "step": 332 }, { "epoch": 0.36, - "learning_rate": 0.0005, - "loss": 1.1845, + "learning_rate": 0.0002, + "loss": 1.2122, "step": 333 }, { "epoch": 0.36, - "learning_rate": 0.0005, - "loss": 1.1326, + "learning_rate": 0.0002, + "loss": 1.1647, "step": 334 }, { "epoch": 0.36, - "learning_rate": 0.0005, - "loss": 1.1221, + "learning_rate": 0.0002, + "loss": 1.1429, "step": 335 }, { "epoch": 0.37, - "learning_rate": 0.0005, - "loss": 1.158, + "learning_rate": 0.0002, + "loss": 1.1903, "step": 336 }, { "epoch": 0.37, - "learning_rate": 0.0005, - "loss": 1.0982, + "learning_rate": 0.0002, + "loss": 1.1253, "step": 337 }, { "epoch": 0.37, - "learning_rate": 0.0005, - "loss": 1.1278, + "learning_rate": 0.0002, + "loss": 1.1484, "step": 338 }, { "epoch": 0.37, - "learning_rate": 0.0005, - "loss": 1.0982, + "learning_rate": 0.0002, + "loss": 1.1276, "step": 339 }, { "epoch": 0.37, - "learning_rate": 0.0005, - "loss": 1.118, + "learning_rate": 0.0002, + "loss": 1.1361, "step": 340 }, { "epoch": 0.37, - "learning_rate": 0.0005, - "loss": 1.1106, + "learning_rate": 0.0002, + "loss": 1.1304, "step": 341 }, { "epoch": 0.37, - "learning_rate": 0.0005, - "loss": 1.129, + "learning_rate": 0.0002, + "loss": 1.1565, "step": 342 }, { "epoch": 0.37, - "learning_rate": 0.0005, - "loss": 1.1754, + "learning_rate": 0.0002, + "loss": 1.1989, "step": 343 }, { "epoch": 0.37, - "learning_rate": 0.0005, - "loss": 1.1159, + "learning_rate": 0.0002, + "loss": 1.1375, "step": 344 }, { "epoch": 0.38, - "learning_rate": 0.0005, - "loss": 1.1448, + "learning_rate": 0.0002, + "loss": 1.1629, "step": 345 }, { "epoch": 0.38, - "learning_rate": 0.0005, - "loss": 1.1971, + "learning_rate": 0.0002, + "loss": 1.2277, "step": 346 }, { "epoch": 0.38, - "learning_rate": 0.0005, - "loss": 1.1513, + "learning_rate": 0.0002, + "loss": 1.1812, "step": 347 }, { "epoch": 0.38, - "learning_rate": 0.0005, - "loss": 1.1271, + "learning_rate": 0.0002, + "loss": 1.1547, "step": 348 }, { "epoch": 0.38, - "learning_rate": 0.0005, - "loss": 1.1378, + "learning_rate": 0.0002, + "loss": 1.165, "step": 349 }, { "epoch": 0.38, - "learning_rate": 0.0005, - "loss": 1.1518, + "learning_rate": 0.0002, + "loss": 1.1777, "step": 350 }, { "epoch": 0.38, - "learning_rate": 0.0005, - "loss": 1.1008, + "learning_rate": 0.0002, + "loss": 1.1205, "step": 351 }, { "epoch": 0.38, - "learning_rate": 0.0005, - "loss": 1.1205, + "learning_rate": 0.0002, + "loss": 1.1498, "step": 352 }, { "epoch": 0.38, - "learning_rate": 0.0005, - "loss": 1.1167, + "learning_rate": 0.0002, + "loss": 1.1497, "step": 353 }, { "epoch": 0.39, - "learning_rate": 0.0005, - "loss": 1.1011, + "learning_rate": 0.0002, + "loss": 1.1209, "step": 354 }, { "epoch": 0.39, - "learning_rate": 0.0005, - "loss": 1.1031, + "learning_rate": 0.0002, + "loss": 1.1398, "step": 355 }, { "epoch": 0.39, - "learning_rate": 0.0005, - "loss": 1.1194, + "learning_rate": 0.0002, + "loss": 1.1389, "step": 356 }, { "epoch": 0.39, - "learning_rate": 0.0005, - "loss": 1.1002, + "learning_rate": 0.0002, + "loss": 1.1242, "step": 357 }, { "epoch": 0.39, - "learning_rate": 0.0005, - "loss": 1.1156, + "learning_rate": 0.0002, + "loss": 1.1337, "step": 358 }, { "epoch": 0.39, - "learning_rate": 0.0005, - "loss": 1.0899, + "learning_rate": 0.0002, + "loss": 1.1146, "step": 359 }, { "epoch": 0.39, - "learning_rate": 0.0005, - "loss": 1.1718, + "learning_rate": 0.0002, + "loss": 1.193, "step": 360 }, { "epoch": 0.39, - "learning_rate": 0.0005, - "loss": 1.1062, + "learning_rate": 0.0002, + "loss": 1.131, "step": 361 }, { "epoch": 0.39, - "learning_rate": 0.0005, - "loss": 1.1618, + "learning_rate": 0.0002, + "loss": 1.1739, "step": 362 }, { "epoch": 0.39, - "learning_rate": 0.0005, - "loss": 1.1149, + "learning_rate": 0.0002, + "loss": 1.1375, "step": 363 }, { "epoch": 0.4, - "learning_rate": 0.0005, - "loss": 1.1086, + "learning_rate": 0.0002, + "loss": 1.136, "step": 364 }, { "epoch": 0.4, - "learning_rate": 0.0005, - "loss": 1.1069, + "learning_rate": 0.0002, + "loss": 1.1323, "step": 365 }, { "epoch": 0.4, - "learning_rate": 0.0005, - "loss": 1.0773, + "learning_rate": 0.0002, + "loss": 1.1089, "step": 366 }, { "epoch": 0.4, - "learning_rate": 0.0005, - "loss": 1.1392, + "learning_rate": 0.0002, + "loss": 1.169, "step": 367 }, { "epoch": 0.4, - "learning_rate": 0.0005, - "loss": 1.1487, + "learning_rate": 0.0002, + "loss": 1.1666, "step": 368 }, { "epoch": 0.4, - "learning_rate": 0.0005, - "loss": 1.1413, + "learning_rate": 0.0002, + "loss": 1.1671, "step": 369 }, { "epoch": 0.4, - "learning_rate": 0.0005, - "loss": 1.1242, + "learning_rate": 0.0002, + "loss": 1.1584, "step": 370 }, { "epoch": 0.4, - "learning_rate": 0.0005, - "loss": 1.0771, + "learning_rate": 0.0002, + "loss": 1.098, "step": 371 }, { "epoch": 0.4, - "learning_rate": 0.0005, - "loss": 1.137, + "learning_rate": 0.0002, + "loss": 1.16, "step": 372 }, { "epoch": 0.41, - "learning_rate": 0.0005, - "loss": 1.0734, + "learning_rate": 0.0002, + "loss": 1.0917, "step": 373 }, { "epoch": 0.41, - "learning_rate": 0.0005, - "loss": 1.1305, + "learning_rate": 0.0002, + "loss": 1.1553, "step": 374 }, { "epoch": 0.41, - "learning_rate": 0.0005, - "loss": 1.1169, + "learning_rate": 0.0002, + "loss": 1.1426, "step": 375 }, { "epoch": 0.41, - "learning_rate": 0.0005, - "loss": 1.1674, + "learning_rate": 0.0002, + "loss": 1.1864, "step": 376 }, { "epoch": 0.41, - "learning_rate": 0.0005, - "loss": 1.1103, + "learning_rate": 0.0002, + "loss": 1.1447, "step": 377 }, { "epoch": 0.41, - "learning_rate": 0.0005, - "loss": 1.1433, + "learning_rate": 0.0002, + "loss": 1.1627, "step": 378 }, { "epoch": 0.41, - "learning_rate": 0.0005, - "loss": 1.1344, + "learning_rate": 0.0002, + "loss": 1.1583, "step": 379 }, { "epoch": 0.41, - "learning_rate": 0.0005, - "loss": 1.123, - "step": 380 + "learning_rate": 0.0002, + "loss": 1.1406, + "step": 380 }, { "epoch": 0.41, - "learning_rate": 0.0005, - "loss": 1.1169, + "learning_rate": 0.0002, + "loss": 1.1462, "step": 381 }, { "epoch": 0.42, - "learning_rate": 0.0005, - "loss": 1.1508, + "learning_rate": 0.0002, + "loss": 1.1725, "step": 382 }, { "epoch": 0.42, - "learning_rate": 0.0005, - "loss": 1.1499, + "learning_rate": 0.0002, + "loss": 1.1701, "step": 383 }, { "epoch": 0.42, - "learning_rate": 0.0005, - "loss": 1.1916, + "learning_rate": 0.0002, + "loss": 1.2148, "step": 384 }, { "epoch": 0.42, - "learning_rate": 0.0005, - "loss": 1.1574, + "learning_rate": 0.0002, + "loss": 1.1841, "step": 385 }, { "epoch": 0.42, - "learning_rate": 0.0005, - "loss": 1.0429, + "learning_rate": 0.0002, + "loss": 1.0701, "step": 386 }, { "epoch": 0.42, - "learning_rate": 0.0005, - "loss": 1.0901, + "learning_rate": 0.0002, + "loss": 1.1142, "step": 387 }, { "epoch": 0.42, - "learning_rate": 0.0005, - "loss": 1.095, + "learning_rate": 0.0002, + "loss": 1.1239, "step": 388 }, { "epoch": 0.42, - "learning_rate": 0.0005, - "loss": 1.0851, + "learning_rate": 0.0002, + "loss": 1.1075, "step": 389 }, { "epoch": 0.42, - "learning_rate": 0.0005, - "loss": 1.083, + "learning_rate": 0.0002, + "loss": 1.1082, "step": 390 }, { "epoch": 0.43, - "learning_rate": 0.0005, - "loss": 1.1137, + "learning_rate": 0.0002, + "loss": 1.1422, "step": 391 }, { "epoch": 0.43, - "learning_rate": 0.0005, - "loss": 1.1507, + "learning_rate": 0.0002, + "loss": 1.1655, "step": 392 }, { "epoch": 0.43, - "learning_rate": 0.0005, - "loss": 1.1454, + "learning_rate": 0.0002, + "loss": 1.1647, "step": 393 }, { "epoch": 0.43, - "learning_rate": 0.0005, - "loss": 1.0831, + "learning_rate": 0.0002, + "loss": 1.108, "step": 394 }, { "epoch": 0.43, - "learning_rate": 0.0005, - "loss": 1.119, + "learning_rate": 0.0002, + "loss": 1.1344, "step": 395 }, { "epoch": 0.43, - "learning_rate": 0.0005, - "loss": 1.1277, + "learning_rate": 0.0002, + "loss": 1.1402, "step": 396 }, { "epoch": 0.43, - "learning_rate": 0.0005, - "loss": 1.156, + "learning_rate": 0.0002, + "loss": 1.1767, "step": 397 }, { "epoch": 0.43, - "learning_rate": 0.0005, - "loss": 1.1328, + "learning_rate": 0.0002, + "loss": 1.1519, "step": 398 }, { "epoch": 0.43, - "learning_rate": 0.0005, - "loss": 1.1374, + "learning_rate": 0.0002, + "loss": 1.1535, "step": 399 }, { "epoch": 0.44, - "learning_rate": 0.0005, - "loss": 1.1243, + "learning_rate": 0.0002, + "loss": 1.1452, "step": 400 }, { "epoch": 0.44, - "learning_rate": 0.0005, - "loss": 1.1207, + "learning_rate": 0.0002, + "loss": 1.1401, "step": 401 }, { "epoch": 0.44, - "learning_rate": 0.0005, - "loss": 1.1422, + "learning_rate": 0.0002, + "loss": 1.1728, "step": 402 }, { "epoch": 0.44, - "learning_rate": 0.0005, - "loss": 1.1359, + "learning_rate": 0.0002, + "loss": 1.1614, "step": 403 }, { "epoch": 0.44, - "learning_rate": 0.0005, - "loss": 1.1686, + "learning_rate": 0.0002, + "loss": 1.1934, "step": 404 }, { "epoch": 0.44, - "learning_rate": 0.0005, - "loss": 1.1114, + "learning_rate": 0.0002, + "loss": 1.13, "step": 405 }, { "epoch": 0.44, - "learning_rate": 0.0005, - "loss": 1.1648, + "learning_rate": 0.0002, + "loss": 1.1839, "step": 406 }, { "epoch": 0.44, - "learning_rate": 0.0005, - "loss": 1.0895, + "learning_rate": 0.0002, + "loss": 1.1132, "step": 407 }, { "epoch": 0.44, - "learning_rate": 0.0005, - "loss": 1.1065, + "learning_rate": 0.0002, + "loss": 1.1297, "step": 408 }, { "epoch": 0.45, - "learning_rate": 0.0005, - "loss": 1.1434, + "learning_rate": 0.0002, + "loss": 1.1606, "step": 409 }, { "epoch": 0.45, - "learning_rate": 0.0005, - "loss": 1.1855, + "learning_rate": 0.0002, + "loss": 1.202, "step": 410 }, { "epoch": 0.45, - "learning_rate": 0.0005, - "loss": 1.1184, + "learning_rate": 0.0002, + "loss": 1.1417, "step": 411 }, { "epoch": 0.45, - "learning_rate": 0.0005, - "loss": 1.0954, + "learning_rate": 0.0002, + "loss": 1.121, "step": 412 }, { "epoch": 0.45, - "learning_rate": 0.0005, - "loss": 1.0693, + "learning_rate": 0.0002, + "loss": 1.0887, "step": 413 }, { "epoch": 0.45, - "learning_rate": 0.0005, - "loss": 1.1026, + "learning_rate": 0.0002, + "loss": 1.1284, "step": 414 }, { "epoch": 0.45, - "learning_rate": 0.0005, - "loss": 1.169, + "learning_rate": 0.0002, + "loss": 1.1896, "step": 415 }, { "epoch": 0.45, - "learning_rate": 0.0005, - "loss": 1.1184, + "learning_rate": 0.0002, + "loss": 1.1336, "step": 416 }, { "epoch": 0.45, - "learning_rate": 0.0005, - "loss": 1.1343, + "learning_rate": 0.0002, + "loss": 1.1557, "step": 417 }, { "epoch": 0.45, - "learning_rate": 0.0005, - "loss": 1.1433, + "learning_rate": 0.0002, + "loss": 1.1685, "step": 418 }, { "epoch": 0.46, - "learning_rate": 0.0005, - "loss": 1.1394, + "learning_rate": 0.0002, + "loss": 1.165, "step": 419 }, { "epoch": 0.46, - "learning_rate": 0.0005, - "loss": 1.1908, + "learning_rate": 0.0002, + "loss": 1.2126, "step": 420 }, { "epoch": 0.46, - "learning_rate": 0.0005, - "loss": 1.1084, + "learning_rate": 0.0002, + "loss": 1.1309, "step": 421 }, { "epoch": 0.46, - "learning_rate": 0.0005, - "loss": 1.1292, + "learning_rate": 0.0002, + "loss": 1.1538, "step": 422 }, { "epoch": 0.46, - "learning_rate": 0.0005, - "loss": 1.1276, + "learning_rate": 0.0002, + "loss": 1.1469, "step": 423 }, { "epoch": 0.46, - "learning_rate": 0.0005, - "loss": 1.1821, + "learning_rate": 0.0002, + "loss": 1.2034, "step": 424 }, { "epoch": 0.46, - "learning_rate": 0.0005, - "loss": 1.075, + "learning_rate": 0.0002, + "loss": 1.091, "step": 425 }, { "epoch": 0.46, - "learning_rate": 0.0005, - "loss": 1.0903, + "learning_rate": 0.0002, + "loss": 1.1149, "step": 426 }, { "epoch": 0.46, - "learning_rate": 0.0005, - "loss": 1.0933, + "learning_rate": 0.0002, + "loss": 1.1083, "step": 427 }, { "epoch": 0.47, - "learning_rate": 0.0005, - "loss": 1.1403, + "learning_rate": 0.0002, + "loss": 1.1615, "step": 428 }, { "epoch": 0.47, - "learning_rate": 0.0005, - "loss": 1.1177, + "learning_rate": 0.0002, + "loss": 1.1426, "step": 429 }, { "epoch": 0.47, - "learning_rate": 0.0005, - "loss": 1.0836, + "learning_rate": 0.0002, + "loss": 1.1066, "step": 430 }, { "epoch": 0.47, - "learning_rate": 0.0005, - "loss": 1.0912, + "learning_rate": 0.0002, + "loss": 1.1123, "step": 431 }, { "epoch": 0.47, - "learning_rate": 0.0005, - "loss": 1.1686, + "learning_rate": 0.0002, + "loss": 1.195, "step": 432 }, { "epoch": 0.47, - "learning_rate": 0.0005, - "loss": 1.1583, + "learning_rate": 0.0002, + "loss": 1.1791, "step": 433 }, { "epoch": 0.47, - "learning_rate": 0.0005, - "loss": 1.1106, + "learning_rate": 0.0002, + "loss": 1.1307, "step": 434 }, { "epoch": 0.47, - "learning_rate": 0.0005, - "loss": 1.1318, + "learning_rate": 0.0002, + "loss": 1.1517, "step": 435 }, { "epoch": 0.47, - "learning_rate": 0.0005, - "loss": 1.0883, + "learning_rate": 0.0002, + "loss": 1.1053, "step": 436 }, { "epoch": 0.48, - "learning_rate": 0.0005, - "loss": 1.1449, + "learning_rate": 0.0002, + "loss": 1.1686, "step": 437 }, { "epoch": 0.48, - "learning_rate": 0.0005, - "loss": 1.1124, + "learning_rate": 0.0002, + "loss": 1.1367, "step": 438 }, { "epoch": 0.48, - "learning_rate": 0.0005, - "loss": 1.1016, + "learning_rate": 0.0002, + "loss": 1.1154, "step": 439 }, { "epoch": 0.48, - "learning_rate": 0.0005, - "loss": 1.0895, + "learning_rate": 0.0002, + "loss": 1.1091, "step": 440 }, { "epoch": 0.48, - "learning_rate": 0.0005, - "loss": 1.0981, + "learning_rate": 0.0002, + "loss": 1.1225, "step": 441 }, { "epoch": 0.48, - "learning_rate": 0.0005, - "loss": 1.1131, + "learning_rate": 0.0002, + "loss": 1.1268, "step": 442 }, { "epoch": 0.48, - "learning_rate": 0.0005, - "loss": 1.1049, + "learning_rate": 0.0002, + "loss": 1.1247, "step": 443 }, { "epoch": 0.48, - "learning_rate": 0.0005, - "loss": 1.0975, + "learning_rate": 0.0002, + "loss": 1.1187, "step": 444 }, { "epoch": 0.48, - "learning_rate": 0.0005, - "loss": 1.1034, + "learning_rate": 0.0002, + "loss": 1.122, "step": 445 }, { "epoch": 0.49, - "learning_rate": 0.0005, - "loss": 1.1582, + "learning_rate": 0.0002, + "loss": 1.1841, "step": 446 }, { "epoch": 0.49, - "learning_rate": 0.0005, - "loss": 1.1558, + "learning_rate": 0.0002, + "loss": 1.1805, "step": 447 }, { "epoch": 0.49, - "learning_rate": 0.0005, - "loss": 1.1147, + "learning_rate": 0.0002, + "loss": 1.1406, "step": 448 }, { "epoch": 0.49, - "learning_rate": 0.0005, - "loss": 1.1363, + "learning_rate": 0.0002, + "loss": 1.1567, "step": 449 }, { "epoch": 0.49, - "learning_rate": 0.0005, - "loss": 1.1115, + "learning_rate": 0.0002, + "loss": 1.1365, "step": 450 }, { "epoch": 0.49, - "learning_rate": 0.0005, - "loss": 1.1537, + "learning_rate": 0.0002, + "loss": 1.1686, "step": 451 }, { "epoch": 0.49, - "learning_rate": 0.0005, - "loss": 1.138, + "learning_rate": 0.0002, + "loss": 1.1462, "step": 452 }, { "epoch": 0.49, - "learning_rate": 0.0005, - "loss": 1.1728, + "learning_rate": 0.0002, + "loss": 1.1872, "step": 453 }, { "epoch": 0.49, - "learning_rate": 0.0005, - "loss": 1.1497, + "learning_rate": 0.0002, + "loss": 1.1659, "step": 454 }, { "epoch": 0.5, - "learning_rate": 0.0005, - "loss": 1.1443, + "learning_rate": 0.0002, + "loss": 1.1684, "step": 455 }, { "epoch": 0.5, - "learning_rate": 0.0005, - "loss": 1.0952, + "learning_rate": 0.0002, + "loss": 1.1268, "step": 456 }, { "epoch": 0.5, - "learning_rate": 0.0005, - "loss": 1.1314, + "learning_rate": 0.0002, + "loss": 1.1503, "step": 457 }, { "epoch": 0.5, - "learning_rate": 0.0005, - "loss": 1.0779, + "learning_rate": 0.0002, + "loss": 1.0989, "step": 458 }, { "epoch": 0.5, - "learning_rate": 0.0005, - "loss": 1.1676, + "learning_rate": 0.0002, + "loss": 1.1882, "step": 459 }, { "epoch": 0.5, - "learning_rate": 0.0005, - "loss": 1.1408, + "learning_rate": 0.0002, + "loss": 1.1581, "step": 460 }, { "epoch": 0.5, - "learning_rate": 0.0005, - "loss": 1.1355, + "learning_rate": 0.0002, + "loss": 1.1563, "step": 461 }, { "epoch": 0.5, - "learning_rate": 0.0005, - "loss": 1.1123, + "learning_rate": 0.0002, + "loss": 1.138, "step": 462 }, { "epoch": 0.5, - "learning_rate": 0.0005, - "loss": 1.0768, + "learning_rate": 0.0002, + "loss": 1.0931, "step": 463 }, { "epoch": 0.5, - "learning_rate": 0.0005, - "loss": 1.073, + "learning_rate": 0.0002, + "loss": 1.0868, "step": 464 }, { "epoch": 0.51, - "learning_rate": 0.0005, - "loss": 1.1002, + "learning_rate": 0.0002, + "loss": 1.1139, "step": 465 }, { "epoch": 0.51, - "learning_rate": 0.0005, - "loss": 1.1408, + "learning_rate": 0.0002, + "loss": 1.1572, "step": 466 }, { "epoch": 0.51, - "learning_rate": 0.0005, - "loss": 1.1524, + "learning_rate": 0.0002, + "loss": 1.1651, "step": 467 }, { "epoch": 0.51, - "learning_rate": 0.0005, - "loss": 1.1628, + "learning_rate": 0.0002, + "loss": 1.1823, "step": 468 }, { "epoch": 0.51, - "learning_rate": 0.0005, - "loss": 1.1003, + "learning_rate": 0.0002, + "loss": 1.1179, "step": 469 }, { "epoch": 0.51, - "learning_rate": 0.0005, - "loss": 1.1811, + "learning_rate": 0.0002, + "loss": 1.1989, "step": 470 }, { "epoch": 0.51, - "learning_rate": 0.0005, - "loss": 1.1205, + "learning_rate": 0.0002, + "loss": 1.1442, "step": 471 }, { "epoch": 0.51, - "learning_rate": 0.0005, - "loss": 1.1096, + "learning_rate": 0.0002, + "loss": 1.1301, "step": 472 }, { "epoch": 0.51, - "learning_rate": 0.0005, - "loss": 1.1496, + "learning_rate": 0.0002, + "loss": 1.1653, "step": 473 }, { "epoch": 0.52, - "learning_rate": 0.0005, - "loss": 1.1841, + "learning_rate": 0.0002, + "loss": 1.2039, "step": 474 }, { "epoch": 0.52, - "learning_rate": 0.0005, - "loss": 1.1353, + "learning_rate": 0.0002, + "loss": 1.1591, "step": 475 }, { "epoch": 0.52, - "learning_rate": 0.0005, - "loss": 1.0934, + "learning_rate": 0.0002, + "loss": 1.1059, "step": 476 }, { "epoch": 0.52, - "learning_rate": 0.0005, - "loss": 1.1177, + "learning_rate": 0.0002, + "loss": 1.1439, "step": 477 }, { "epoch": 0.52, - "learning_rate": 0.0005, - "loss": 1.1786, + "learning_rate": 0.0002, + "loss": 1.2052, "step": 478 }, { "epoch": 0.52, - "learning_rate": 0.0005, - "loss": 1.1388, + "learning_rate": 0.0002, + "loss": 1.1572, "step": 479 }, { "epoch": 0.52, - "learning_rate": 0.0005, - "loss": 1.1186, + "learning_rate": 0.0002, + "loss": 1.1353, "step": 480 }, { "epoch": 0.52, - "learning_rate": 0.0005, - "loss": 1.1976, + "learning_rate": 0.0002, + "loss": 1.213, "step": 481 }, { "epoch": 0.52, - "learning_rate": 0.0005, - "loss": 1.0979, + "learning_rate": 0.0002, + "loss": 1.1147, "step": 482 }, { "epoch": 0.53, - "learning_rate": 0.0005, - "loss": 1.1238, + "learning_rate": 0.0002, + "loss": 1.1403, "step": 483 }, { "epoch": 0.53, - "learning_rate": 0.0005, - "loss": 1.1358, + "learning_rate": 0.0002, + "loss": 1.1432, "step": 484 }, { "epoch": 0.53, - "learning_rate": 0.0005, - "loss": 1.1277, + "learning_rate": 0.0002, + "loss": 1.1429, "step": 485 }, { "epoch": 0.53, - "learning_rate": 0.0005, - "loss": 1.0825, + "learning_rate": 0.0002, + "loss": 1.105, "step": 486 }, { "epoch": 0.53, - "learning_rate": 0.0005, - "loss": 1.1324, + "learning_rate": 0.0002, + "loss": 1.1572, "step": 487 }, { "epoch": 0.53, - "learning_rate": 0.0005, - "loss": 1.1685, + "learning_rate": 0.0002, + "loss": 1.1908, "step": 488 }, { "epoch": 0.53, - "learning_rate": 0.0005, - "loss": 1.1475, + "learning_rate": 0.0002, + "loss": 1.1661, "step": 489 }, { "epoch": 0.53, - "learning_rate": 0.0005, - "loss": 1.1074, + "learning_rate": 0.0002, + "loss": 1.1233, "step": 490 }, { "epoch": 0.53, - "learning_rate": 0.0005, - "loss": 1.1237, + "learning_rate": 0.0002, + "loss": 1.1417, "step": 491 }, { "epoch": 0.54, - "learning_rate": 0.0005, - "loss": 1.1007, + "learning_rate": 0.0002, + "loss": 1.1213, "step": 492 }, { "epoch": 0.54, - "learning_rate": 0.0005, - "loss": 1.0941, + "learning_rate": 0.0002, + "loss": 1.1083, "step": 493 }, { "epoch": 0.54, - "learning_rate": 0.0005, - "loss": 1.171, + "learning_rate": 0.0002, + "loss": 1.1934, "step": 494 }, { "epoch": 0.54, - "learning_rate": 0.0005, - "loss": 1.1143, + "learning_rate": 0.0002, + "loss": 1.136, "step": 495 }, { "epoch": 0.54, - "learning_rate": 0.0005, - "loss": 1.1332, + "learning_rate": 0.0002, + "loss": 1.1528, "step": 496 }, { "epoch": 0.54, - "learning_rate": 0.0005, - "loss": 1.1109, + "learning_rate": 0.0002, + "loss": 1.1251, "step": 497 }, { "epoch": 0.54, - "learning_rate": 0.0005, - "loss": 1.1606, + "learning_rate": 0.0002, + "loss": 1.1822, "step": 498 }, { "epoch": 0.54, - "learning_rate": 0.0005, - "loss": 1.0875, + "learning_rate": 0.0002, + "loss": 1.1139, "step": 499 }, { "epoch": 0.54, - "learning_rate": 0.0005, - "loss": 1.093, + "learning_rate": 0.0002, + "loss": 1.109, "step": 500 }, { "epoch": 0.55, - "learning_rate": 0.0005, - "loss": 1.1218, + "learning_rate": 0.0002, + "loss": 1.1382, "step": 501 }, { "epoch": 0.55, - "learning_rate": 0.0005, - "loss": 1.1523, + "learning_rate": 0.0002, + "loss": 1.1715, "step": 502 }, { "epoch": 0.55, - "learning_rate": 0.0005, - "loss": 1.0864, + "learning_rate": 0.0002, + "loss": 1.0998, "step": 503 }, { "epoch": 0.55, - "learning_rate": 0.0005, - "loss": 1.1312, + "learning_rate": 0.0002, + "loss": 1.1553, "step": 504 }, { "epoch": 0.55, - "learning_rate": 0.0005, - "loss": 1.1031, + "learning_rate": 0.0002, + "loss": 1.1283, "step": 505 }, { "epoch": 0.55, - "learning_rate": 0.0005, - "loss": 1.0873, + "learning_rate": 0.0002, + "loss": 1.1016, "step": 506 }, { "epoch": 0.55, - "learning_rate": 0.0005, - "loss": 1.1007, + "learning_rate": 0.0002, + "loss": 1.115, "step": 507 }, { "epoch": 0.55, - "learning_rate": 0.0005, - "loss": 1.1589, + "learning_rate": 0.0002, + "loss": 1.1688, "step": 508 }, { "epoch": 0.55, - "learning_rate": 0.0005, - "loss": 1.1274, + "learning_rate": 0.0002, + "loss": 1.145, "step": 509 }, { "epoch": 0.55, - "learning_rate": 0.0005, - "loss": 1.0488, + "learning_rate": 0.0002, + "loss": 1.0592, "step": 510 }, { "epoch": 0.56, - "learning_rate": 0.0005, - "loss": 1.1344, + "learning_rate": 0.0002, + "loss": 1.1558, "step": 511 }, { "epoch": 0.56, - "learning_rate": 0.0005, - "loss": 1.1485, + "learning_rate": 0.0002, + "loss": 1.1715, "step": 512 }, { "epoch": 0.56, - "learning_rate": 0.0005, - "loss": 1.073, + "learning_rate": 0.0002, + "loss": 1.0879, "step": 513 }, { "epoch": 0.56, - "learning_rate": 0.0005, - "loss": 1.1457, + "learning_rate": 0.0002, + "loss": 1.1648, "step": 514 }, { "epoch": 0.56, - "learning_rate": 0.0005, - "loss": 1.1524, + "learning_rate": 0.0002, + "loss": 1.1689, "step": 515 }, { "epoch": 0.56, - "learning_rate": 0.0005, - "loss": 1.1407, + "learning_rate": 0.0002, + "loss": 1.1551, "step": 516 }, { "epoch": 0.56, - "learning_rate": 0.0005, - "loss": 1.1652, + "learning_rate": 0.0002, + "loss": 1.1889, "step": 517 }, { "epoch": 0.56, - "learning_rate": 0.0005, - "loss": 1.1391, + "learning_rate": 0.0002, + "loss": 1.1602, "step": 518 }, { "epoch": 0.56, - "learning_rate": 0.0005, - "loss": 1.1251, + "learning_rate": 0.0002, + "loss": 1.1499, "step": 519 }, { "epoch": 0.57, - "learning_rate": 0.0005, - "loss": 1.136, + "learning_rate": 0.0002, + "loss": 1.1619, "step": 520 }, { "epoch": 0.57, - "learning_rate": 0.0005, - "loss": 1.0951, + "learning_rate": 0.0002, + "loss": 1.1184, "step": 521 }, { "epoch": 0.57, - "learning_rate": 0.0005, - "loss": 1.122, + "learning_rate": 0.0002, + "loss": 1.1436, "step": 522 }, { "epoch": 0.57, - "learning_rate": 0.0005, - "loss": 1.097, + "learning_rate": 0.0002, + "loss": 1.1171, "step": 523 }, { "epoch": 0.57, - "learning_rate": 0.0005, - "loss": 1.1416, + "learning_rate": 0.0002, + "loss": 1.1618, "step": 524 }, { "epoch": 0.57, - "learning_rate": 0.0005, - "loss": 1.0924, + "learning_rate": 0.0002, + "loss": 1.1099, "step": 525 }, { "epoch": 0.57, - "learning_rate": 0.0005, - "loss": 1.1187, + "learning_rate": 0.0002, + "loss": 1.1311, "step": 526 }, { "epoch": 0.57, - "learning_rate": 0.0005, - "loss": 1.1388, + "learning_rate": 0.0002, + "loss": 1.1554, "step": 527 }, { "epoch": 0.57, - "learning_rate": 0.0005, - "loss": 1.1068, + "learning_rate": 0.0002, + "loss": 1.1237, "step": 528 }, { "epoch": 0.58, - "learning_rate": 0.0005, - "loss": 1.1338, + "learning_rate": 0.0002, + "loss": 1.1507, "step": 529 }, { "epoch": 0.58, - "learning_rate": 0.0005, - "loss": 1.089, + "learning_rate": 0.0002, + "loss": 1.1125, "step": 530 }, { "epoch": 0.58, - "learning_rate": 0.0005, - "loss": 1.1547, + "learning_rate": 0.0002, + "loss": 1.1743, "step": 531 }, { "epoch": 0.58, - "learning_rate": 0.0005, - "loss": 1.1474, + "learning_rate": 0.0002, + "loss": 1.1636, "step": 532 }, { "epoch": 0.58, - "learning_rate": 0.0005, - "loss": 1.1267, + "learning_rate": 0.0002, + "loss": 1.1443, "step": 533 }, { "epoch": 0.58, - "learning_rate": 0.0005, - "loss": 1.1456, + "learning_rate": 0.0002, + "loss": 1.1704, "step": 534 }, { "epoch": 0.58, - "learning_rate": 0.0005, - "loss": 1.1456, + "learning_rate": 0.0002, + "loss": 1.173, "step": 535 }, { "epoch": 0.58, - "learning_rate": 0.0005, - "loss": 1.0993, + "learning_rate": 0.0002, + "loss": 1.1179, "step": 536 }, { "epoch": 0.58, - "learning_rate": 0.0005, - "loss": 1.0625, + "learning_rate": 0.0002, + "loss": 1.0829, "step": 537 }, { "epoch": 0.59, - "learning_rate": 0.0005, - "loss": 1.1282, + "learning_rate": 0.0002, + "loss": 1.1566, "step": 538 }, { "epoch": 0.59, - "learning_rate": 0.0005, - "loss": 1.1154, + "learning_rate": 0.0002, + "loss": 1.1324, "step": 539 }, { "epoch": 0.59, - "learning_rate": 0.0005, - "loss": 1.1428, + "learning_rate": 0.0002, + "loss": 1.1637, "step": 540 }, { "epoch": 0.59, - "learning_rate": 0.0005, - "loss": 1.1507, + "learning_rate": 0.0002, + "loss": 1.1686, "step": 541 }, { "epoch": 0.59, - "learning_rate": 0.0005, - "loss": 1.1471, + "learning_rate": 0.0002, + "loss": 1.159, "step": 542 }, { "epoch": 0.59, - "learning_rate": 0.0005, - "loss": 1.1485, + "learning_rate": 0.0002, + "loss": 1.1662, "step": 543 }, { "epoch": 0.59, - "learning_rate": 0.0005, - "loss": 1.1608, + "learning_rate": 0.0002, + "loss": 1.172, "step": 544 }, { "epoch": 0.59, - "learning_rate": 0.0005, - "loss": 1.1688, + "learning_rate": 0.0002, + "loss": 1.1885, "step": 545 }, { "epoch": 0.59, - "learning_rate": 0.0005, - "loss": 1.1432, + "learning_rate": 0.0002, + "loss": 1.1561, "step": 546 }, { "epoch": 0.6, - "learning_rate": 0.0005, - "loss": 1.0744, + "learning_rate": 0.0002, + "loss": 1.0992, "step": 547 }, { "epoch": 0.6, - "learning_rate": 0.0005, - "loss": 1.1473, + "learning_rate": 0.0002, + "loss": 1.171, "step": 548 }, { "epoch": 0.6, - "learning_rate": 0.0005, - "loss": 1.129, + "learning_rate": 0.0002, + "loss": 1.1343, "step": 549 }, { "epoch": 0.6, - "learning_rate": 0.0005, - "loss": 1.0627, + "learning_rate": 0.0002, + "loss": 1.0924, "step": 550 }, { "epoch": 0.6, - "learning_rate": 0.0005, - "loss": 1.119, + "learning_rate": 0.0002, + "loss": 1.13, "step": 551 }, { "epoch": 0.6, - "learning_rate": 0.0005, - "loss": 1.0922, + "learning_rate": 0.0002, + "loss": 1.1109, "step": 552 }, { "epoch": 0.6, - "learning_rate": 0.0005, - "loss": 1.1226, + "learning_rate": 0.0002, + "loss": 1.1409, "step": 553 }, { "epoch": 0.6, - "learning_rate": 0.0005, - "loss": 1.1088, + "learning_rate": 0.0002, + "loss": 1.1313, "step": 554 }, { "epoch": 0.6, - "learning_rate": 0.0005, - "loss": 1.1086, + "learning_rate": 0.0002, + "loss": 1.1369, "step": 555 }, { "epoch": 0.61, - "learning_rate": 0.0005, - "loss": 1.1223, + "learning_rate": 0.0002, + "loss": 1.1443, "step": 556 }, { "epoch": 0.61, - "learning_rate": 0.0005, - "loss": 1.1532, + "learning_rate": 0.0002, + "loss": 1.1725, "step": 557 }, { "epoch": 0.61, - "learning_rate": 0.0005, - "loss": 1.0867, + "learning_rate": 0.0002, + "loss": 1.1089, "step": 558 }, { "epoch": 0.61, - "learning_rate": 0.0005, - "loss": 1.1249, + "learning_rate": 0.0002, + "loss": 1.1453, "step": 559 }, { "epoch": 0.61, - "learning_rate": 0.0005, - "loss": 1.0589, + "learning_rate": 0.0002, + "loss": 1.0835, "step": 560 }, { "epoch": 0.61, - "learning_rate": 0.0005, - "loss": 1.0997, + "learning_rate": 0.0002, + "loss": 1.1176, "step": 561 }, { "epoch": 0.61, - "learning_rate": 0.0005, - "loss": 1.106, + "learning_rate": 0.0002, + "loss": 1.1235, "step": 562 }, { "epoch": 0.61, - "learning_rate": 0.0005, - "loss": 1.1232, + "learning_rate": 0.0002, + "loss": 1.1362, "step": 563 }, { "epoch": 0.61, - "learning_rate": 0.0005, - "loss": 1.15, + "learning_rate": 0.0002, + "loss": 1.1725, "step": 564 }, { "epoch": 0.61, - "learning_rate": 0.0005, - "loss": 1.13, + "learning_rate": 0.0002, + "loss": 1.1467, "step": 565 }, { "epoch": 0.62, - "learning_rate": 0.0005, - "loss": 1.1356, + "learning_rate": 0.0002, + "loss": 1.1597, "step": 566 }, { "epoch": 0.62, - "learning_rate": 0.0005, - "loss": 1.0605, + "learning_rate": 0.0002, + "loss": 1.0761, "step": 567 }, { "epoch": 0.62, - "learning_rate": 0.0005, - "loss": 1.1148, + "learning_rate": 0.0002, + "loss": 1.1347, "step": 568 }, { "epoch": 0.62, - "learning_rate": 0.0005, - "loss": 1.0978, + "learning_rate": 0.0002, + "loss": 1.1097, "step": 569 }, { "epoch": 0.62, - "learning_rate": 0.0005, - "loss": 1.1588, + "learning_rate": 0.0002, + "loss": 1.187, "step": 570 }, { "epoch": 0.62, - "learning_rate": 0.0005, - "loss": 1.1003, + "learning_rate": 0.0002, + "loss": 1.117, "step": 571 }, { "epoch": 0.62, - "learning_rate": 0.0005, - "loss": 1.134, + "learning_rate": 0.0002, + "loss": 1.1491, "step": 572 }, { "epoch": 0.62, - "learning_rate": 0.0005, - "loss": 1.1209, + "learning_rate": 0.0002, + "loss": 1.1299, "step": 573 }, { "epoch": 0.62, - "learning_rate": 0.0005, - "loss": 1.1147, + "learning_rate": 0.0002, + "loss": 1.1295, "step": 574 }, { "epoch": 0.63, - "learning_rate": 0.0005, - "loss": 1.0861, + "learning_rate": 0.0002, + "loss": 1.0978, "step": 575 }, { "epoch": 0.63, - "learning_rate": 0.0005, - "loss": 1.1172, + "learning_rate": 0.0002, + "loss": 1.1331, "step": 576 }, { "epoch": 0.63, - "learning_rate": 0.0005, - "loss": 1.0743, + "learning_rate": 0.0002, + "loss": 1.0863, "step": 577 }, { "epoch": 0.63, - "learning_rate": 0.0005, - "loss": 1.1118, + "learning_rate": 0.0002, + "loss": 1.1329, "step": 578 }, { "epoch": 0.63, - "learning_rate": 0.0005, - "loss": 1.1286, + "learning_rate": 0.0002, + "loss": 1.1463, "step": 579 }, { "epoch": 0.63, - "learning_rate": 0.0005, - "loss": 1.1274, + "learning_rate": 0.0002, + "loss": 1.1398, "step": 580 }, { "epoch": 0.63, - "learning_rate": 0.0005, - "loss": 1.1367, + "learning_rate": 0.0002, + "loss": 1.1603, "step": 581 }, { "epoch": 0.63, - "learning_rate": 0.0005, - "loss": 1.1314, + "learning_rate": 0.0002, + "loss": 1.1542, "step": 582 }, { "epoch": 0.63, - "learning_rate": 0.0005, - "loss": 1.1111, + "learning_rate": 0.0002, + "loss": 1.1217, "step": 583 }, { "epoch": 0.64, - "learning_rate": 0.0005, - "loss": 1.0931, + "learning_rate": 0.0002, + "loss": 1.1041, "step": 584 }, { "epoch": 0.64, - "learning_rate": 0.0005, - "loss": 1.0855, + "learning_rate": 0.0002, + "loss": 1.0967, "step": 585 }, { "epoch": 0.64, - "learning_rate": 0.0005, - "loss": 1.1351, + "learning_rate": 0.0002, + "loss": 1.1547, "step": 586 }, { "epoch": 0.64, - "learning_rate": 0.0005, - "loss": 1.1783, + "learning_rate": 0.0002, + "loss": 1.1949, "step": 587 }, { "epoch": 0.64, - "learning_rate": 0.0005, - "loss": 1.1478, + "learning_rate": 0.0002, + "loss": 1.1662, "step": 588 }, { "epoch": 0.64, - "learning_rate": 0.0005, - "loss": 1.1735, + "learning_rate": 0.0002, + "loss": 1.1899, "step": 589 }, { "epoch": 0.64, - "learning_rate": 0.0005, - "loss": 1.1211, + "learning_rate": 0.0002, + "loss": 1.1376, "step": 590 }, { "epoch": 0.64, - "learning_rate": 0.0005, - "loss": 1.0519, + "learning_rate": 0.0002, + "loss": 1.074, "step": 591 }, { "epoch": 0.64, - "learning_rate": 0.0005, - "loss": 1.0837, + "learning_rate": 0.0002, + "loss": 1.1014, "step": 592 }, { "epoch": 0.65, - "learning_rate": 0.0005, - "loss": 1.1075, + "learning_rate": 0.0002, + "loss": 1.1223, "step": 593 }, { "epoch": 0.65, - "learning_rate": 0.0005, - "loss": 1.1191, + "learning_rate": 0.0002, + "loss": 1.1405, "step": 594 }, { "epoch": 0.65, - "learning_rate": 0.0005, - "loss": 1.1093, + "learning_rate": 0.0002, + "loss": 1.1346, "step": 595 }, { "epoch": 0.65, - "learning_rate": 0.0005, - "loss": 1.1389, + "learning_rate": 0.0002, + "loss": 1.1676, "step": 596 }, { "epoch": 0.65, - "learning_rate": 0.0005, - "loss": 1.1264, + "learning_rate": 0.0002, + "loss": 1.151, "step": 597 }, { "epoch": 0.65, - "learning_rate": 0.0005, - "loss": 1.1058, + "learning_rate": 0.0002, + "loss": 1.1129, "step": 598 }, { "epoch": 0.65, - "learning_rate": 0.0005, - "loss": 1.0963, + "learning_rate": 0.0002, + "loss": 1.1166, "step": 599 }, { "epoch": 0.65, - "learning_rate": 0.0005, - "loss": 1.1002, + "learning_rate": 0.0002, + "loss": 1.1173, "step": 600 }, { "epoch": 0.65, - "learning_rate": 0.0005, - "loss": 1.1513, + "learning_rate": 0.0002, + "loss": 1.1773, "step": 601 }, { "epoch": 0.66, - "learning_rate": 0.0005, - "loss": 1.0723, + "learning_rate": 0.0002, + "loss": 1.086, "step": 602 }, { "epoch": 0.66, - "learning_rate": 0.0005, - "loss": 1.1105, + "learning_rate": 0.0002, + "loss": 1.1213, "step": 603 }, { "epoch": 0.66, - "learning_rate": 0.0005, - "loss": 1.113, + "learning_rate": 0.0002, + "loss": 1.1316, "step": 604 }, { "epoch": 0.66, - "learning_rate": 0.0005, - "loss": 1.0865, + "learning_rate": 0.0002, + "loss": 1.1073, "step": 605 }, { "epoch": 0.66, - "learning_rate": 0.0005, - "loss": 1.1057, + "learning_rate": 0.0002, + "loss": 1.1201, "step": 606 }, { "epoch": 0.66, - "learning_rate": 0.0005, - "loss": 1.1192, + "learning_rate": 0.0002, + "loss": 1.1399, "step": 607 }, { "epoch": 0.66, - "learning_rate": 0.0005, - "loss": 1.1289, + "learning_rate": 0.0002, + "loss": 1.1432, "step": 608 }, { "epoch": 0.66, - "learning_rate": 0.0005, - "loss": 1.1108, + "learning_rate": 0.0002, + "loss": 1.1287, "step": 609 }, { "epoch": 0.66, - "learning_rate": 0.0005, - "loss": 1.1258, + "learning_rate": 0.0002, + "loss": 1.1471, "step": 610 }, { "epoch": 0.66, - "learning_rate": 0.0005, - "loss": 1.1143, + "learning_rate": 0.0002, + "loss": 1.1299, "step": 611 }, { "epoch": 0.67, - "learning_rate": 0.0005, - "loss": 1.1101, + "learning_rate": 0.0002, + "loss": 1.1307, "step": 612 }, { "epoch": 0.67, - "learning_rate": 0.0005, - "loss": 1.1329, + "learning_rate": 0.0002, + "loss": 1.1575, "step": 613 }, { "epoch": 0.67, - "learning_rate": 0.0005, - "loss": 1.1358, + "learning_rate": 0.0002, + "loss": 1.1489, "step": 614 }, { "epoch": 0.67, - "learning_rate": 0.0005, - "loss": 1.1403, + "learning_rate": 0.0002, + "loss": 1.1569, "step": 615 }, { "epoch": 0.67, - "learning_rate": 0.0005, - "loss": 1.1541, + "learning_rate": 0.0002, + "loss": 1.1597, "step": 616 }, { "epoch": 0.67, - "learning_rate": 0.0005, - "loss": 1.127, + "learning_rate": 0.0002, + "loss": 1.1337, "step": 617 }, { "epoch": 0.67, - "learning_rate": 0.0005, - "loss": 1.1154, + "learning_rate": 0.0002, + "loss": 1.1378, "step": 618 }, { "epoch": 0.67, - "learning_rate": 0.0005, - "loss": 1.1153, + "learning_rate": 0.0002, + "loss": 1.1262, "step": 619 }, { "epoch": 0.67, - "learning_rate": 0.0005, - "loss": 1.1383, + "learning_rate": 0.0002, + "loss": 1.144, "step": 620 }, { "epoch": 0.68, - "learning_rate": 0.0005, - "loss": 1.1302, + "learning_rate": 0.0002, + "loss": 1.1415, "step": 621 }, { "epoch": 0.68, - "learning_rate": 0.0005, - "loss": 1.1266, + "learning_rate": 0.0002, + "loss": 1.1377, "step": 622 }, { "epoch": 0.68, - "learning_rate": 0.0005, - "loss": 1.0733, + "learning_rate": 0.0002, + "loss": 1.0907, "step": 623 }, { "epoch": 0.68, - "learning_rate": 0.0005, - "loss": 1.0846, + "learning_rate": 0.0002, + "loss": 1.0939, "step": 624 }, { "epoch": 0.68, - "learning_rate": 0.0005, - "loss": 1.1305, + "learning_rate": 0.0002, + "loss": 1.1369, "step": 625 }, { "epoch": 0.68, - "learning_rate": 0.0005, - "loss": 1.1089, + "learning_rate": 0.0002, + "loss": 1.124, "step": 626 }, { "epoch": 0.68, - "learning_rate": 0.0005, - "loss": 1.1499, + "learning_rate": 0.0002, + "loss": 1.1665, "step": 627 }, { "epoch": 0.68, - "learning_rate": 0.0005, - "loss": 1.0957, + "learning_rate": 0.0002, + "loss": 1.1095, "step": 628 }, { "epoch": 0.68, - "learning_rate": 0.0005, - "loss": 1.1104, + "learning_rate": 0.0002, + "loss": 1.1243, "step": 629 }, { "epoch": 0.69, - "learning_rate": 0.0005, - "loss": 1.068, + "learning_rate": 0.0002, + "loss": 1.0787, "step": 630 }, { "epoch": 0.69, - "learning_rate": 0.0005, - "loss": 1.128, + "learning_rate": 0.0002, + "loss": 1.1527, "step": 631 }, { "epoch": 0.69, - "learning_rate": 0.0005, - "loss": 1.128, + "learning_rate": 0.0002, + "loss": 1.1394, "step": 632 }, { "epoch": 0.69, - "learning_rate": 0.0005, - "loss": 1.1251, + "learning_rate": 0.0002, + "loss": 1.1479, "step": 633 }, { "epoch": 0.69, - "learning_rate": 0.0005, - "loss": 1.0731, + "learning_rate": 0.0002, + "loss": 1.0814, "step": 634 }, { "epoch": 0.69, - "learning_rate": 0.0005, - "loss": 1.1001, + "learning_rate": 0.0002, + "loss": 1.1101, "step": 635 }, { "epoch": 0.69, - "learning_rate": 0.0005, - "loss": 1.122, + "learning_rate": 0.0002, + "loss": 1.1329, "step": 636 }, { "epoch": 0.69, - "learning_rate": 0.0005, - "loss": 1.1294, + "learning_rate": 0.0002, + "loss": 1.1443, "step": 637 }, { "epoch": 0.69, - "learning_rate": 0.0005, - "loss": 1.0869, + "learning_rate": 0.0002, + "loss": 1.1064, "step": 638 }, { "epoch": 0.7, - "learning_rate": 0.0005, - "loss": 1.0959, + "learning_rate": 0.0002, + "loss": 1.1215, "step": 639 }, { "epoch": 0.7, - "learning_rate": 0.0005, - "loss": 1.1069, + "learning_rate": 0.0002, + "loss": 1.1251, "step": 640 }, { "epoch": 0.7, - "learning_rate": 0.0005, - "loss": 1.1385, + "learning_rate": 0.0002, + "loss": 1.1555, "step": 641 }, { "epoch": 0.7, - "learning_rate": 0.0005, - "loss": 1.1011, + "learning_rate": 0.0002, + "loss": 1.1209, "step": 642 }, { "epoch": 0.7, - "learning_rate": 0.0005, - "loss": 1.0785, + "learning_rate": 0.0002, + "loss": 1.0904, "step": 643 }, { "epoch": 0.7, - "learning_rate": 0.0005, - "loss": 1.1236, + "learning_rate": 0.0002, + "loss": 1.1376, "step": 644 }, { "epoch": 0.7, - "learning_rate": 0.0005, - "loss": 1.1247, + "learning_rate": 0.0002, + "loss": 1.1488, "step": 645 }, { "epoch": 0.7, - "learning_rate": 0.0005, - "loss": 1.1202, + "learning_rate": 0.0002, + "loss": 1.1461, "step": 646 }, { "epoch": 0.7, - "learning_rate": 0.0005, - "loss": 1.0821, + "learning_rate": 0.0002, + "loss": 1.102, "step": 647 }, { "epoch": 0.71, - "learning_rate": 0.0005, - "loss": 1.1, + "learning_rate": 0.0002, + "loss": 1.1272, "step": 648 }, { "epoch": 0.71, - "learning_rate": 0.0005, - "loss": 1.1027, + "learning_rate": 0.0002, + "loss": 1.1237, "step": 649 }, { "epoch": 0.71, - "learning_rate": 0.0005, - "loss": 1.0754, + "learning_rate": 0.0002, + "loss": 1.0932, "step": 650 }, { "epoch": 0.71, - "learning_rate": 0.0005, - "loss": 1.0978, + "learning_rate": 0.0002, + "loss": 1.1167, "step": 651 }, { "epoch": 0.71, - "learning_rate": 0.0005, - "loss": 1.1193, + "learning_rate": 0.0002, + "loss": 1.141, "step": 652 }, { "epoch": 0.71, - "learning_rate": 0.0005, - "loss": 1.1039, + "learning_rate": 0.0002, + "loss": 1.118, "step": 653 }, { "epoch": 0.71, - "learning_rate": 0.0005, - "loss": 1.0711, + "learning_rate": 0.0002, + "loss": 1.0894, "step": 654 }, { "epoch": 0.71, - "learning_rate": 0.0005, - "loss": 1.1076, + "learning_rate": 0.0002, + "loss": 1.1214, "step": 655 }, { "epoch": 0.71, - "learning_rate": 0.0005, - "loss": 1.1399, + "learning_rate": 0.0002, + "loss": 1.1576, "step": 656 }, { "epoch": 0.71, - "learning_rate": 0.0005, - "loss": 1.1101, + "learning_rate": 0.0002, + "loss": 1.1256, "step": 657 }, { "epoch": 0.72, - "learning_rate": 0.0005, - "loss": 1.0897, + "learning_rate": 0.0002, + "loss": 1.0988, "step": 658 }, { "epoch": 0.72, - "learning_rate": 0.0005, - "loss": 1.1471, + "learning_rate": 0.0002, + "loss": 1.1661, "step": 659 }, { "epoch": 0.72, - "learning_rate": 0.0005, - "loss": 1.0853, + "learning_rate": 0.0002, + "loss": 1.101, "step": 660 }, { "epoch": 0.72, - "learning_rate": 0.0005, - "loss": 1.1272, + "learning_rate": 0.0002, + "loss": 1.1447, "step": 661 }, { "epoch": 0.72, - "learning_rate": 0.0005, - "loss": 1.1351, + "learning_rate": 0.0002, + "loss": 1.1472, "step": 662 }, { "epoch": 0.72, - "learning_rate": 0.0005, - "loss": 1.0876, + "learning_rate": 0.0002, + "loss": 1.1014, "step": 663 }, { "epoch": 0.72, - "learning_rate": 0.0005, - "loss": 1.0701, + "learning_rate": 0.0002, + "loss": 1.0913, "step": 664 }, { "epoch": 0.72, - "learning_rate": 0.0005, - "loss": 1.1206, + "learning_rate": 0.0002, + "loss": 1.1373, "step": 665 }, { "epoch": 0.72, - "learning_rate": 0.0005, - "loss": 1.0792, + "learning_rate": 0.0002, + "loss": 1.0946, "step": 666 }, { "epoch": 0.73, - "learning_rate": 0.0005, - "loss": 1.0793, + "learning_rate": 0.0002, + "loss": 1.0996, "step": 667 }, { "epoch": 0.73, - "learning_rate": 0.0005, - "loss": 1.0608, + "learning_rate": 0.0002, + "loss": 1.0844, "step": 668 }, { "epoch": 0.73, - "learning_rate": 0.0005, - "loss": 1.162, + "learning_rate": 0.0002, + "loss": 1.177, "step": 669 }, { "epoch": 0.73, - "learning_rate": 0.0005, - "loss": 1.1206, + "learning_rate": 0.0002, + "loss": 1.1377, "step": 670 }, { "epoch": 0.73, - "learning_rate": 0.0005, - "loss": 1.0972, + "learning_rate": 0.0002, + "loss": 1.1153, "step": 671 }, { "epoch": 0.73, - "learning_rate": 0.0005, - "loss": 1.0974, + "learning_rate": 0.0002, + "loss": 1.1156, "step": 672 }, { "epoch": 0.73, - "learning_rate": 0.0005, - "loss": 1.0726, + "learning_rate": 0.0002, + "loss": 1.0874, "step": 673 }, { "epoch": 0.73, - "learning_rate": 0.0005, - "loss": 1.0959, + "learning_rate": 0.0002, + "loss": 1.1152, "step": 674 }, { "epoch": 0.73, - "learning_rate": 0.0005, - "loss": 1.0941, + "learning_rate": 0.0002, + "loss": 1.1152, "step": 675 }, { "epoch": 0.74, - "learning_rate": 0.0005, - "loss": 1.1039, + "learning_rate": 0.0002, + "loss": 1.1221, "step": 676 }, { "epoch": 0.74, - "learning_rate": 0.0005, - "loss": 1.0615, + "learning_rate": 0.0002, + "loss": 1.0685, "step": 677 }, { "epoch": 0.74, - "learning_rate": 0.0005, - "loss": 1.1342, + "learning_rate": 0.0002, + "loss": 1.1453, "step": 678 }, { "epoch": 0.74, - "learning_rate": 0.0005, - "loss": 1.1919, + "learning_rate": 0.0002, + "loss": 1.2116, "step": 679 }, { "epoch": 0.74, - "learning_rate": 0.0005, - "loss": 1.1385, + "learning_rate": 0.0002, + "loss": 1.1607, "step": 680 }, { "epoch": 0.74, - "learning_rate": 0.0005, - "loss": 1.131, + "learning_rate": 0.0002, + "loss": 1.1446, "step": 681 }, { "epoch": 0.74, - "learning_rate": 0.0005, - "loss": 1.0772, + "learning_rate": 0.0002, + "loss": 1.0934, "step": 682 }, { "epoch": 0.74, - "learning_rate": 0.0005, - "loss": 1.0776, + "learning_rate": 0.0002, + "loss": 1.0891, "step": 683 }, { "epoch": 0.74, - "learning_rate": 0.0005, - "loss": 1.0766, + "learning_rate": 0.0002, + "loss": 1.0969, "step": 684 }, { "epoch": 0.75, - "learning_rate": 0.0005, - "loss": 1.0667, + "learning_rate": 0.0002, + "loss": 1.0864, "step": 685 }, { "epoch": 0.75, - "learning_rate": 0.0005, - "loss": 1.1359, + "learning_rate": 0.0002, + "loss": 1.1535, "step": 686 }, { "epoch": 0.75, - "learning_rate": 0.0005, - "loss": 1.0814, + "learning_rate": 0.0002, + "loss": 1.0933, "step": 687 }, { "epoch": 0.75, - "learning_rate": 0.0005, - "loss": 1.0671, + "learning_rate": 0.0002, + "loss": 1.0774, "step": 688 }, { "epoch": 0.75, - "learning_rate": 0.0005, - "loss": 1.1235, + "learning_rate": 0.0002, + "loss": 1.1354, "step": 689 }, { "epoch": 0.75, - "learning_rate": 0.0005, - "loss": 1.0756, + "learning_rate": 0.0002, + "loss": 1.0941, "step": 690 }, { "epoch": 0.75, - "learning_rate": 0.0005, - "loss": 1.1154, + "learning_rate": 0.0002, + "loss": 1.123, "step": 691 }, { "epoch": 0.75, - "learning_rate": 0.0005, - "loss": 1.1088, + "learning_rate": 0.0002, + "loss": 1.1245, "step": 692 }, { "epoch": 0.75, - "learning_rate": 0.0005, - "loss": 1.1147, + "learning_rate": 0.0002, + "loss": 1.1358, "step": 693 }, { "epoch": 0.76, - "learning_rate": 0.0005, - "loss": 1.095, + "learning_rate": 0.0002, + "loss": 1.1075, "step": 694 }, { "epoch": 0.76, - "learning_rate": 0.0005, - "loss": 1.1245, + "learning_rate": 0.0002, + "loss": 1.1392, "step": 695 }, { "epoch": 0.76, - "learning_rate": 0.0005, - "loss": 1.0816, + "learning_rate": 0.0002, + "loss": 1.0955, "step": 696 }, { "epoch": 0.76, - "learning_rate": 0.0005, - "loss": 1.1043, + "learning_rate": 0.0002, + "loss": 1.118, "step": 697 }, { "epoch": 0.76, - "learning_rate": 0.0005, - "loss": 1.075, + "learning_rate": 0.0002, + "loss": 1.0857, "step": 698 }, { "epoch": 0.76, - "learning_rate": 0.0005, - "loss": 1.0842, + "learning_rate": 0.0002, + "loss": 1.1049, "step": 699 }, { "epoch": 0.76, - "learning_rate": 0.0005, - "loss": 1.1424, + "learning_rate": 0.0002, + "loss": 1.1626, "step": 700 }, { "epoch": 0.76, - "learning_rate": 0.0005, - "loss": 1.1195, + "learning_rate": 0.0002, + "loss": 1.1302, "step": 701 }, { "epoch": 0.76, - "learning_rate": 0.0005, - "loss": 1.1016, + "learning_rate": 0.0002, + "loss": 1.1272, "step": 702 }, { "epoch": 0.76, - "learning_rate": 0.0005, - "loss": 1.1193, + "learning_rate": 0.0002, + "loss": 1.1297, "step": 703 }, { "epoch": 0.77, - "learning_rate": 0.0005, - "loss": 1.0593, + "learning_rate": 0.0002, + "loss": 1.0772, "step": 704 }, { "epoch": 0.77, - "learning_rate": 0.0005, - "loss": 1.0937, + "learning_rate": 0.0002, + "loss": 1.1172, "step": 705 }, { "epoch": 0.77, - "learning_rate": 0.0005, - "loss": 1.076, + "learning_rate": 0.0002, + "loss": 1.0905, "step": 706 }, { "epoch": 0.77, - "learning_rate": 0.0005, - "loss": 1.1058, + "learning_rate": 0.0002, + "loss": 1.1248, "step": 707 }, { "epoch": 0.77, - "learning_rate": 0.0005, - "loss": 1.1066, + "learning_rate": 0.0002, + "loss": 1.1193, "step": 708 }, { "epoch": 0.77, - "learning_rate": 0.0005, - "loss": 1.0824, + "learning_rate": 0.0002, + "loss": 1.0947, "step": 709 }, { "epoch": 0.77, - "learning_rate": 0.0005, - "loss": 1.1151, + "learning_rate": 0.0002, + "loss": 1.1283, "step": 710 }, { "epoch": 0.77, - "learning_rate": 0.0005, - "loss": 1.1501, + "learning_rate": 0.0002, + "loss": 1.1714, "step": 711 }, { "epoch": 0.77, - "learning_rate": 0.0005, - "loss": 1.1372, + "learning_rate": 0.0002, + "loss": 1.1622, "step": 712 }, { "epoch": 0.78, - "learning_rate": 0.0005, - "loss": 1.0637, + "learning_rate": 0.0002, + "loss": 1.0761, "step": 713 }, { "epoch": 0.78, - "learning_rate": 0.0005, - "loss": 1.1195, + "learning_rate": 0.0002, + "loss": 1.1363, "step": 714 }, { "epoch": 0.78, - "learning_rate": 0.0005, - "loss": 1.1021, + "learning_rate": 0.0002, + "loss": 1.1175, "step": 715 }, { "epoch": 0.78, - "learning_rate": 0.0005, - "loss": 1.1089, + "learning_rate": 0.0002, + "loss": 1.1182, "step": 716 }, { "epoch": 0.78, - "learning_rate": 0.0005, - "loss": 1.0846, + "learning_rate": 0.0002, + "loss": 1.1037, "step": 717 }, { "epoch": 0.78, - "learning_rate": 0.0005, - "loss": 1.1233, + "learning_rate": 0.0002, + "loss": 1.14, "step": 718 }, { "epoch": 0.78, - "learning_rate": 0.0005, - "loss": 1.1031, + "learning_rate": 0.0002, + "loss": 1.1198, "step": 719 }, { "epoch": 0.78, - "learning_rate": 0.0005, - "loss": 1.1211, + "learning_rate": 0.0002, + "loss": 1.1355, "step": 720 }, { "epoch": 0.78, - "learning_rate": 0.0005, - "loss": 1.0841, + "learning_rate": 0.0002, + "loss": 1.1009, "step": 721 }, { "epoch": 0.79, - "learning_rate": 0.0005, - "loss": 1.0998, + "learning_rate": 0.0002, + "loss": 1.115, "step": 722 }, { "epoch": 0.79, - "learning_rate": 0.0005, - "loss": 1.0864, + "learning_rate": 0.0002, + "loss": 1.1052, "step": 723 }, { "epoch": 0.79, - "learning_rate": 0.0005, - "loss": 1.1018, + "learning_rate": 0.0002, + "loss": 1.1241, "step": 724 }, { "epoch": 0.79, - "learning_rate": 0.0005, - "loss": 1.0778, + "learning_rate": 0.0002, + "loss": 1.0856, "step": 725 }, { "epoch": 0.79, - "learning_rate": 0.0005, - "loss": 1.1049, + "learning_rate": 0.0002, + "loss": 1.112, "step": 726 }, { "epoch": 0.79, - "learning_rate": 0.0005, - "loss": 1.1231, + "learning_rate": 0.0002, + "loss": 1.1407, "step": 727 }, { "epoch": 0.79, - "learning_rate": 0.0005, - "loss": 1.1025, + "learning_rate": 0.0002, + "loss": 1.1228, "step": 728 }, { "epoch": 0.79, - "learning_rate": 0.0005, - "loss": 1.111, + "learning_rate": 0.0002, + "loss": 1.1308, "step": 729 }, { "epoch": 0.79, - "learning_rate": 0.0005, - "loss": 1.1082, + "learning_rate": 0.0002, + "loss": 1.1223, "step": 730 }, { "epoch": 0.8, - "learning_rate": 0.0005, - "loss": 1.0837, + "learning_rate": 0.0002, + "loss": 1.1016, "step": 731 }, { "epoch": 0.8, - "learning_rate": 0.0005, - "loss": 1.0867, + "learning_rate": 0.0002, + "loss": 1.0964, "step": 732 }, { "epoch": 0.8, - "learning_rate": 0.0005, - "loss": 1.1125, + "learning_rate": 0.0002, + "loss": 1.1349, "step": 733 }, { "epoch": 0.8, - "learning_rate": 0.0005, - "loss": 1.0909, + "learning_rate": 0.0002, + "loss": 1.1099, "step": 734 }, { "epoch": 0.8, - "learning_rate": 0.0005, - "loss": 1.1443, + "learning_rate": 0.0002, + "loss": 1.1502, "step": 735 }, { "epoch": 0.8, - "learning_rate": 0.0005, - "loss": 1.1003, + "learning_rate": 0.0002, + "loss": 1.1155, "step": 736 }, { "epoch": 0.8, - "learning_rate": 0.0005, - "loss": 1.1581, + "learning_rate": 0.0002, + "loss": 1.1842, "step": 737 }, { "epoch": 0.8, - "learning_rate": 0.0005, - "loss": 1.0736, + "learning_rate": 0.0002, + "loss": 1.0961, "step": 738 }, { "epoch": 0.8, - "learning_rate": 0.0005, - "loss": 1.0974, + "learning_rate": 0.0002, + "loss": 1.1162, "step": 739 }, { "epoch": 0.81, - "learning_rate": 0.0005, - "loss": 1.0872, + "learning_rate": 0.0002, + "loss": 1.1027, "step": 740 }, { "epoch": 0.81, - "learning_rate": 0.0005, - "loss": 1.1186, + "learning_rate": 0.0002, + "loss": 1.1243, "step": 741 }, { "epoch": 0.81, - "learning_rate": 0.0005, - "loss": 1.0866, + "learning_rate": 0.0002, + "loss": 1.1069, "step": 742 }, { "epoch": 0.81, - "learning_rate": 0.0005, - "loss": 1.1553, + "learning_rate": 0.0002, + "loss": 1.1685, "step": 743 }, { "epoch": 0.81, - "learning_rate": 0.0005, - "loss": 1.078, + "learning_rate": 0.0002, + "loss": 1.106, "step": 744 }, { "epoch": 0.81, - "learning_rate": 0.0005, - "loss": 1.1081, + "learning_rate": 0.0002, + "loss": 1.1223, "step": 745 }, { "epoch": 0.81, - "learning_rate": 0.0005, - "loss": 1.0674, + "learning_rate": 0.0002, + "loss": 1.0892, "step": 746 }, { "epoch": 0.81, - "learning_rate": 0.0005, - "loss": 1.0973, + "learning_rate": 0.0002, + "loss": 1.1124, "step": 747 }, { "epoch": 0.81, - "learning_rate": 0.0005, - "loss": 1.0922, + "learning_rate": 0.0002, + "loss": 1.1033, "step": 748 }, { "epoch": 0.82, - "learning_rate": 0.0005, - "loss": 1.1481, + "learning_rate": 0.0002, + "loss": 1.1494, "step": 749 }, { "epoch": 0.82, - "learning_rate": 0.0005, - "loss": 1.1071, + "learning_rate": 0.0002, + "loss": 1.1153, "step": 750 }, { "epoch": 0.82, - "learning_rate": 0.0005, - "loss": 1.1343, + "learning_rate": 0.0002, + "loss": 1.1509, "step": 751 }, { "epoch": 0.82, - "learning_rate": 0.0005, - "loss": 1.1072, + "learning_rate": 0.0002, + "loss": 1.1332, "step": 752 }, { "epoch": 0.82, - "learning_rate": 0.0005, - "loss": 1.0918, + "learning_rate": 0.0002, + "loss": 1.0981, "step": 753 }, { "epoch": 0.82, - "learning_rate": 0.0005, - "loss": 1.1667, + "learning_rate": 0.0002, + "loss": 1.1835, "step": 754 }, { "epoch": 0.82, - "learning_rate": 0.0005, - "loss": 1.165, + "learning_rate": 0.0002, + "loss": 1.1804, "step": 755 }, { "epoch": 0.82, - "learning_rate": 0.0005, - "loss": 1.131, + "learning_rate": 0.0002, + "loss": 1.1495, "step": 756 }, { "epoch": 0.82, - "learning_rate": 0.0005, - "loss": 1.1252, + "learning_rate": 0.0002, + "loss": 1.1419, "step": 757 }, { "epoch": 0.82, - "learning_rate": 0.0005, - "loss": 1.1433, + "learning_rate": 0.0002, + "loss": 1.156, "step": 758 }, { "epoch": 0.83, - "learning_rate": 0.0005, - "loss": 1.1037, + "learning_rate": 0.0002, + "loss": 1.1254, "step": 759 }, { "epoch": 0.83, - "learning_rate": 0.0005, - "loss": 1.0903, + "learning_rate": 0.0002, + "loss": 1.1012, "step": 760 }, { "epoch": 0.83, - "learning_rate": 0.0005, - "loss": 1.1368, + "learning_rate": 0.0002, + "loss": 1.1527, "step": 761 }, { "epoch": 0.83, - "learning_rate": 0.0005, - "loss": 1.1149, + "learning_rate": 0.0002, + "loss": 1.13, "step": 762 }, { "epoch": 0.83, - "learning_rate": 0.0005, - "loss": 1.0738, + "learning_rate": 0.0002, + "loss": 1.0893, "step": 763 }, { "epoch": 0.83, - "learning_rate": 0.0005, - "loss": 1.0749, + "learning_rate": 0.0002, + "loss": 1.0871, "step": 764 }, { "epoch": 0.83, - "learning_rate": 0.0005, - "loss": 1.1186, + "learning_rate": 0.0002, + "loss": 1.1377, "step": 765 }, { "epoch": 0.83, - "learning_rate": 0.0005, - "loss": 1.0967, + "learning_rate": 0.0002, + "loss": 1.1105, "step": 766 }, { "epoch": 0.83, - "learning_rate": 0.0005, - "loss": 1.0909, + "learning_rate": 0.0002, + "loss": 1.1068, "step": 767 }, { "epoch": 0.84, - "learning_rate": 0.0005, - "loss": 1.0847, + "learning_rate": 0.0002, + "loss": 1.1039, "step": 768 }, { "epoch": 0.84, - "learning_rate": 0.0005, - "loss": 1.1469, + "learning_rate": 0.0002, + "loss": 1.1578, "step": 769 }, { "epoch": 0.84, - "learning_rate": 0.0005, - "loss": 1.118, + "learning_rate": 0.0002, + "loss": 1.1345, "step": 770 }, { "epoch": 0.84, - "learning_rate": 0.0005, - "loss": 1.1076, + "learning_rate": 0.0002, + "loss": 1.1201, "step": 771 }, { "epoch": 0.84, - "learning_rate": 0.0005, - "loss": 1.0834, + "learning_rate": 0.0002, + "loss": 1.0844, "step": 772 }, { "epoch": 0.84, - "learning_rate": 0.0005, - "loss": 1.0864, + "learning_rate": 0.0002, + "loss": 1.1009, "step": 773 }, { "epoch": 0.84, - "learning_rate": 0.0005, - "loss": 1.0932, + "learning_rate": 0.0002, + "loss": 1.1133, "step": 774 }, { "epoch": 0.84, - "learning_rate": 0.0005, - "loss": 1.0861, + "learning_rate": 0.0002, + "loss": 1.1073, "step": 775 }, { "epoch": 0.84, - "learning_rate": 0.0005, - "loss": 1.0539, + "learning_rate": 0.0002, + "loss": 1.0743, "step": 776 }, { "epoch": 0.85, - "learning_rate": 0.0005, - "loss": 1.1354, + "learning_rate": 0.0002, + "loss": 1.1437, "step": 777 }, { "epoch": 0.85, - "learning_rate": 0.0005, - "loss": 1.1273, + "learning_rate": 0.0002, + "loss": 1.1425, "step": 778 }, { "epoch": 0.85, - "learning_rate": 0.0005, - "loss": 1.1437, + "learning_rate": 0.0002, + "loss": 1.1573, "step": 779 }, { "epoch": 0.85, - "learning_rate": 0.0005, - "loss": 1.1279, + "learning_rate": 0.0002, + "loss": 1.1476, "step": 780 }, { "epoch": 0.85, - "learning_rate": 0.0005, - "loss": 1.0666, + "learning_rate": 0.0002, + "loss": 1.0792, "step": 781 }, { "epoch": 0.85, - "learning_rate": 0.0005, - "loss": 1.1058, + "learning_rate": 0.0002, + "loss": 1.1254, "step": 782 }, { "epoch": 0.85, - "learning_rate": 0.0005, - "loss": 1.1323, + "learning_rate": 0.0002, + "loss": 1.1422, "step": 783 }, { "epoch": 0.85, - "learning_rate": 0.0005, - "loss": 1.1055, + "learning_rate": 0.0002, + "loss": 1.1244, "step": 784 }, { "epoch": 0.85, - "learning_rate": 0.0005, - "loss": 1.0792, + "learning_rate": 0.0002, + "loss": 1.0957, "step": 785 }, { "epoch": 0.86, - "learning_rate": 0.0005, - "loss": 1.125, + "learning_rate": 0.0002, + "loss": 1.1384, "step": 786 }, { "epoch": 0.86, - "learning_rate": 0.0005, - "loss": 1.1169, + "learning_rate": 0.0002, + "loss": 1.1277, "step": 787 }, { "epoch": 0.86, - "learning_rate": 0.0005, - "loss": 1.1032, + "learning_rate": 0.0002, + "loss": 1.1184, "step": 788 }, { "epoch": 0.86, - "learning_rate": 0.0005, - "loss": 1.0784, + "learning_rate": 0.0002, + "loss": 1.0984, "step": 789 }, { "epoch": 0.86, - "learning_rate": 0.0005, - "loss": 1.1013, + "learning_rate": 0.0002, + "loss": 1.1069, "step": 790 }, { "epoch": 0.86, - "learning_rate": 0.0005, - "loss": 1.0913, + "learning_rate": 0.0002, + "loss": 1.113, "step": 791 }, { "epoch": 0.86, - "learning_rate": 0.0005, - "loss": 1.1077, + "learning_rate": 0.0002, + "loss": 1.1236, "step": 792 }, { "epoch": 0.86, - "learning_rate": 0.0005, - "loss": 1.1228, + "learning_rate": 0.0002, + "loss": 1.1315, "step": 793 }, { "epoch": 0.86, - "learning_rate": 0.0005, - "loss": 1.0615, + "learning_rate": 0.0002, + "loss": 1.0735, "step": 794 }, { "epoch": 0.87, - "learning_rate": 0.0005, - "loss": 1.1173, + "learning_rate": 0.0002, + "loss": 1.1308, "step": 795 }, { "epoch": 0.87, - "learning_rate": 0.0005, - "loss": 1.0966, + "learning_rate": 0.0002, + "loss": 1.1206, "step": 796 }, { "epoch": 0.87, - "learning_rate": 0.0005, - "loss": 1.1056, + "learning_rate": 0.0002, + "loss": 1.1305, "step": 797 }, { "epoch": 0.87, - "learning_rate": 0.0005, - "loss": 1.0328, + "learning_rate": 0.0002, + "loss": 1.0445, "step": 798 }, { "epoch": 0.87, - "learning_rate": 0.0005, - "loss": 1.1116, + "learning_rate": 0.0002, + "loss": 1.1355, "step": 799 }, { "epoch": 0.87, - "learning_rate": 0.0005, - "loss": 1.0858, + "learning_rate": 0.0002, + "loss": 1.105, "step": 800 }, { "epoch": 0.87, - "learning_rate": 0.0005, - "loss": 1.1091, + "learning_rate": 0.0002, + "loss": 1.1284, "step": 801 }, { "epoch": 0.87, - "learning_rate": 0.0005, - "loss": 1.0493, + "learning_rate": 0.0002, + "loss": 1.0728, "step": 802 }, { "epoch": 0.87, - "learning_rate": 0.0005, - "loss": 1.105, + "learning_rate": 0.0002, + "loss": 1.1192, "step": 803 }, { "epoch": 0.87, - "learning_rate": 0.0005, - "loss": 1.1777, + "learning_rate": 0.0002, + "loss": 1.1998, "step": 804 }, { "epoch": 0.88, - "learning_rate": 0.0005, - "loss": 1.1451, + "learning_rate": 0.0002, + "loss": 1.1652, "step": 805 }, { "epoch": 0.88, - "learning_rate": 0.0005, - "loss": 1.0778, + "learning_rate": 0.0002, + "loss": 1.095, "step": 806 }, { "epoch": 0.88, - "learning_rate": 0.0005, - "loss": 1.0932, + "learning_rate": 0.0002, + "loss": 1.1081, "step": 807 }, { "epoch": 0.88, - "learning_rate": 0.0005, - "loss": 1.1024, + "learning_rate": 0.0002, + "loss": 1.118, "step": 808 }, { "epoch": 0.88, - "learning_rate": 0.0005, - "loss": 1.111, + "learning_rate": 0.0002, + "loss": 1.133, "step": 809 }, { "epoch": 0.88, - "learning_rate": 0.0005, - "loss": 1.1229, + "learning_rate": 0.0002, + "loss": 1.1408, "step": 810 }, { "epoch": 0.88, - "learning_rate": 0.0005, - "loss": 1.0897, + "learning_rate": 0.0002, + "loss": 1.1111, "step": 811 }, { "epoch": 0.88, - "learning_rate": 0.0005, - "loss": 1.1202, + "learning_rate": 0.0002, + "loss": 1.1341, "step": 812 }, { "epoch": 0.88, - "learning_rate": 0.0005, - "loss": 1.1137, + "learning_rate": 0.0002, + "loss": 1.137, "step": 813 }, { "epoch": 0.89, - "learning_rate": 0.0005, - "loss": 1.1186, + "learning_rate": 0.0002, + "loss": 1.1298, "step": 814 }, { "epoch": 0.89, - "learning_rate": 0.0005, - "loss": 1.1214, + "learning_rate": 0.0002, + "loss": 1.1341, "step": 815 }, { "epoch": 0.89, - "learning_rate": 0.0005, - "loss": 1.0675, + "learning_rate": 0.0002, + "loss": 1.0827, "step": 816 }, { "epoch": 0.89, - "learning_rate": 0.0005, - "loss": 1.1189, + "learning_rate": 0.0002, + "loss": 1.1288, "step": 817 }, { "epoch": 0.89, - "learning_rate": 0.0005, - "loss": 1.0946, + "learning_rate": 0.0002, + "loss": 1.1108, "step": 818 }, { "epoch": 0.89, - "learning_rate": 0.0005, - "loss": 1.1108, + "learning_rate": 0.0002, + "loss": 1.1341, "step": 819 }, { "epoch": 0.89, - "learning_rate": 0.0005, - "loss": 1.0859, + "learning_rate": 0.0002, + "loss": 1.1025, "step": 820 }, { "epoch": 0.89, - "learning_rate": 0.0005, - "loss": 1.1146, + "learning_rate": 0.0002, + "loss": 1.1312, "step": 821 }, { "epoch": 0.89, - "learning_rate": 0.0005, - "loss": 1.0901, + "learning_rate": 0.0002, + "loss": 1.1076, "step": 822 }, { "epoch": 0.9, - "learning_rate": 0.0005, - "loss": 1.102, + "learning_rate": 0.0002, + "loss": 1.1128, "step": 823 }, { "epoch": 0.9, - "learning_rate": 0.0005, - "loss": 1.11, + "learning_rate": 0.0002, + "loss": 1.1222, "step": 824 }, { "epoch": 0.9, - "learning_rate": 0.0005, - "loss": 1.0579, + "learning_rate": 0.0002, + "loss": 1.071, "step": 825 }, { "epoch": 0.9, - "learning_rate": 0.0005, - "loss": 1.1133, + "learning_rate": 0.0002, + "loss": 1.1343, "step": 826 }, { "epoch": 0.9, - "learning_rate": 0.0005, - "loss": 1.0737, + "learning_rate": 0.0002, + "loss": 1.0913, "step": 827 }, { "epoch": 0.9, - "learning_rate": 0.0005, - "loss": 1.0851, + "learning_rate": 0.0002, + "loss": 1.1079, "step": 828 }, { "epoch": 0.9, - "learning_rate": 0.0005, - "loss": 1.0861, + "learning_rate": 0.0002, + "loss": 1.1028, "step": 829 }, { "epoch": 0.9, - "learning_rate": 0.0005, - "loss": 1.1276, + "learning_rate": 0.0002, + "loss": 1.145, "step": 830 }, { "epoch": 0.9, - "learning_rate": 0.0005, - "loss": 1.1065, + "learning_rate": 0.0002, + "loss": 1.1247, "step": 831 }, { "epoch": 0.91, - "learning_rate": 0.0005, - "loss": 1.071, + "learning_rate": 0.0002, + "loss": 1.0956, "step": 832 }, { "epoch": 0.91, - "learning_rate": 0.0005, - "loss": 1.1132, + "learning_rate": 0.0002, + "loss": 1.1283, "step": 833 }, { "epoch": 0.91, - "learning_rate": 0.0005, - "loss": 1.0474, + "learning_rate": 0.0002, + "loss": 1.0676, "step": 834 }, { "epoch": 0.91, - "learning_rate": 0.0005, - "loss": 1.0713, + "learning_rate": 0.0002, + "loss": 1.0809, "step": 835 }, { "epoch": 0.91, - "learning_rate": 0.0005, - "loss": 1.0895, + "learning_rate": 0.0002, + "loss": 1.1104, "step": 836 }, { "epoch": 0.91, - "learning_rate": 0.0005, - "loss": 1.1, + "learning_rate": 0.0002, + "loss": 1.1173, "step": 837 }, { "epoch": 0.91, - "learning_rate": 0.0005, - "loss": 1.0584, + "learning_rate": 0.0002, + "loss": 1.0672, "step": 838 }, { "epoch": 0.91, - "learning_rate": 0.0005, - "loss": 1.1124, + "learning_rate": 0.0002, + "loss": 1.1331, "step": 839 }, { "epoch": 0.91, - "learning_rate": 0.0005, - "loss": 1.0842, + "learning_rate": 0.0002, + "loss": 1.0964, "step": 840 }, { "epoch": 0.92, - "learning_rate": 0.0005, - "loss": 1.1105, + "learning_rate": 0.0002, + "loss": 1.1244, "step": 841 }, { "epoch": 0.92, - "learning_rate": 0.0005, - "loss": 1.0448, + "learning_rate": 0.0002, + "loss": 1.0605, "step": 842 }, { "epoch": 0.92, - "learning_rate": 0.0005, - "loss": 1.0846, + "learning_rate": 0.0002, + "loss": 1.0989, "step": 843 }, { "epoch": 0.92, - "learning_rate": 0.0005, - "loss": 1.1041, + "learning_rate": 0.0002, + "loss": 1.1209, "step": 844 }, { "epoch": 0.92, - "learning_rate": 0.0005, - "loss": 1.1228, + "learning_rate": 0.0002, + "loss": 1.1429, "step": 845 }, { "epoch": 0.92, - "learning_rate": 0.0005, - "loss": 1.0668, + "learning_rate": 0.0002, + "loss": 1.0857, "step": 846 }, { "epoch": 0.92, - "learning_rate": 0.0005, - "loss": 1.0761, + "learning_rate": 0.0002, + "loss": 1.0997, "step": 847 }, { "epoch": 0.92, - "learning_rate": 0.0005, - "loss": 1.065, + "learning_rate": 0.0002, + "loss": 1.083, "step": 848 }, { "epoch": 0.92, - "learning_rate": 0.0005, - "loss": 1.1331, + "learning_rate": 0.0002, + "loss": 1.1503, "step": 849 }, { "epoch": 0.92, - "learning_rate": 0.0005, - "loss": 1.068, + "learning_rate": 0.0002, + "loss": 1.0814, "step": 850 }, { "epoch": 0.93, - "learning_rate": 0.0005, - "loss": 1.0584, + "learning_rate": 0.0002, + "loss": 1.0757, "step": 851 }, { "epoch": 0.93, - "learning_rate": 0.0005, - "loss": 1.1088, + "learning_rate": 0.0002, + "loss": 1.1207, "step": 852 }, { "epoch": 0.93, - "learning_rate": 0.0005, - "loss": 1.0701, + "learning_rate": 0.0002, + "loss": 1.0929, "step": 853 }, { "epoch": 0.93, - "learning_rate": 0.0005, - "loss": 1.1073, + "learning_rate": 0.0002, + "loss": 1.1238, "step": 854 }, { "epoch": 0.93, - "learning_rate": 0.0005, - "loss": 1.1508, + "learning_rate": 0.0002, + "loss": 1.1706, "step": 855 }, { "epoch": 0.93, - "learning_rate": 0.0005, - "loss": 1.1345, + "learning_rate": 0.0002, + "loss": 1.1495, "step": 856 }, { "epoch": 0.93, - "learning_rate": 0.0005, - "loss": 1.1011, + "learning_rate": 0.0002, + "loss": 1.1186, "step": 857 }, { "epoch": 0.93, - "learning_rate": 0.0005, - "loss": 1.1197, + "learning_rate": 0.0002, + "loss": 1.1455, "step": 858 }, { "epoch": 0.93, - "learning_rate": 0.0005, - "loss": 1.1237, + "learning_rate": 0.0002, + "loss": 1.1397, "step": 859 }, { "epoch": 0.94, - "learning_rate": 0.0005, - "loss": 1.1065, + "learning_rate": 0.0002, + "loss": 1.1187, "step": 860 }, { "epoch": 0.94, - "learning_rate": 0.0005, - "loss": 1.0708, + "learning_rate": 0.0002, + "loss": 1.0875, "step": 861 }, { "epoch": 0.94, - "learning_rate": 0.0005, - "loss": 1.1131, + "learning_rate": 0.0002, + "loss": 1.1339, "step": 862 }, { "epoch": 0.94, - "learning_rate": 0.0005, - "loss": 1.1019, + "learning_rate": 0.0002, + "loss": 1.112, "step": 863 }, { "epoch": 0.94, - "learning_rate": 0.0005, - "loss": 1.1584, + "learning_rate": 0.0002, + "loss": 1.172, "step": 864 }, { "epoch": 0.94, - "learning_rate": 0.0005, - "loss": 1.0882, + "learning_rate": 0.0002, + "loss": 1.0988, "step": 865 }, { "epoch": 0.94, - "learning_rate": 0.0005, - "loss": 1.0999, + "learning_rate": 0.0002, + "loss": 1.1128, "step": 866 }, { "epoch": 0.94, - "learning_rate": 0.0005, - "loss": 1.076, + "learning_rate": 0.0002, + "loss": 1.0914, "step": 867 }, { "epoch": 0.94, - "learning_rate": 0.0005, - "loss": 1.088, + "learning_rate": 0.0002, + "loss": 1.1053, "step": 868 }, { "epoch": 0.95, - "learning_rate": 0.0005, - "loss": 1.1374, + "learning_rate": 0.0002, + "loss": 1.1523, "step": 869 }, { "epoch": 0.95, - "learning_rate": 0.0005, - "loss": 1.0797, + "learning_rate": 0.0002, + "loss": 1.0923, "step": 870 }, { "epoch": 0.95, - "learning_rate": 0.0005, - "loss": 1.1058, + "learning_rate": 0.0002, + "loss": 1.1284, "step": 871 }, { "epoch": 0.95, - "learning_rate": 0.0005, - "loss": 1.1146, + "learning_rate": 0.0002, + "loss": 1.1363, "step": 872 }, { "epoch": 0.95, - "learning_rate": 0.0005, - "loss": 1.0843, + "learning_rate": 0.0002, + "loss": 1.1032, "step": 873 }, { "epoch": 0.95, - "learning_rate": 0.0005, - "loss": 1.1031, + "learning_rate": 0.0002, + "loss": 1.1185, "step": 874 }, { "epoch": 0.95, - "learning_rate": 0.0005, - "loss": 1.0865, + "learning_rate": 0.0002, + "loss": 1.1039, "step": 875 }, { "epoch": 0.95, - "learning_rate": 0.0005, - "loss": 1.0452, + "learning_rate": 0.0002, + "loss": 1.058, "step": 876 }, { "epoch": 0.95, - "learning_rate": 0.0005, - "loss": 1.0778, + "learning_rate": 0.0002, + "loss": 1.0896, "step": 877 }, { "epoch": 0.96, - "learning_rate": 0.0005, - "loss": 1.0756, + "learning_rate": 0.0002, + "loss": 1.0941, "step": 878 }, { "epoch": 0.96, - "learning_rate": 0.0005, - "loss": 1.1474, + "learning_rate": 0.0002, + "loss": 1.1697, "step": 879 }, { "epoch": 0.96, - "learning_rate": 0.0005, - "loss": 1.103, + "learning_rate": 0.0002, + "loss": 1.1224, "step": 880 }, { "epoch": 0.96, - "learning_rate": 0.0005, - "loss": 1.0811, + "learning_rate": 0.0002, + "loss": 1.0918, "step": 881 }, { "epoch": 0.96, - "learning_rate": 0.0005, - "loss": 1.0889, + "learning_rate": 0.0002, + "loss": 1.1016, "step": 882 }, { "epoch": 0.96, - "learning_rate": 0.0005, - "loss": 1.0838, + "learning_rate": 0.0002, + "loss": 1.0979, "step": 883 }, { "epoch": 0.96, - "learning_rate": 0.0005, - "loss": 1.0785, + "learning_rate": 0.0002, + "loss": 1.0914, "step": 884 }, { "epoch": 0.96, - "learning_rate": 0.0005, - "loss": 1.0681, + "learning_rate": 0.0002, + "loss": 1.0869, "step": 885 }, { "epoch": 0.96, - "learning_rate": 0.0005, - "loss": 1.0803, + "learning_rate": 0.0002, + "loss": 1.0952, "step": 886 }, { "epoch": 0.97, - "learning_rate": 0.0005, - "loss": 1.059, + "learning_rate": 0.0002, + "loss": 1.0792, "step": 887 }, { "epoch": 0.97, - "learning_rate": 0.0005, - "loss": 1.061, + "learning_rate": 0.0002, + "loss": 1.0787, "step": 888 }, { "epoch": 0.97, - "learning_rate": 0.0005, - "loss": 1.0898, + "learning_rate": 0.0002, + "loss": 1.1077, "step": 889 }, { "epoch": 0.97, - "learning_rate": 0.0005, - "loss": 1.1096, + "learning_rate": 0.0002, + "loss": 1.1303, "step": 890 }, { "epoch": 0.97, - "learning_rate": 0.0005, - "loss": 1.0858, + "learning_rate": 0.0002, + "loss": 1.1088, "step": 891 }, { "epoch": 0.97, - "learning_rate": 0.0005, - "loss": 1.075, + "learning_rate": 0.0002, + "loss": 1.0881, "step": 892 }, { "epoch": 0.97, - "learning_rate": 0.0005, - "loss": 1.0722, + "learning_rate": 0.0002, + "loss": 1.0903, "step": 893 }, { "epoch": 0.97, - "learning_rate": 0.0005, - "loss": 1.0998, + "learning_rate": 0.0002, + "loss": 1.1193, "step": 894 }, { "epoch": 0.97, - "learning_rate": 0.0005, - "loss": 1.073, + "learning_rate": 0.0002, + "loss": 1.0901, "step": 895 }, { "epoch": 0.97, - "learning_rate": 0.0005, - "loss": 1.132, + "learning_rate": 0.0002, + "loss": 1.1433, "step": 896 }, { "epoch": 0.98, - "learning_rate": 0.0005, - "loss": 1.1191, + "learning_rate": 0.0002, + "loss": 1.1341, "step": 897 }, { "epoch": 0.98, - "learning_rate": 0.0005, - "loss": 1.0656, + "learning_rate": 0.0002, + "loss": 1.0764, "step": 898 }, { "epoch": 0.98, - "learning_rate": 0.0005, - "loss": 1.0935, + "learning_rate": 0.0002, + "loss": 1.095, "step": 899 }, { "epoch": 0.98, - "learning_rate": 0.0005, - "loss": 1.0748, + "learning_rate": 0.0002, + "loss": 1.0829, "step": 900 }, { "epoch": 0.98, - "learning_rate": 0.0005, - "loss": 1.1, + "learning_rate": 0.0002, + "loss": 1.1135, "step": 901 }, { "epoch": 0.98, - "learning_rate": 0.0005, - "loss": 1.0594, + "learning_rate": 0.0002, + "loss": 1.0825, "step": 902 }, { "epoch": 0.98, - "learning_rate": 0.0005, - "loss": 1.1052, + "learning_rate": 0.0002, + "loss": 1.1274, "step": 903 }, { "epoch": 0.98, - "learning_rate": 0.0005, - "loss": 1.0865, + "learning_rate": 0.0002, + "loss": 1.0939, "step": 904 }, { "epoch": 0.98, - "learning_rate": 0.0005, - "loss": 1.1092, + "learning_rate": 0.0002, + "loss": 1.1339, "step": 905 }, { "epoch": 0.99, - "learning_rate": 0.0005, - "loss": 1.0992, + "learning_rate": 0.0002, + "loss": 1.1178, "step": 906 }, { "epoch": 0.99, - "learning_rate": 0.0005, - "loss": 1.0669, + "learning_rate": 0.0002, + "loss": 1.0873, "step": 907 }, { "epoch": 0.99, - "learning_rate": 0.0005, - "loss": 1.0876, + "learning_rate": 0.0002, + "loss": 1.0974, "step": 908 }, { "epoch": 0.99, - "learning_rate": 0.0005, - "loss": 1.0512, + "learning_rate": 0.0002, + "loss": 1.0636, "step": 909 }, { "epoch": 0.99, - "learning_rate": 0.0005, - "loss": 1.0696, + "learning_rate": 0.0002, + "loss": 1.084, "step": 910 }, { "epoch": 0.99, - "learning_rate": 0.0005, - "loss": 1.1406, + "learning_rate": 0.0002, + "loss": 1.1569, "step": 911 }, { "epoch": 0.99, - "learning_rate": 0.0005, - "loss": 1.113, + "learning_rate": 0.0002, + "loss": 1.1193, "step": 912 }, { "epoch": 0.99, - "learning_rate": 0.0005, - "loss": 1.1111, + "learning_rate": 0.0002, + "loss": 1.1275, "step": 913 }, { "epoch": 0.99, - "learning_rate": 0.0005, - "loss": 1.1172, + "learning_rate": 0.0002, + "loss": 1.1333, "step": 914 }, { "epoch": 1.0, - "learning_rate": 0.0005, - "loss": 1.0936, + "learning_rate": 0.0002, + "loss": 1.108, "step": 915 }, { "epoch": 1.0, - "learning_rate": 0.0005, - "loss": 1.0646, + "learning_rate": 0.0002, + "loss": 1.0811, "step": 916 }, { "epoch": 1.0, - "learning_rate": 0.0005, - "loss": 1.0913, + "learning_rate": 0.0002, + "loss": 1.0994, "step": 917 }, { "epoch": 1.0, - "learning_rate": 0.0005, - "loss": 1.1158, + "learning_rate": 0.0002, + "loss": 1.1285, "step": 918 }, { "epoch": 1.0, - "learning_rate": 0.0005, - "loss": 1.2066, + "learning_rate": 0.0002, + "loss": 1.2129, "step": 919 }, { "epoch": 1.0, - "learning_rate": 0.0005, - "loss": 1.0921, + "learning_rate": 0.0002, + "loss": 1.1188, "step": 920 }, { "epoch": 1.0, - "learning_rate": 0.0005, - "loss": 1.0417, + "learning_rate": 0.0002, + "loss": 1.0622, "step": 921 }, { "epoch": 1.0, - "learning_rate": 0.0005, - "loss": 1.0623, + "learning_rate": 0.0002, + "loss": 1.0937, "step": 922 }, { "epoch": 1.0, - "learning_rate": 0.0005, - "loss": 1.0693, + "learning_rate": 0.0002, + "loss": 1.0902, "step": 923 }, { "epoch": 1.01, - "learning_rate": 0.0005, - "loss": 1.1329, + "learning_rate": 0.0002, + "loss": 1.1583, "step": 924 }, { "epoch": 1.01, - "learning_rate": 0.0005, - "loss": 1.0405, + "learning_rate": 0.0002, + "loss": 1.0703, "step": 925 }, { "epoch": 1.01, - "learning_rate": 0.0005, - "loss": 1.123, + "learning_rate": 0.0002, + "loss": 1.1429, "step": 926 }, { "epoch": 1.01, - "learning_rate": 0.0005, - "loss": 1.0907, + "learning_rate": 0.0002, + "loss": 1.1143, "step": 927 }, { "epoch": 1.01, - "learning_rate": 0.0005, - "loss": 1.1027, + "learning_rate": 0.0002, + "loss": 1.1339, "step": 928 }, { "epoch": 1.01, - "learning_rate": 0.0005, - "loss": 1.0361, + "learning_rate": 0.0002, + "loss": 1.0587, "step": 929 }, { "epoch": 1.01, - "learning_rate": 0.0005, - "loss": 1.0717, + "learning_rate": 0.0002, + "loss": 1.0953, "step": 930 }, { "epoch": 1.01, - "learning_rate": 0.0005, - "loss": 1.0983, + "learning_rate": 0.0002, + "loss": 1.122, "step": 931 }, { "epoch": 1.01, - "learning_rate": 0.0005, - "loss": 1.0931, + "learning_rate": 0.0002, + "loss": 1.1205, "step": 932 }, { "epoch": 1.02, - "learning_rate": 0.0005, - "loss": 1.0287, + "learning_rate": 0.0002, + "loss": 1.0554, "step": 933 }, { "epoch": 1.02, - "learning_rate": 0.0005, - "loss": 1.0251, + "learning_rate": 0.0002, + "loss": 1.062, "step": 934 }, { "epoch": 1.02, - "learning_rate": 0.0005, - "loss": 1.083, + "learning_rate": 0.0002, + "loss": 1.1055, "step": 935 }, { "epoch": 1.02, - "learning_rate": 0.0005, - "loss": 1.0168, + "learning_rate": 0.0002, + "loss": 1.039, "step": 936 }, { "epoch": 1.02, - "learning_rate": 0.0005, - "loss": 1.0348, + "learning_rate": 0.0002, + "loss": 1.057, "step": 937 }, { "epoch": 1.02, - "learning_rate": 0.0005, - "loss": 1.0208, + "learning_rate": 0.0002, + "loss": 1.0456, "step": 938 }, { "epoch": 1.02, - "learning_rate": 0.0005, - "loss": 1.0745, + "learning_rate": 0.0002, + "loss": 1.0998, "step": 939 }, { "epoch": 1.02, - "learning_rate": 0.0005, - "loss": 1.0597, + "learning_rate": 0.0002, + "loss": 1.0815, "step": 940 }, { "epoch": 1.02, - "learning_rate": 0.0005, - "loss": 1.086, + "learning_rate": 0.0002, + "loss": 1.1086, "step": 941 }, { "epoch": 1.03, - "learning_rate": 0.0005, - "loss": 1.1229, + "learning_rate": 0.0002, + "loss": 1.1553, "step": 942 }, { "epoch": 1.03, - "learning_rate": 0.0005, - "loss": 1.0629, + "learning_rate": 0.0002, + "loss": 1.0887, "step": 943 }, { "epoch": 1.03, - "learning_rate": 0.0005, - "loss": 1.0613, + "learning_rate": 0.0002, + "loss": 1.0755, "step": 944 }, { "epoch": 1.03, - "learning_rate": 0.0005, - "loss": 1.1087, + "learning_rate": 0.0002, + "loss": 1.1345, "step": 945 }, { "epoch": 1.03, - "learning_rate": 0.0005, - "loss": 1.0878, + "learning_rate": 0.0002, + "loss": 1.1152, "step": 946 }, { "epoch": 1.03, - "learning_rate": 0.0005, - "loss": 1.0584, + "learning_rate": 0.0002, + "loss": 1.0971, "step": 947 }, { "epoch": 1.03, - "learning_rate": 0.0005, - "loss": 1.0745, + "learning_rate": 0.0002, + "loss": 1.1036, "step": 948 }, { "epoch": 1.03, - "learning_rate": 0.0005, - "loss": 1.0703, + "learning_rate": 0.0002, + "loss": 1.0959, "step": 949 }, { "epoch": 1.03, - "learning_rate": 0.0005, - "loss": 1.0519, + "learning_rate": 0.0002, + "loss": 1.0774, "step": 950 }, { "epoch": 1.03, - "learning_rate": 0.0005, - "loss": 1.0344, + "learning_rate": 0.0002, + "loss": 1.0614, "step": 951 }, { "epoch": 1.04, - "learning_rate": 0.0005, - "loss": 1.0741, + "learning_rate": 0.0002, + "loss": 1.1111, "step": 952 }, { "epoch": 1.04, - "learning_rate": 0.0005, - "loss": 1.0876, + "learning_rate": 0.0002, + "loss": 1.1109, "step": 953 }, { "epoch": 1.04, - "learning_rate": 0.0005, - "loss": 1.1, + "learning_rate": 0.0002, + "loss": 1.1275, "step": 954 }, { "epoch": 1.04, - "learning_rate": 0.0005, - "loss": 1.0372, + "learning_rate": 0.0002, + "loss": 1.0608, "step": 955 }, { "epoch": 1.04, - "learning_rate": 0.0005, - "loss": 1.0536, + "learning_rate": 0.0002, + "loss": 1.0764, "step": 956 }, { "epoch": 1.04, - "learning_rate": 0.0005, - "loss": 1.1005, + "learning_rate": 0.0002, + "loss": 1.1193, "step": 957 }, { "epoch": 1.04, - "learning_rate": 0.0005, - "loss": 1.0845, + "learning_rate": 0.0002, + "loss": 1.1043, "step": 958 }, { "epoch": 1.04, - "learning_rate": 0.0005, - "loss": 1.0262, + "learning_rate": 0.0002, + "loss": 1.0525, "step": 959 }, { "epoch": 1.04, - "learning_rate": 0.0005, - "loss": 1.0848, + "learning_rate": 0.0002, + "loss": 1.1126, "step": 960 }, { "epoch": 1.05, - "learning_rate": 0.0005, - "loss": 1.0271, + "learning_rate": 0.0002, + "loss": 1.0502, "step": 961 }, { "epoch": 1.05, - "learning_rate": 0.0005, - "loss": 1.1024, + "learning_rate": 0.0002, + "loss": 1.1266, "step": 962 }, { "epoch": 1.05, - "learning_rate": 0.0005, - "loss": 1.0802, + "learning_rate": 0.0002, + "loss": 1.1193, "step": 963 }, { "epoch": 1.05, - "learning_rate": 0.0005, - "loss": 1.115, + "learning_rate": 0.0002, + "loss": 1.1345, "step": 964 }, { "epoch": 1.05, - "learning_rate": 0.0005, - "loss": 1.0878, + "learning_rate": 0.0002, + "loss": 1.1187, "step": 965 }, { "epoch": 1.05, - "learning_rate": 0.0005, - "loss": 1.0238, + "learning_rate": 0.0002, + "loss": 1.0485, "step": 966 }, { "epoch": 1.05, - "learning_rate": 0.0005, - "loss": 1.059, + "learning_rate": 0.0002, + "loss": 1.0865, "step": 967 }, { "epoch": 1.05, - "learning_rate": 0.0005, - "loss": 1.0751, + "learning_rate": 0.0002, + "loss": 1.1045, "step": 968 }, { "epoch": 1.05, - "learning_rate": 0.0005, - "loss": 1.0384, + "learning_rate": 0.0002, + "loss": 1.0616, "step": 969 }, { "epoch": 1.06, - "learning_rate": 0.0005, - "loss": 1.07, + "learning_rate": 0.0002, + "loss": 1.0937, "step": 970 }, { "epoch": 1.06, - "learning_rate": 0.0005, - "loss": 1.0762, + "learning_rate": 0.0002, + "loss": 1.1092, "step": 971 }, { "epoch": 1.06, - "learning_rate": 0.0005, - "loss": 1.0458, + "learning_rate": 0.0002, + "loss": 1.0723, "step": 972 }, { "epoch": 1.06, - "learning_rate": 0.0005, - "loss": 1.0867, + "learning_rate": 0.0002, + "loss": 1.112, "step": 973 }, { "epoch": 1.06, - "learning_rate": 0.0005, - "loss": 1.0888, + "learning_rate": 0.0002, + "loss": 1.118, "step": 974 }, { "epoch": 1.06, - "learning_rate": 0.0005, - "loss": 1.1093, + "learning_rate": 0.0002, + "loss": 1.1305, "step": 975 }, { "epoch": 1.06, - "learning_rate": 0.0005, - "loss": 1.0892, + "learning_rate": 0.0002, + "loss": 1.1113, "step": 976 }, { "epoch": 1.06, - "learning_rate": 0.0005, - "loss": 1.0605, + "learning_rate": 0.0002, + "loss": 1.0855, "step": 977 }, { "epoch": 1.06, - "learning_rate": 0.0005, - "loss": 1.084, + "learning_rate": 0.0002, + "loss": 1.1144, "step": 978 }, { "epoch": 1.07, - "learning_rate": 0.0005, - "loss": 1.0844, + "learning_rate": 0.0002, + "loss": 1.1109, "step": 979 }, { "epoch": 1.07, - "learning_rate": 0.0005, - "loss": 1.0767, + "learning_rate": 0.0002, + "loss": 1.1075, "step": 980 }, { "epoch": 1.07, - "learning_rate": 0.0005, - "loss": 1.0712, + "learning_rate": 0.0002, + "loss": 1.0965, "step": 981 }, { "epoch": 1.07, - "learning_rate": 0.0005, - "loss": 1.0664, + "learning_rate": 0.0002, + "loss": 1.0825, "step": 982 }, { "epoch": 1.07, - "learning_rate": 0.0005, - "loss": 1.0287, + "learning_rate": 0.0002, + "loss": 1.0585, "step": 983 }, { "epoch": 1.07, - "learning_rate": 0.0005, - "loss": 1.0598, + "learning_rate": 0.0002, + "loss": 1.086, "step": 984 }, { "epoch": 1.07, - "learning_rate": 0.0005, - "loss": 1.1338, + "learning_rate": 0.0002, + "loss": 1.1588, "step": 985 }, { "epoch": 1.07, - "learning_rate": 0.0005, - "loss": 1.0842, + "learning_rate": 0.0002, + "loss": 1.1164, "step": 986 }, { "epoch": 1.07, - "learning_rate": 0.0005, - "loss": 1.0454, + "learning_rate": 0.0002, + "loss": 1.0727, "step": 987 }, { "epoch": 1.08, - "learning_rate": 0.0005, - "loss": 1.06, + "learning_rate": 0.0002, + "loss": 1.0857, "step": 988 }, { "epoch": 1.08, - "learning_rate": 0.0005, - "loss": 1.0736, + "learning_rate": 0.0002, + "loss": 1.0951, "step": 989 }, { "epoch": 1.08, - "learning_rate": 0.0005, - "loss": 1.095, + "learning_rate": 0.0002, + "loss": 1.12, "step": 990 }, { "epoch": 1.08, - "learning_rate": 0.0005, - "loss": 1.0626, + "learning_rate": 0.0002, + "loss": 1.0828, "step": 991 }, { "epoch": 1.08, - "learning_rate": 0.0005, - "loss": 1.0652, + "learning_rate": 0.0002, + "loss": 1.0921, "step": 992 }, { "epoch": 1.08, - "learning_rate": 0.0005, - "loss": 1.1389, + "learning_rate": 0.0002, + "loss": 1.1596, "step": 993 }, { "epoch": 1.08, - "learning_rate": 0.0005, - "loss": 1.0765, + "learning_rate": 0.0002, + "loss": 1.1029, "step": 994 }, { "epoch": 1.08, - "learning_rate": 0.0005, - "loss": 1.0853, + "learning_rate": 0.0002, + "loss": 1.1173, "step": 995 }, { "epoch": 1.08, - "learning_rate": 0.0005, - "loss": 1.081, + "learning_rate": 0.0002, + "loss": 1.0991, "step": 996 }, { "epoch": 1.08, - "learning_rate": 0.0005, - "loss": 1.0418, + "learning_rate": 0.0002, + "loss": 1.0622, "step": 997 }, { "epoch": 1.09, - "learning_rate": 0.0005, - "loss": 1.1099, + "learning_rate": 0.0002, + "loss": 1.1356, "step": 998 }, { "epoch": 1.09, - "learning_rate": 0.0005, - "loss": 1.0864, + "learning_rate": 0.0002, + "loss": 1.1144, "step": 999 }, { "epoch": 1.09, - "learning_rate": 0.0005, - "loss": 1.0563, + "learning_rate": 0.0002, + "loss": 1.0752, "step": 1000 }, { "epoch": 1.09, - "learning_rate": 0.0005, - "loss": 1.0611, + "learning_rate": 0.0002, + "loss": 1.0893, "step": 1001 }, { "epoch": 1.09, - "learning_rate": 0.0005, - "loss": 1.0848, + "learning_rate": 0.0002, + "loss": 1.1038, "step": 1002 }, { "epoch": 1.09, - "learning_rate": 0.0005, - "loss": 1.0647, + "learning_rate": 0.0002, + "loss": 1.0889, "step": 1003 }, { "epoch": 1.09, - "learning_rate": 0.0005, - "loss": 1.0548, + "learning_rate": 0.0002, + "loss": 1.0839, "step": 1004 }, { "epoch": 1.09, - "learning_rate": 0.0005, - "loss": 1.0707, + "learning_rate": 0.0002, + "loss": 1.0968, "step": 1005 }, { "epoch": 1.09, - "learning_rate": 0.0005, - "loss": 1.0998, + "learning_rate": 0.0002, + "loss": 1.1157, "step": 1006 }, { "epoch": 1.1, - "learning_rate": 0.0005, - "loss": 1.0808, + "learning_rate": 0.0002, + "loss": 1.1145, "step": 1007 }, { "epoch": 1.1, - "learning_rate": 0.0005, - "loss": 1.1005, + "learning_rate": 0.0002, + "loss": 1.1203, "step": 1008 }, { "epoch": 1.1, - "learning_rate": 0.0005, - "loss": 1.0063, + "learning_rate": 0.0002, + "loss": 1.0326, "step": 1009 }, { "epoch": 1.1, - "learning_rate": 0.0005, - "loss": 1.0937, + "learning_rate": 0.0002, + "loss": 1.1152, "step": 1010 }, { "epoch": 1.1, - "learning_rate": 0.0005, - "loss": 1.0599, + "learning_rate": 0.0002, + "loss": 1.0844, "step": 1011 }, { "epoch": 1.1, - "learning_rate": 0.0005, - "loss": 1.0372, + "learning_rate": 0.0002, + "loss": 1.0556, "step": 1012 }, { "epoch": 1.1, - "learning_rate": 0.0005, - "loss": 1.0683, + "learning_rate": 0.0002, + "loss": 1.0952, "step": 1013 }, { "epoch": 1.1, - "learning_rate": 0.0005, - "loss": 1.0776, + "learning_rate": 0.0002, + "loss": 1.1081, "step": 1014 }, { "epoch": 1.1, - "learning_rate": 0.0005, - "loss": 1.107, + "learning_rate": 0.0002, + "loss": 1.1323, "step": 1015 }, { "epoch": 1.11, - "learning_rate": 0.0005, - "loss": 1.0666, + "learning_rate": 0.0002, + "loss": 1.0845, "step": 1016 }, { "epoch": 1.11, - "learning_rate": 0.0005, - "loss": 1.0708, + "learning_rate": 0.0002, + "loss": 1.1048, "step": 1017 }, { "epoch": 1.11, - "learning_rate": 0.0005, - "loss": 1.0231, + "learning_rate": 0.0002, + "loss": 1.0536, "step": 1018 }, { "epoch": 1.11, - "learning_rate": 0.0005, - "loss": 1.1159, + "learning_rate": 0.0002, + "loss": 1.1401, "step": 1019 }, { "epoch": 1.11, - "learning_rate": 0.0005, - "loss": 1.0887, + "learning_rate": 0.0002, + "loss": 1.1149, "step": 1020 }, { "epoch": 1.11, - "learning_rate": 0.0005, - "loss": 1.0604, + "learning_rate": 0.0002, + "loss": 1.0908, "step": 1021 }, { "epoch": 1.11, - "learning_rate": 0.0005, - "loss": 1.0647, + "learning_rate": 0.0002, + "loss": 1.089, "step": 1022 }, { "epoch": 1.11, - "learning_rate": 0.0005, - "loss": 1.0885, + "learning_rate": 0.0002, + "loss": 1.1178, "step": 1023 }, { "epoch": 1.11, - "learning_rate": 0.0005, - "loss": 1.0994, + "learning_rate": 0.0002, + "loss": 1.1328, "step": 1024 }, { "epoch": 1.12, - "learning_rate": 0.0005, - "loss": 1.1123, + "learning_rate": 0.0002, + "loss": 1.1377, "step": 1025 }, { "epoch": 1.12, - "learning_rate": 0.0005, - "loss": 1.0419, + "learning_rate": 0.0002, + "loss": 1.0703, "step": 1026 }, { "epoch": 1.12, - "learning_rate": 0.0005, - "loss": 1.0666, + "learning_rate": 0.0002, + "loss": 1.0938, "step": 1027 }, { "epoch": 1.12, - "learning_rate": 0.0005, - "loss": 1.0461, + "learning_rate": 0.0002, + "loss": 1.0779, "step": 1028 }, { "epoch": 1.12, - "learning_rate": 0.0005, - "loss": 1.0478, + "learning_rate": 0.0002, + "loss": 1.0769, "step": 1029 }, { "epoch": 1.12, - "learning_rate": 0.0005, - "loss": 1.0503, + "learning_rate": 0.0002, + "loss": 1.0828, "step": 1030 }, { "epoch": 1.12, - "learning_rate": 0.0005, - "loss": 1.0637, + "learning_rate": 0.0002, + "loss": 1.0991, "step": 1031 }, { "epoch": 1.12, - "learning_rate": 0.0005, - "loss": 1.0565, + "learning_rate": 0.0002, + "loss": 1.0851, "step": 1032 }, { "epoch": 1.12, - "learning_rate": 0.0005, - "loss": 1.0977, + "learning_rate": 0.0002, + "loss": 1.1167, "step": 1033 }, { "epoch": 1.13, - "learning_rate": 0.0005, - "loss": 1.0618, + "learning_rate": 0.0002, + "loss": 1.0851, "step": 1034 }, { "epoch": 1.13, - "learning_rate": 0.0005, - "loss": 1.0344, + "learning_rate": 0.0002, + "loss": 1.0663, "step": 1035 }, { "epoch": 1.13, - "learning_rate": 0.0005, - "loss": 1.0871, + "learning_rate": 0.0002, + "loss": 1.1154, "step": 1036 }, { "epoch": 1.13, - "learning_rate": 0.0005, - "loss": 1.0463, + "learning_rate": 0.0002, + "loss": 1.0739, "step": 1037 }, { "epoch": 1.13, - "learning_rate": 0.0005, - "loss": 1.1187, + "learning_rate": 0.0002, + "loss": 1.1509, "step": 1038 }, { "epoch": 1.13, - "learning_rate": 0.0005, - "loss": 1.0801, + "learning_rate": 0.0002, + "loss": 1.1103, "step": 1039 }, { "epoch": 1.13, - "learning_rate": 0.0005, - "loss": 1.0614, + "learning_rate": 0.0002, + "loss": 1.0907, "step": 1040 }, { "epoch": 1.13, - "learning_rate": 0.0005, - "loss": 1.086, + "learning_rate": 0.0002, + "loss": 1.1107, "step": 1041 }, { "epoch": 1.13, - "learning_rate": 0.0005, - "loss": 1.0538, + "learning_rate": 0.0002, + "loss": 1.0789, "step": 1042 }, { "epoch": 1.13, - "learning_rate": 0.0005, - "loss": 1.0509, + "learning_rate": 0.0002, + "loss": 1.0765, "step": 1043 }, { "epoch": 1.14, - "learning_rate": 0.0005, - "loss": 1.0721, + "learning_rate": 0.0002, + "loss": 1.0878, "step": 1044 }, { "epoch": 1.14, - "learning_rate": 0.0005, - "loss": 1.0618, + "learning_rate": 0.0002, + "loss": 1.085, "step": 1045 }, { "epoch": 1.14, - "learning_rate": 0.0005, - "loss": 1.0721, + "learning_rate": 0.0002, + "loss": 1.096, "step": 1046 }, { "epoch": 1.14, - "learning_rate": 0.0005, - "loss": 1.0543, + "learning_rate": 0.0002, + "loss": 1.0774, "step": 1047 }, { "epoch": 1.14, - "learning_rate": 0.0005, - "loss": 1.0546, + "learning_rate": 0.0002, + "loss": 1.0733, "step": 1048 }, { "epoch": 1.14, - "learning_rate": 0.0005, - "loss": 1.0844, + "learning_rate": 0.0002, + "loss": 1.1002, "step": 1049 }, { "epoch": 1.14, - "learning_rate": 0.0005, - "loss": 1.0564, + "learning_rate": 0.0002, + "loss": 1.0946, "step": 1050 }, { "epoch": 1.14, - "learning_rate": 0.0005, - "loss": 1.1279, + "learning_rate": 0.0002, + "loss": 1.1495, "step": 1051 }, { "epoch": 1.14, - "learning_rate": 0.0005, - "loss": 1.1062, + "learning_rate": 0.0002, + "loss": 1.1257, "step": 1052 }, { "epoch": 1.15, - "learning_rate": 0.0005, - "loss": 1.0518, + "learning_rate": 0.0002, + "loss": 1.0852, "step": 1053 }, { "epoch": 1.15, - "learning_rate": 0.0005, - "loss": 1.108, + "learning_rate": 0.0002, + "loss": 1.1395, "step": 1054 }, { "epoch": 1.15, - "learning_rate": 0.0005, - "loss": 1.0449, + "learning_rate": 0.0002, + "loss": 1.0665, "step": 1055 }, { "epoch": 1.15, - "learning_rate": 0.0005, - "loss": 1.0863, + "learning_rate": 0.0002, + "loss": 1.112, "step": 1056 }, { "epoch": 1.15, - "learning_rate": 0.0005, - "loss": 1.1088, + "learning_rate": 0.0002, + "loss": 1.1364, "step": 1057 }, { "epoch": 1.15, - "learning_rate": 0.0005, - "loss": 1.068, + "learning_rate": 0.0002, + "loss": 1.0915, "step": 1058 }, { "epoch": 1.15, - "learning_rate": 0.0005, - "loss": 1.0614, + "learning_rate": 0.0002, + "loss": 1.0753, "step": 1059 }, { "epoch": 1.15, - "learning_rate": 0.0005, - "loss": 1.0344, + "learning_rate": 0.0002, + "loss": 1.0537, "step": 1060 }, { "epoch": 1.15, - "learning_rate": 0.0005, - "loss": 1.0553, + "learning_rate": 0.0002, + "loss": 1.0767, "step": 1061 }, { "epoch": 1.16, - "learning_rate": 0.0005, - "loss": 1.0474, + "learning_rate": 0.0002, + "loss": 1.0749, "step": 1062 }, { "epoch": 1.16, - "learning_rate": 0.0005, - "loss": 1.0608, + "learning_rate": 0.0002, + "loss": 1.0819, "step": 1063 }, { "epoch": 1.16, - "learning_rate": 0.0005, - "loss": 1.098, + "learning_rate": 0.0002, + "loss": 1.135, "step": 1064 }, { "epoch": 1.16, - "learning_rate": 0.0005, - "loss": 1.1034, + "learning_rate": 0.0002, + "loss": 1.1323, "step": 1065 }, { "epoch": 1.16, - "learning_rate": 0.0005, - "loss": 1.0347, + "learning_rate": 0.0002, + "loss": 1.0692, "step": 1066 }, { "epoch": 1.16, - "learning_rate": 0.0005, - "loss": 1.087, + "learning_rate": 0.0002, + "loss": 1.1057, "step": 1067 }, { "epoch": 1.16, - "learning_rate": 0.0005, - "loss": 1.0697, + "learning_rate": 0.0002, + "loss": 1.0977, "step": 1068 }, { "epoch": 1.16, - "learning_rate": 0.0005, - "loss": 1.1041, + "learning_rate": 0.0002, + "loss": 1.1224, "step": 1069 }, { "epoch": 1.16, - "learning_rate": 0.0005, - "loss": 1.0404, + "learning_rate": 0.0002, + "loss": 1.0641, "step": 1070 }, { "epoch": 1.17, - "learning_rate": 0.0005, - "loss": 1.0604, + "learning_rate": 0.0002, + "loss": 1.0827, "step": 1071 }, { "epoch": 1.17, - "learning_rate": 0.0005, - "loss": 1.0919, + "learning_rate": 0.0002, + "loss": 1.1165, "step": 1072 }, { "epoch": 1.17, - "learning_rate": 0.0005, - "loss": 1.0197, + "learning_rate": 0.0002, + "loss": 1.0466, "step": 1073 }, { "epoch": 1.17, - "learning_rate": 0.0005, - "loss": 1.0883, + "learning_rate": 0.0002, + "loss": 1.107, "step": 1074 }, { "epoch": 1.17, - "learning_rate": 0.0005, - "loss": 1.0987, + "learning_rate": 0.0002, + "loss": 1.1155, "step": 1075 }, { "epoch": 1.17, - "learning_rate": 0.0005, - "loss": 1.0301, + "learning_rate": 0.0002, + "loss": 1.0548, "step": 1076 }, { "epoch": 1.17, - "learning_rate": 0.0005, - "loss": 1.0506, + "learning_rate": 0.0002, + "loss": 1.0823, "step": 1077 }, { "epoch": 1.17, - "learning_rate": 0.0005, - "loss": 1.0704, + "learning_rate": 0.0002, + "loss": 1.0951, "step": 1078 }, { "epoch": 1.17, - "learning_rate": 0.0005, - "loss": 1.0905, + "learning_rate": 0.0002, + "loss": 1.1071, "step": 1079 }, { "epoch": 1.18, - "learning_rate": 0.0005, - "loss": 1.0012, + "learning_rate": 0.0002, + "loss": 1.0346, "step": 1080 }, { "epoch": 1.18, - "learning_rate": 0.0005, - "loss": 1.0614, + "learning_rate": 0.0002, + "loss": 1.0876, "step": 1081 }, { "epoch": 1.18, - "learning_rate": 0.0005, - "loss": 1.0517, + "learning_rate": 0.0002, + "loss": 1.0752, "step": 1082 }, { "epoch": 1.18, - "learning_rate": 0.0005, - "loss": 1.0578, + "learning_rate": 0.0002, + "loss": 1.0862, "step": 1083 }, { "epoch": 1.18, - "learning_rate": 0.0005, - "loss": 1.0045, + "learning_rate": 0.0002, + "loss": 1.0272, "step": 1084 }, { "epoch": 1.18, - "learning_rate": 0.0005, - "loss": 1.0284, + "learning_rate": 0.0002, + "loss": 1.0487, "step": 1085 }, { "epoch": 1.18, - "learning_rate": 0.0005, - "loss": 1.0342, + "learning_rate": 0.0002, + "loss": 1.0642, "step": 1086 }, { "epoch": 1.18, - "learning_rate": 0.0005, - "loss": 1.0729, + "learning_rate": 0.0002, + "loss": 1.092, "step": 1087 }, { "epoch": 1.18, - "learning_rate": 0.0005, - "loss": 1.0833, + "learning_rate": 0.0002, + "loss": 1.1027, "step": 1088 }, { "epoch": 1.18, - "learning_rate": 0.0005, - "loss": 1.0798, + "learning_rate": 0.0002, + "loss": 1.1004, "step": 1089 }, { "epoch": 1.19, - "learning_rate": 0.0005, - "loss": 1.057, + "learning_rate": 0.0002, + "loss": 1.0805, "step": 1090 }, { "epoch": 1.19, - "learning_rate": 0.0005, - "loss": 1.0654, + "learning_rate": 0.0002, + "loss": 1.0853, "step": 1091 }, { "epoch": 1.19, - "learning_rate": 0.0005, - "loss": 1.0629, + "learning_rate": 0.0002, + "loss": 1.1023, "step": 1092 }, { "epoch": 1.19, - "learning_rate": 0.0005, - "loss": 1.0981, + "learning_rate": 0.0002, + "loss": 1.1154, "step": 1093 }, { "epoch": 1.19, - "learning_rate": 0.0005, - "loss": 1.0692, + "learning_rate": 0.0002, + "loss": 1.0868, "step": 1094 }, { "epoch": 1.19, - "learning_rate": 0.0005, - "loss": 1.0578, + "learning_rate": 0.0002, + "loss": 1.0719, "step": 1095 }, { "epoch": 1.19, - "learning_rate": 0.0005, - "loss": 1.0683, + "learning_rate": 0.0002, + "loss": 1.0968, "step": 1096 }, { "epoch": 1.19, - "learning_rate": 0.0005, - "loss": 1.0889, + "learning_rate": 0.0002, + "loss": 1.1058, "step": 1097 }, { "epoch": 1.19, - "learning_rate": 0.0005, - "loss": 1.0578, + "learning_rate": 0.0002, + "loss": 1.0893, "step": 1098 }, { "epoch": 1.2, - "learning_rate": 0.0005, - "loss": 1.0944, + "learning_rate": 0.0002, + "loss": 1.1195, "step": 1099 }, { "epoch": 1.2, - "learning_rate": 0.0005, - "loss": 1.0321, + "learning_rate": 0.0002, + "loss": 1.0469, "step": 1100 }, { "epoch": 1.2, - "learning_rate": 0.0005, - "loss": 1.0346, + "learning_rate": 0.0002, + "loss": 1.0604, "step": 1101 }, { "epoch": 1.2, - "learning_rate": 0.0005, - "loss": 1.0643, + "learning_rate": 0.0002, + "loss": 1.0921, "step": 1102 }, { "epoch": 1.2, - "learning_rate": 0.0005, - "loss": 1.0597, + "learning_rate": 0.0002, + "loss": 1.087, "step": 1103 }, { "epoch": 1.2, - "learning_rate": 0.0005, - "loss": 1.079, + "learning_rate": 0.0002, + "loss": 1.1023, "step": 1104 }, { "epoch": 1.2, - "learning_rate": 0.0005, - "loss": 1.0648, + "learning_rate": 0.0002, + "loss": 1.0916, "step": 1105 }, { "epoch": 1.2, - "learning_rate": 0.0005, - "loss": 1.0313, + "learning_rate": 0.0002, + "loss": 1.0526, "step": 1106 }, { "epoch": 1.2, - "learning_rate": 0.0005, - "loss": 1.0512, + "learning_rate": 0.0002, + "loss": 1.07, "step": 1107 }, { "epoch": 1.21, - "learning_rate": 0.0005, - "loss": 1.0734, + "learning_rate": 0.0002, + "loss": 1.0964, "step": 1108 }, { "epoch": 1.21, - "learning_rate": 0.0005, - "loss": 1.1025, + "learning_rate": 0.0002, + "loss": 1.128, "step": 1109 }, { "epoch": 1.21, - "learning_rate": 0.0005, - "loss": 1.0752, + "learning_rate": 0.0002, + "loss": 1.0928, "step": 1110 }, { "epoch": 1.21, - "learning_rate": 0.0005, - "loss": 1.0338, + "learning_rate": 0.0002, + "loss": 1.0533, "step": 1111 }, { "epoch": 1.21, - "learning_rate": 0.0005, - "loss": 1.0582, + "learning_rate": 0.0002, + "loss": 1.0848, "step": 1112 }, { "epoch": 1.21, - "learning_rate": 0.0005, - "loss": 1.0729, + "learning_rate": 0.0002, + "loss": 1.1033, "step": 1113 }, { "epoch": 1.21, - "learning_rate": 0.0005, - "loss": 1.102, + "learning_rate": 0.0002, + "loss": 1.132, "step": 1114 }, { "epoch": 1.21, - "learning_rate": 0.0005, - "loss": 1.0787, + "learning_rate": 0.0002, + "loss": 1.1041, "step": 1115 }, { "epoch": 1.21, - "learning_rate": 0.0005, - "loss": 1.0914, + "learning_rate": 0.0002, + "loss": 1.1144, "step": 1116 }, { "epoch": 1.22, - "learning_rate": 0.0005, - "loss": 1.0752, + "learning_rate": 0.0002, + "loss": 1.1015, "step": 1117 }, { "epoch": 1.22, - "learning_rate": 0.0005, - "loss": 1.0648, + "learning_rate": 0.0002, + "loss": 1.0891, "step": 1118 }, { "epoch": 1.22, - "learning_rate": 0.0005, - "loss": 1.0307, + "learning_rate": 0.0002, + "loss": 1.0572, "step": 1119 }, { "epoch": 1.22, - "learning_rate": 0.0005, - "loss": 1.0335, + "learning_rate": 0.0002, + "loss": 1.0476, "step": 1120 }, { "epoch": 1.22, - "learning_rate": 0.0005, - "loss": 1.0107, + "learning_rate": 0.0002, + "loss": 1.0274, "step": 1121 }, { "epoch": 1.22, - "learning_rate": 0.0005, - "loss": 1.0614, + "learning_rate": 0.0002, + "loss": 1.0873, "step": 1122 }, { "epoch": 1.22, - "learning_rate": 0.0005, - "loss": 1.0587, + "learning_rate": 0.0002, + "loss": 1.0774, "step": 1123 }, { "epoch": 1.22, - "learning_rate": 0.0005, - "loss": 1.1154, + "learning_rate": 0.0002, + "loss": 1.1324, "step": 1124 }, { "epoch": 1.22, - "learning_rate": 0.0005, - "loss": 1.0476, + "learning_rate": 0.0002, + "loss": 1.0705, "step": 1125 }, { "epoch": 1.23, - "learning_rate": 0.0005, - "loss": 1.0971, + "learning_rate": 0.0002, + "loss": 1.1202, "step": 1126 }, { "epoch": 1.23, - "learning_rate": 0.0005, - "loss": 1.0974, + "learning_rate": 0.0002, + "loss": 1.1173, "step": 1127 }, { "epoch": 1.23, - "learning_rate": 0.0005, - "loss": 1.0648, + "learning_rate": 0.0002, + "loss": 1.0876, "step": 1128 }, { "epoch": 1.23, - "learning_rate": 0.0005, - "loss": 1.03, + "learning_rate": 0.0002, + "loss": 1.0518, "step": 1129 }, { "epoch": 1.23, - "learning_rate": 0.0005, - "loss": 1.0515, + "learning_rate": 0.0002, + "loss": 1.0797, "step": 1130 }, { "epoch": 1.23, - "learning_rate": 0.0005, - "loss": 1.071, + "learning_rate": 0.0002, + "loss": 1.0936, "step": 1131 }, { "epoch": 1.23, - "learning_rate": 0.0005, - "loss": 1.0464, + "learning_rate": 0.0002, + "loss": 1.0691, "step": 1132 }, { "epoch": 1.23, - "learning_rate": 0.0005, - "loss": 1.0593, + "learning_rate": 0.0002, + "loss": 1.0802, "step": 1133 }, { "epoch": 1.23, - "learning_rate": 0.0005, - "loss": 1.0827, + "learning_rate": 0.0002, + "loss": 1.109, "step": 1134 }, { "epoch": 1.24, - "learning_rate": 0.0005, - "loss": 1.084, + "learning_rate": 0.0002, + "loss": 1.122, "step": 1135 }, { "epoch": 1.24, - "learning_rate": 0.0005, - "loss": 1.0582, + "learning_rate": 0.0002, + "loss": 1.0838, "step": 1136 }, { "epoch": 1.24, - "learning_rate": 0.0005, - "loss": 1.0828, + "learning_rate": 0.0002, + "loss": 1.1001, "step": 1137 }, { "epoch": 1.24, - "learning_rate": 0.0005, - "loss": 1.0359, + "learning_rate": 0.0002, + "loss": 1.0584, "step": 1138 }, { "epoch": 1.24, - "learning_rate": 0.0005, - "loss": 1.04, + "learning_rate": 0.0002, + "loss": 1.0653, "step": 1139 }, { "epoch": 1.24, - "learning_rate": 0.0005, - "loss": 1.0821, + "learning_rate": 0.0002, + "loss": 1.1073, "step": 1140 }, { "epoch": 1.24, - "learning_rate": 0.0005, - "loss": 1.055, + "learning_rate": 0.0002, + "loss": 1.0886, "step": 1141 }, { "epoch": 1.24, - "learning_rate": 0.0005, - "loss": 1.0494, + "learning_rate": 0.0002, + "loss": 1.0699, "step": 1142 }, { "epoch": 1.24, - "learning_rate": 0.0005, - "loss": 1.0516, + "learning_rate": 0.0002, + "loss": 1.0669, "step": 1143 }, { "epoch": 1.24, - "learning_rate": 0.0005, - "loss": 1.0399, + "learning_rate": 0.0002, + "loss": 1.0615, "step": 1144 }, { "epoch": 1.25, - "learning_rate": 0.0005, - "loss": 1.0859, + "learning_rate": 0.0002, + "loss": 1.1159, "step": 1145 }, { "epoch": 1.25, - "learning_rate": 0.0005, - "loss": 1.0906, + "learning_rate": 0.0002, + "loss": 1.1094, "step": 1146 }, { "epoch": 1.25, - "learning_rate": 0.0005, - "loss": 1.0176, + "learning_rate": 0.0002, + "loss": 1.048, "step": 1147 }, { "epoch": 1.25, - "learning_rate": 0.0005, - "loss": 1.0895, + "learning_rate": 0.0002, + "loss": 1.1154, "step": 1148 }, { "epoch": 1.25, - "learning_rate": 0.0005, - "loss": 1.1346, + "learning_rate": 0.0002, + "loss": 1.1598, "step": 1149 }, { "epoch": 1.25, - "learning_rate": 0.0005, - "loss": 0.9874, + "learning_rate": 0.0002, + "loss": 1.003, "step": 1150 }, { "epoch": 1.25, - "learning_rate": 0.0005, - "loss": 1.0728, + "learning_rate": 0.0002, + "loss": 1.0937, "step": 1151 }, { "epoch": 1.25, - "learning_rate": 0.0005, - "loss": 1.088, + "learning_rate": 0.0002, + "loss": 1.1144, "step": 1152 }, { "epoch": 1.25, - "learning_rate": 0.0005, - "loss": 1.0471, + "learning_rate": 0.0002, + "loss": 1.0818, "step": 1153 }, { "epoch": 1.26, - "learning_rate": 0.0005, - "loss": 1.0449, + "learning_rate": 0.0002, + "loss": 1.0709, "step": 1154 }, { "epoch": 1.26, - "learning_rate": 0.0005, - "loss": 1.0969, + "learning_rate": 0.0002, + "loss": 1.121, "step": 1155 }, { "epoch": 1.26, - "learning_rate": 0.0005, - "loss": 1.0267, + "learning_rate": 0.0002, + "loss": 1.0539, "step": 1156 }, { "epoch": 1.26, - "learning_rate": 0.0005, - "loss": 1.0707, + "learning_rate": 0.0002, + "loss": 1.091, "step": 1157 }, { "epoch": 1.26, - "learning_rate": 0.0005, - "loss": 1.0641, + "learning_rate": 0.0002, + "loss": 1.0911, "step": 1158 }, { "epoch": 1.26, - "learning_rate": 0.0005, - "loss": 1.0523, + "learning_rate": 0.0002, + "loss": 1.0768, "step": 1159 }, { "epoch": 1.26, - "learning_rate": 0.0005, - "loss": 1.0956, + "learning_rate": 0.0002, + "loss": 1.1184, "step": 1160 }, { "epoch": 1.26, - "learning_rate": 0.0005, - "loss": 1.0258, + "learning_rate": 0.0002, + "loss": 1.0584, "step": 1161 }, { "epoch": 1.26, - "learning_rate": 0.0005, - "loss": 1.0777, + "learning_rate": 0.0002, + "loss": 1.0985, "step": 1162 }, { "epoch": 1.27, - "learning_rate": 0.0005, - "loss": 1.0976, + "learning_rate": 0.0002, + "loss": 1.1171, "step": 1163 }, { "epoch": 1.27, - "learning_rate": 0.0005, - "loss": 1.0987, + "learning_rate": 0.0002, + "loss": 1.117, "step": 1164 }, { "epoch": 1.27, - "learning_rate": 0.0005, - "loss": 0.9969, + "learning_rate": 0.0002, + "loss": 1.0154, "step": 1165 }, { "epoch": 1.27, - "learning_rate": 0.0005, - "loss": 1.1103, + "learning_rate": 0.0002, + "loss": 1.1356, "step": 1166 }, { "epoch": 1.27, - "learning_rate": 0.0005, - "loss": 1.055, + "learning_rate": 0.0002, + "loss": 1.0843, "step": 1167 }, { "epoch": 1.27, - "learning_rate": 0.0005, - "loss": 1.0698, + "learning_rate": 0.0002, + "loss": 1.1001, "step": 1168 }, { "epoch": 1.27, - "learning_rate": 0.0005, - "loss": 1.0972, + "learning_rate": 0.0002, + "loss": 1.1239, "step": 1169 }, { "epoch": 1.27, - "learning_rate": 0.0005, - "loss": 1.0393, + "learning_rate": 0.0002, + "loss": 1.0623, "step": 1170 }, { "epoch": 1.27, - "learning_rate": 0.0005, - "loss": 1.0385, + "learning_rate": 0.0002, + "loss": 1.0628, "step": 1171 }, { "epoch": 1.28, - "learning_rate": 0.0005, - "loss": 1.0801, + "learning_rate": 0.0002, + "loss": 1.1121, "step": 1172 }, { "epoch": 1.28, - "learning_rate": 0.0005, - "loss": 1.034, + "learning_rate": 0.0002, + "loss": 1.0633, "step": 1173 }, { "epoch": 1.28, - "learning_rate": 0.0005, - "loss": 1.079, + "learning_rate": 0.0002, + "loss": 1.1103, "step": 1174 }, { "epoch": 1.28, - "learning_rate": 0.0005, - "loss": 1.0459, + "learning_rate": 0.0002, + "loss": 1.0601, "step": 1175 }, { "epoch": 1.28, - "learning_rate": 0.0005, - "loss": 1.0632, + "learning_rate": 0.0002, + "loss": 1.0771, "step": 1176 }, { "epoch": 1.28, - "learning_rate": 0.0005, - "loss": 1.0375, + "learning_rate": 0.0002, + "loss": 1.0639, "step": 1177 }, { "epoch": 1.28, - "learning_rate": 0.0005, - "loss": 1.0229, + "learning_rate": 0.0002, + "loss": 1.0457, "step": 1178 }, { "epoch": 1.28, - "learning_rate": 0.0005, - "loss": 1.0751, + "learning_rate": 0.0002, + "loss": 1.1017, "step": 1179 }, { "epoch": 1.28, - "learning_rate": 0.0005, - "loss": 1.0328, + "learning_rate": 0.0002, + "loss": 1.0558, "step": 1180 }, { "epoch": 1.29, - "learning_rate": 0.0005, - "loss": 1.0652, + "learning_rate": 0.0002, + "loss": 1.0853, "step": 1181 }, { "epoch": 1.29, - "learning_rate": 0.0005, - "loss": 1.0632, + "learning_rate": 0.0002, + "loss": 1.0854, "step": 1182 }, { "epoch": 1.29, - "learning_rate": 0.0005, - "loss": 1.1036, + "learning_rate": 0.0002, + "loss": 1.1233, "step": 1183 }, { "epoch": 1.29, - "learning_rate": 0.0005, - "loss": 1.0792, + "learning_rate": 0.0002, + "loss": 1.0982, "step": 1184 }, { "epoch": 1.29, - "learning_rate": 0.0005, - "loss": 1.0104, + "learning_rate": 0.0002, + "loss": 1.0383, "step": 1185 }, { "epoch": 1.29, - "learning_rate": 0.0005, - "loss": 1.1514, + "learning_rate": 0.0002, + "loss": 1.1803, "step": 1186 }, { "epoch": 1.29, - "learning_rate": 0.0005, - "loss": 1.0525, + "learning_rate": 0.0002, + "loss": 1.0721, "step": 1187 }, { "epoch": 1.29, - "learning_rate": 0.0005, - "loss": 1.0701, + "learning_rate": 0.0002, + "loss": 1.0981, "step": 1188 }, { "epoch": 1.29, - "learning_rate": 0.0005, - "loss": 1.0327, + "learning_rate": 0.0002, + "loss": 1.0619, "step": 1189 }, { "epoch": 1.29, - "learning_rate": 0.0005, - "loss": 1.0686, + "learning_rate": 0.0002, + "loss": 1.0889, "step": 1190 }, { "epoch": 1.3, - "learning_rate": 0.0005, - "loss": 1.0711, + "learning_rate": 0.0002, + "loss": 1.0929, "step": 1191 }, { "epoch": 1.3, - "learning_rate": 0.0005, - "loss": 1.013, + "learning_rate": 0.0002, + "loss": 1.0375, "step": 1192 }, { "epoch": 1.3, - "learning_rate": 0.0005, - "loss": 1.1186, + "learning_rate": 0.0002, + "loss": 1.1453, "step": 1193 }, { "epoch": 1.3, - "learning_rate": 0.0005, - "loss": 1.0468, + "learning_rate": 0.0002, + "loss": 1.0716, "step": 1194 }, { "epoch": 1.3, - "learning_rate": 0.0005, - "loss": 1.0736, + "learning_rate": 0.0002, + "loss": 1.0961, "step": 1195 }, { "epoch": 1.3, - "learning_rate": 0.0005, - "loss": 1.1029, + "learning_rate": 0.0002, + "loss": 1.1293, "step": 1196 }, { "epoch": 1.3, - "learning_rate": 0.0005, - "loss": 1.1157, + "learning_rate": 0.0002, + "loss": 1.1314, "step": 1197 }, { "epoch": 1.3, - "learning_rate": 0.0005, - "loss": 1.0616, + "learning_rate": 0.0002, + "loss": 1.0868, "step": 1198 }, { "epoch": 1.3, - "learning_rate": 0.0005, - "loss": 1.0653, + "learning_rate": 0.0002, + "loss": 1.0942, "step": 1199 }, { "epoch": 1.31, - "learning_rate": 0.0005, - "loss": 1.0215, + "learning_rate": 0.0002, + "loss": 1.0454, "step": 1200 }, { "epoch": 1.31, - "learning_rate": 0.0005, - "loss": 1.0674, + "learning_rate": 0.0002, + "loss": 1.0816, "step": 1201 }, { "epoch": 1.31, - "learning_rate": 0.0005, - "loss": 1.0615, + "learning_rate": 0.0002, + "loss": 1.0902, "step": 1202 }, { "epoch": 1.31, - "learning_rate": 0.0005, - "loss": 1.0367, + "learning_rate": 0.0002, + "loss": 1.057, "step": 1203 }, { "epoch": 1.31, - "learning_rate": 0.0005, - "loss": 1.0976, + "learning_rate": 0.0002, + "loss": 1.1205, "step": 1204 }, { "epoch": 1.31, - "learning_rate": 0.0005, - "loss": 1.075, + "learning_rate": 0.0002, + "loss": 1.0991, "step": 1205 }, { "epoch": 1.31, - "learning_rate": 0.0005, - "loss": 1.0679, + "learning_rate": 0.0002, + "loss": 1.0898, "step": 1206 }, { "epoch": 1.31, - "learning_rate": 0.0005, - "loss": 1.0133, + "learning_rate": 0.0002, + "loss": 1.0455, "step": 1207 }, { "epoch": 1.31, - "learning_rate": 0.0005, - "loss": 1.0781, + "learning_rate": 0.0002, + "loss": 1.1118, "step": 1208 }, { "epoch": 1.32, - "learning_rate": 0.0005, - "loss": 1.0093, + "learning_rate": 0.0002, + "loss": 1.0345, "step": 1209 }, { "epoch": 1.32, - "learning_rate": 0.0005, - "loss": 1.085, + "learning_rate": 0.0002, + "loss": 1.1066, "step": 1210 }, { "epoch": 1.32, - "learning_rate": 0.0005, - "loss": 1.082, + "learning_rate": 0.0002, + "loss": 1.1073, "step": 1211 }, { "epoch": 1.32, - "learning_rate": 0.0005, - "loss": 1.0458, + "learning_rate": 0.0002, + "loss": 1.059, "step": 1212 }, { "epoch": 1.32, - "learning_rate": 0.0005, - "loss": 1.0701, + "learning_rate": 0.0002, + "loss": 1.0998, "step": 1213 }, { "epoch": 1.32, - "learning_rate": 0.0005, - "loss": 1.0318, + "learning_rate": 0.0002, + "loss": 1.0524, "step": 1214 }, { "epoch": 1.32, - "learning_rate": 0.0005, - "loss": 1.0882, + "learning_rate": 0.0002, + "loss": 1.1186, "step": 1215 }, { "epoch": 1.32, - "learning_rate": 0.0005, - "loss": 1.0086, + "learning_rate": 0.0002, + "loss": 1.0225, "step": 1216 }, { "epoch": 1.32, - "learning_rate": 0.0005, - "loss": 1.0535, + "learning_rate": 0.0002, + "loss": 1.0739, "step": 1217 }, { "epoch": 1.33, - "learning_rate": 0.0005, - "loss": 1.068, + "learning_rate": 0.0002, + "loss": 1.0997, "step": 1218 }, { "epoch": 1.33, - "learning_rate": 0.0005, - "loss": 1.0198, + "learning_rate": 0.0002, + "loss": 1.0369, "step": 1219 }, { "epoch": 1.33, - "learning_rate": 0.0005, - "loss": 1.0786, + "learning_rate": 0.0002, + "loss": 1.1186, "step": 1220 }, { "epoch": 1.33, - "learning_rate": 0.0005, - "loss": 1.0328, + "learning_rate": 0.0002, + "loss": 1.0557, "step": 1221 }, { "epoch": 1.33, - "learning_rate": 0.0005, - "loss": 1.0948, + "learning_rate": 0.0002, + "loss": 1.1197, "step": 1222 }, { "epoch": 1.33, - "learning_rate": 0.0005, - "loss": 1.0813, + "learning_rate": 0.0002, + "loss": 1.0968, "step": 1223 }, { "epoch": 1.33, - "learning_rate": 0.0005, - "loss": 1.0312, + "learning_rate": 0.0002, + "loss": 1.0576, "step": 1224 }, { "epoch": 1.33, - "learning_rate": 0.0005, - "loss": 1.0704, + "learning_rate": 0.0002, + "loss": 1.0912, "step": 1225 }, { "epoch": 1.33, - "learning_rate": 0.0005, - "loss": 1.0442, + "learning_rate": 0.0002, + "loss": 1.0796, "step": 1226 }, { "epoch": 1.34, - "learning_rate": 0.0005, - "loss": 1.0691, + "learning_rate": 0.0002, + "loss": 1.0879, "step": 1227 }, { "epoch": 1.34, - "learning_rate": 0.0005, - "loss": 1.094, + "learning_rate": 0.0002, + "loss": 1.1195, "step": 1228 }, { "epoch": 1.34, - "learning_rate": 0.0005, - "loss": 1.0345, + "learning_rate": 0.0002, + "loss": 1.0606, "step": 1229 }, { "epoch": 1.34, - "learning_rate": 0.0005, - "loss": 1.0822, + "learning_rate": 0.0002, + "loss": 1.1071, "step": 1230 }, { "epoch": 1.34, - "learning_rate": 0.0005, - "loss": 1.0984, + "learning_rate": 0.0002, + "loss": 1.1185, "step": 1231 }, { "epoch": 1.34, - "learning_rate": 0.0005, - "loss": 1.0884, + "learning_rate": 0.0002, + "loss": 1.1008, "step": 1232 }, { "epoch": 1.34, - "learning_rate": 0.0005, - "loss": 1.0661, + "learning_rate": 0.0002, + "loss": 1.0889, "step": 1233 }, { "epoch": 1.34, - "learning_rate": 0.0005, - "loss": 1.0907, + "learning_rate": 0.0002, + "loss": 1.1162, "step": 1234 }, { "epoch": 1.34, - "learning_rate": 0.0005, - "loss": 1.1057, + "learning_rate": 0.0002, + "loss": 1.1265, "step": 1235 }, { "epoch": 1.34, - "learning_rate": 0.0005, - "loss": 1.0705, + "learning_rate": 0.0002, + "loss": 1.0949, "step": 1236 }, { "epoch": 1.35, - "learning_rate": 0.0005, - "loss": 1.0246, + "learning_rate": 0.0002, + "loss": 1.0559, "step": 1237 }, { "epoch": 1.35, - "learning_rate": 0.0005, - "loss": 1.0649, + "learning_rate": 0.0002, + "loss": 1.0914, "step": 1238 }, { "epoch": 1.35, - "learning_rate": 0.0005, - "loss": 1.095, + "learning_rate": 0.0002, + "loss": 1.1171, "step": 1239 }, { "epoch": 1.35, - "learning_rate": 0.0005, - "loss": 1.0966, + "learning_rate": 0.0002, + "loss": 1.1155, "step": 1240 }, { "epoch": 1.35, - "learning_rate": 0.0005, - "loss": 1.0393, + "learning_rate": 0.0002, + "loss": 1.0704, "step": 1241 }, { "epoch": 1.35, - "learning_rate": 0.0005, - "loss": 1.0895, + "learning_rate": 0.0002, + "loss": 1.1076, "step": 1242 }, { "epoch": 1.35, - "learning_rate": 0.0005, - "loss": 1.0684, + "learning_rate": 0.0002, + "loss": 1.1009, "step": 1243 }, { "epoch": 1.35, - "learning_rate": 0.0005, - "loss": 1.0556, + "learning_rate": 0.0002, + "loss": 1.0726, "step": 1244 }, { "epoch": 1.35, - "learning_rate": 0.0005, - "loss": 1.0403, + "learning_rate": 0.0002, + "loss": 1.064, "step": 1245 }, { "epoch": 1.36, - "learning_rate": 0.0005, - "loss": 1.1037, + "learning_rate": 0.0002, + "loss": 1.1263, "step": 1246 }, { "epoch": 1.36, - "learning_rate": 0.0005, - "loss": 1.0558, + "learning_rate": 0.0002, + "loss": 1.0847, "step": 1247 }, { "epoch": 1.36, - "learning_rate": 0.0005, - "loss": 1.0618, + "learning_rate": 0.0002, + "loss": 1.0938, "step": 1248 }, { "epoch": 1.36, - "learning_rate": 0.0005, - "loss": 1.1237, + "learning_rate": 0.0002, + "loss": 1.1514, "step": 1249 }, { "epoch": 1.36, - "learning_rate": 0.0005, - "loss": 1.035, + "learning_rate": 0.0002, + "loss": 1.0542, "step": 1250 }, { "epoch": 1.36, - "learning_rate": 0.0005, - "loss": 1.0771, + "learning_rate": 0.0002, + "loss": 1.0994, "step": 1251 }, { "epoch": 1.36, - "learning_rate": 0.0005, - "loss": 1.1024, + "learning_rate": 0.0002, + "loss": 1.1199, "step": 1252 }, { "epoch": 1.36, - "learning_rate": 0.0005, - "loss": 1.0719, + "learning_rate": 0.0002, + "loss": 1.091, "step": 1253 }, { "epoch": 1.36, - "learning_rate": 0.0005, - "loss": 1.0476, + "learning_rate": 0.0002, + "loss": 1.0676, "step": 1254 }, { "epoch": 1.37, - "learning_rate": 0.0005, - "loss": 1.1005, + "learning_rate": 0.0002, + "loss": 1.1268, "step": 1255 }, { "epoch": 1.37, - "learning_rate": 0.0005, - "loss": 1.0351, + "learning_rate": 0.0002, + "loss": 1.0487, "step": 1256 }, { "epoch": 1.37, - "learning_rate": 0.0005, - "loss": 1.068, + "learning_rate": 0.0002, + "loss": 1.0935, "step": 1257 }, { "epoch": 1.37, - "learning_rate": 0.0005, - "loss": 1.0712, + "learning_rate": 0.0002, + "loss": 1.105, "step": 1258 }, { "epoch": 1.37, - "learning_rate": 0.0005, - "loss": 1.0786, + "learning_rate": 0.0002, + "loss": 1.1004, "step": 1259 }, { "epoch": 1.37, - "learning_rate": 0.0005, - "loss": 1.0289, + "learning_rate": 0.0002, + "loss": 1.0579, "step": 1260 }, { "epoch": 1.37, - "learning_rate": 0.0005, - "loss": 1.0623, + "learning_rate": 0.0002, + "loss": 1.0755, "step": 1261 }, { "epoch": 1.37, - "learning_rate": 0.0005, - "loss": 1.0388, + "learning_rate": 0.0002, + "loss": 1.0667, "step": 1262 }, { "epoch": 1.37, - "learning_rate": 0.0005, - "loss": 1.0722, + "learning_rate": 0.0002, + "loss": 1.0912, "step": 1263 }, { "epoch": 1.38, - "learning_rate": 0.0005, - "loss": 1.0128, + "learning_rate": 0.0002, + "loss": 1.0353, "step": 1264 }, { "epoch": 1.38, - "learning_rate": 0.0005, - "loss": 1.0787, + "learning_rate": 0.0002, + "loss": 1.1045, "step": 1265 }, { "epoch": 1.38, - "learning_rate": 0.0005, - "loss": 1.0822, + "learning_rate": 0.0002, + "loss": 1.0957, "step": 1266 }, { "epoch": 1.38, - "learning_rate": 0.0005, - "loss": 1.0747, + "learning_rate": 0.0002, + "loss": 1.0931, "step": 1267 }, { "epoch": 1.38, - "learning_rate": 0.0005, - "loss": 0.9848, + "learning_rate": 0.0002, + "loss": 1.0061, "step": 1268 }, { "epoch": 1.38, - "learning_rate": 0.0005, - "loss": 1.0504, + "learning_rate": 0.0002, + "loss": 1.074, "step": 1269 }, { "epoch": 1.38, - "learning_rate": 0.0005, - "loss": 1.0539, + "learning_rate": 0.0002, + "loss": 1.0835, "step": 1270 }, { "epoch": 1.38, - "learning_rate": 0.0005, - "loss": 1.088, + "learning_rate": 0.0002, + "loss": 1.1189, "step": 1271 }, { "epoch": 1.38, - "learning_rate": 0.0005, - "loss": 1.0466, + "learning_rate": 0.0002, + "loss": 1.0725, "step": 1272 }, { "epoch": 1.39, - "learning_rate": 0.0005, - "loss": 1.0625, + "learning_rate": 0.0002, + "loss": 1.09, "step": 1273 }, { "epoch": 1.39, - "learning_rate": 0.0005, - "loss": 1.1106, + "learning_rate": 0.0002, + "loss": 1.1374, "step": 1274 }, { "epoch": 1.39, - "learning_rate": 0.0005, - "loss": 1.1052, + "learning_rate": 0.0002, + "loss": 1.1292, "step": 1275 }, { "epoch": 1.39, - "learning_rate": 0.0005, - "loss": 1.0933, + "learning_rate": 0.0002, + "loss": 1.1122, "step": 1276 }, { "epoch": 1.39, - "learning_rate": 0.0005, - "loss": 1.0791, + "learning_rate": 0.0002, + "loss": 1.099, "step": 1277 }, { "epoch": 1.39, - "learning_rate": 0.0005, - "loss": 1.0338, + "learning_rate": 0.0002, + "loss": 1.0576, "step": 1278 }, { "epoch": 1.39, - "learning_rate": 0.0005, - "loss": 1.0017, + "learning_rate": 0.0002, + "loss": 1.0325, "step": 1279 }, { "epoch": 1.39, - "learning_rate": 0.0005, - "loss": 1.0188, + "learning_rate": 0.0002, + "loss": 1.0468, "step": 1280 }, { "epoch": 1.39, - "learning_rate": 0.0005, - "loss": 1.055, + "learning_rate": 0.0002, + "loss": 1.089, "step": 1281 }, { "epoch": 1.39, - "learning_rate": 0.0005, - "loss": 1.0607, + "learning_rate": 0.0002, + "loss": 1.089, "step": 1282 }, { "epoch": 1.4, - "learning_rate": 0.0005, - "loss": 1.0479, + "learning_rate": 0.0002, + "loss": 1.0762, "step": 1283 }, { "epoch": 1.4, - "learning_rate": 0.0005, - "loss": 1.0837, + "learning_rate": 0.0002, + "loss": 1.0964, "step": 1284 }, { "epoch": 1.4, - "learning_rate": 0.0005, - "loss": 1.0498, + "learning_rate": 0.0002, + "loss": 1.068, "step": 1285 }, { "epoch": 1.4, - "learning_rate": 0.0005, - "loss": 1.064, + "learning_rate": 0.0002, + "loss": 1.0826, "step": 1286 }, { "epoch": 1.4, - "learning_rate": 0.0005, - "loss": 1.0856, + "learning_rate": 0.0002, + "loss": 1.1005, "step": 1287 }, { "epoch": 1.4, - "learning_rate": 0.0005, - "loss": 1.0653, + "learning_rate": 0.0002, + "loss": 1.0857, "step": 1288 }, { "epoch": 1.4, - "learning_rate": 0.0005, - "loss": 1.0629, + "learning_rate": 0.0002, + "loss": 1.0875, "step": 1289 }, { "epoch": 1.4, - "learning_rate": 0.0005, - "loss": 1.059, + "learning_rate": 0.0002, + "loss": 1.0721, "step": 1290 }, { "epoch": 1.4, - "learning_rate": 0.0005, - "loss": 1.0581, + "learning_rate": 0.0002, + "loss": 1.0804, "step": 1291 }, { "epoch": 1.41, - "learning_rate": 0.0005, - "loss": 1.0526, + "learning_rate": 0.0002, + "loss": 1.0739, "step": 1292 }, { "epoch": 1.41, - "learning_rate": 0.0005, - "loss": 1.0702, + "learning_rate": 0.0002, + "loss": 1.0913, "step": 1293 }, { "epoch": 1.41, - "learning_rate": 0.0005, - "loss": 1.0553, + "learning_rate": 0.0002, + "loss": 1.0666, "step": 1294 }, { "epoch": 1.41, - "learning_rate": 0.0005, - "loss": 1.0037, + "learning_rate": 0.0002, + "loss": 1.0323, "step": 1295 }, { "epoch": 1.41, - "learning_rate": 0.0005, - "loss": 1.0772, + "learning_rate": 0.0002, + "loss": 1.0968, "step": 1296 }, { "epoch": 1.41, - "learning_rate": 0.0005, - "loss": 1.0831, + "learning_rate": 0.0002, + "loss": 1.1019, "step": 1297 }, { "epoch": 1.41, - "learning_rate": 0.0005, - "loss": 1.075, + "learning_rate": 0.0002, + "loss": 1.0966, "step": 1298 }, { "epoch": 1.41, - "learning_rate": 0.0005, - "loss": 1.0583, + "learning_rate": 0.0002, + "loss": 1.0713, "step": 1299 }, { "epoch": 1.41, - "learning_rate": 0.0005, - "loss": 1.0555, + "learning_rate": 0.0002, + "loss": 1.0781, "step": 1300 }, { "epoch": 1.42, - "learning_rate": 0.0005, - "loss": 1.0598, + "learning_rate": 0.0002, + "loss": 1.0805, "step": 1301 }, { "epoch": 1.42, - "learning_rate": 0.0005, - "loss": 1.0815, + "learning_rate": 0.0002, + "loss": 1.1085, "step": 1302 }, { "epoch": 1.42, - "learning_rate": 0.0005, - "loss": 1.0179, + "learning_rate": 0.0002, + "loss": 1.0381, "step": 1303 }, { "epoch": 1.42, - "learning_rate": 0.0005, - "loss": 1.0668, + "learning_rate": 0.0002, + "loss": 1.0907, "step": 1304 }, { "epoch": 1.42, - "learning_rate": 0.0005, - "loss": 1.0583, + "learning_rate": 0.0002, + "loss": 1.0887, "step": 1305 }, { "epoch": 1.42, - "learning_rate": 0.0005, - "loss": 1.0852, + "learning_rate": 0.0002, + "loss": 1.0966, "step": 1306 }, { "epoch": 1.42, - "learning_rate": 0.0005, - "loss": 1.0491, + "learning_rate": 0.0002, + "loss": 1.0709, "step": 1307 }, { "epoch": 1.42, - "learning_rate": 0.0005, - "loss": 1.0892, + "learning_rate": 0.0002, + "loss": 1.1183, "step": 1308 }, { "epoch": 1.42, - "learning_rate": 0.0005, - "loss": 1.0711, + "learning_rate": 0.0002, + "loss": 1.099, "step": 1309 }, { "epoch": 1.43, - "learning_rate": 0.0005, - "loss": 1.0319, + "learning_rate": 0.0002, + "loss": 1.0597, "step": 1310 }, { "epoch": 1.43, - "learning_rate": 0.0005, - "loss": 1.0818, + "learning_rate": 0.0002, + "loss": 1.1089, "step": 1311 }, { "epoch": 1.43, - "learning_rate": 0.0005, - "loss": 1.0547, + "learning_rate": 0.0002, + "loss": 1.0733, "step": 1312 }, { "epoch": 1.43, - "learning_rate": 0.0005, - "loss": 1.07, + "learning_rate": 0.0002, + "loss": 1.1004, "step": 1313 }, { "epoch": 1.43, - "learning_rate": 0.0005, - "loss": 1.0604, + "learning_rate": 0.0002, + "loss": 1.0774, "step": 1314 }, { "epoch": 1.43, - "learning_rate": 0.0005, - "loss": 1.0485, + "learning_rate": 0.0002, + "loss": 1.0681, "step": 1315 }, { "epoch": 1.43, - "learning_rate": 0.0005, - "loss": 1.0233, + "learning_rate": 0.0002, + "loss": 1.0491, "step": 1316 }, { "epoch": 1.43, - "learning_rate": 0.0005, - "loss": 1.0618, + "learning_rate": 0.0002, + "loss": 1.0863, "step": 1317 }, { "epoch": 1.43, - "learning_rate": 0.0005, - "loss": 1.0451, + "learning_rate": 0.0002, + "loss": 1.0625, "step": 1318 }, { "epoch": 1.44, - "learning_rate": 0.0005, - "loss": 1.097, + "learning_rate": 0.0002, + "loss": 1.1248, "step": 1319 }, { "epoch": 1.44, - "learning_rate": 0.0005, - "loss": 1.0598, + "learning_rate": 0.0002, + "loss": 1.0906, "step": 1320 }, { "epoch": 1.44, - "learning_rate": 0.0005, - "loss": 1.0574, + "learning_rate": 0.0002, + "loss": 1.0817, "step": 1321 }, { "epoch": 1.44, - "learning_rate": 0.0005, - "loss": 1.0377, + "learning_rate": 0.0002, + "loss": 1.0543, "step": 1322 }, { "epoch": 1.44, - "learning_rate": 0.0005, - "loss": 1.0169, + "learning_rate": 0.0002, + "loss": 1.0406, "step": 1323 }, { "epoch": 1.44, - "learning_rate": 0.0005, - "loss": 1.085, + "learning_rate": 0.0002, + "loss": 1.1063, "step": 1324 }, { "epoch": 1.44, - "learning_rate": 0.0005, - "loss": 1.0952, + "learning_rate": 0.0002, + "loss": 1.1143, "step": 1325 }, { "epoch": 1.44, - "learning_rate": 0.0005, - "loss": 1.1142, + "learning_rate": 0.0002, + "loss": 1.1421, "step": 1326 }, { "epoch": 1.44, - "learning_rate": 0.0005, - "loss": 1.1141, + "learning_rate": 0.0002, + "loss": 1.1367, "step": 1327 }, { "epoch": 1.45, - "learning_rate": 0.0005, - "loss": 1.036, + "learning_rate": 0.0002, + "loss": 1.0619, "step": 1328 }, { "epoch": 1.45, - "learning_rate": 0.0005, - "loss": 1.0425, + "learning_rate": 0.0002, + "loss": 1.0681, "step": 1329 }, { "epoch": 1.45, - "learning_rate": 0.0005, - "loss": 1.0701, + "learning_rate": 0.0002, + "loss": 1.0938, "step": 1330 }, { "epoch": 1.45, - "learning_rate": 0.0005, - "loss": 1.0607, + "learning_rate": 0.0002, + "loss": 1.0754, "step": 1331 }, { "epoch": 1.45, - "learning_rate": 0.0005, - "loss": 1.0233, + "learning_rate": 0.0002, + "loss": 1.0455, "step": 1332 }, { "epoch": 1.45, - "learning_rate": 0.0005, - "loss": 1.0209, + "learning_rate": 0.0002, + "loss": 1.0456, "step": 1333 }, { "epoch": 1.45, - "learning_rate": 0.0005, - "loss": 1.0791, + "learning_rate": 0.0002, + "loss": 1.0934, "step": 1334 }, { "epoch": 1.45, - "learning_rate": 0.0005, - "loss": 1.0621, + "learning_rate": 0.0002, + "loss": 1.0865, "step": 1335 }, { "epoch": 1.45, - "learning_rate": 0.0005, - "loss": 1.0137, + "learning_rate": 0.0002, + "loss": 1.0399, "step": 1336 }, { "epoch": 1.45, - "learning_rate": 0.0005, - "loss": 1.0938, + "learning_rate": 0.0002, + "loss": 1.1112, "step": 1337 }, { "epoch": 1.46, - "learning_rate": 0.0005, - "loss": 1.1374, + "learning_rate": 0.0002, + "loss": 1.159, "step": 1338 }, { "epoch": 1.46, - "learning_rate": 0.0005, - "loss": 1.0544, + "learning_rate": 0.0002, + "loss": 1.0813, "step": 1339 }, { "epoch": 1.46, - "learning_rate": 0.0005, - "loss": 1.0733, + "learning_rate": 0.0002, + "loss": 1.1037, "step": 1340 }, { "epoch": 1.46, - "learning_rate": 0.0005, - "loss": 1.0569, + "learning_rate": 0.0002, + "loss": 1.0772, "step": 1341 }, { "epoch": 1.46, - "learning_rate": 0.0005, - "loss": 1.0882, + "learning_rate": 0.0002, + "loss": 1.1013, "step": 1342 }, { "epoch": 1.46, - "learning_rate": 0.0005, - "loss": 0.9811, + "learning_rate": 0.0002, + "loss": 0.9976, "step": 1343 }, { "epoch": 1.46, - "learning_rate": 0.0005, - "loss": 1.0249, + "learning_rate": 0.0002, + "loss": 1.0401, "step": 1344 }, { "epoch": 1.46, - "learning_rate": 0.0005, - "loss": 1.0328, + "learning_rate": 0.0002, + "loss": 1.0566, "step": 1345 }, { "epoch": 1.46, - "learning_rate": 0.0005, - "loss": 1.0661, + "learning_rate": 0.0002, + "loss": 1.0937, "step": 1346 }, { "epoch": 1.47, - "learning_rate": 0.0005, - "loss": 1.066, + "learning_rate": 0.0002, + "loss": 1.0871, "step": 1347 }, { "epoch": 1.47, - "learning_rate": 0.0005, - "loss": 1.082, + "learning_rate": 0.0002, + "loss": 1.1169, "step": 1348 }, { "epoch": 1.47, - "learning_rate": 0.0005, - "loss": 1.0913, + "learning_rate": 0.0002, + "loss": 1.1145, "step": 1349 }, { "epoch": 1.47, - "learning_rate": 0.0005, - "loss": 1.0429, + "learning_rate": 0.0002, + "loss": 1.0648, "step": 1350 }, { "epoch": 1.47, - "learning_rate": 0.0005, - "loss": 1.111, + "learning_rate": 0.0002, + "loss": 1.1332, "step": 1351 }, { "epoch": 1.47, - "learning_rate": 0.0005, - "loss": 1.1039, + "learning_rate": 0.0002, + "loss": 1.1249, "step": 1352 }, { "epoch": 1.47, - "learning_rate": 0.0005, - "loss": 1.0746, + "learning_rate": 0.0002, + "loss": 1.1006, "step": 1353 }, { "epoch": 1.47, - "learning_rate": 0.0005, - "loss": 1.0829, + "learning_rate": 0.0002, + "loss": 1.0908, "step": 1354 }, { "epoch": 1.47, - "learning_rate": 0.0005, - "loss": 1.0163, + "learning_rate": 0.0002, + "loss": 1.0343, "step": 1355 }, { "epoch": 1.48, - "learning_rate": 0.0005, - "loss": 1.0654, + "learning_rate": 0.0002, + "loss": 1.0991, "step": 1356 }, { "epoch": 1.48, - "learning_rate": 0.0005, - "loss": 1.11, + "learning_rate": 0.0002, + "loss": 1.1304, "step": 1357 }, { "epoch": 1.48, - "learning_rate": 0.0005, - "loss": 1.0651, + "learning_rate": 0.0002, + "loss": 1.0854, "step": 1358 }, { "epoch": 1.48, - "learning_rate": 0.0005, - "loss": 1.0776, + "learning_rate": 0.0002, + "loss": 1.1089, "step": 1359 }, { "epoch": 1.48, - "learning_rate": 0.0005, - "loss": 1.0499, + "learning_rate": 0.0002, + "loss": 1.0785, "step": 1360 }, { "epoch": 1.48, - "learning_rate": 0.0005, - "loss": 1.0814, + "learning_rate": 0.0002, + "loss": 1.1074, "step": 1361 }, { "epoch": 1.48, - "learning_rate": 0.0005, - "loss": 1.0874, + "learning_rate": 0.0002, + "loss": 1.1154, "step": 1362 }, { "epoch": 1.48, - "learning_rate": 0.0005, - "loss": 1.0779, + "learning_rate": 0.0002, + "loss": 1.093, "step": 1363 }, { "epoch": 1.48, - "learning_rate": 0.0005, - "loss": 1.1042, + "learning_rate": 0.0002, + "loss": 1.1167, "step": 1364 }, { "epoch": 1.49, - "learning_rate": 0.0005, - "loss": 1.0985, + "learning_rate": 0.0002, + "loss": 1.1162, "step": 1365 }, { "epoch": 1.49, - "learning_rate": 0.0005, - "loss": 1.0287, + "learning_rate": 0.0002, + "loss": 1.0555, "step": 1366 }, { "epoch": 1.49, - "learning_rate": 0.0005, - "loss": 1.0516, + "learning_rate": 0.0002, + "loss": 1.0782, "step": 1367 }, { "epoch": 1.49, - "learning_rate": 0.0005, - "loss": 1.0923, + "learning_rate": 0.0002, + "loss": 1.1213, "step": 1368 }, { "epoch": 1.49, - "learning_rate": 0.0005, - "loss": 1.0307, + "learning_rate": 0.0002, + "loss": 1.0474, "step": 1369 }, { "epoch": 1.49, - "learning_rate": 0.0005, - "loss": 1.0664, + "learning_rate": 0.0002, + "loss": 1.0954, "step": 1370 }, { "epoch": 1.49, - "learning_rate": 0.0005, - "loss": 1.057, + "learning_rate": 0.0002, + "loss": 1.0792, "step": 1371 }, { "epoch": 1.49, - "learning_rate": 0.0005, - "loss": 1.0665, + "learning_rate": 0.0002, + "loss": 1.0914, "step": 1372 }, { "epoch": 1.49, - "learning_rate": 0.0005, - "loss": 1.088, + "learning_rate": 0.0002, + "loss": 1.1135, "step": 1373 }, { "epoch": 1.5, - "learning_rate": 0.0005, - "loss": 1.0416, + "learning_rate": 0.0002, + "loss": 1.0664, "step": 1374 }, { "epoch": 1.5, - "learning_rate": 0.0005, - "loss": 1.0232, + "learning_rate": 0.0002, + "loss": 1.0437, "step": 1375 }, { "epoch": 1.5, - "learning_rate": 0.0005, - "loss": 1.084, + "learning_rate": 0.0002, + "loss": 1.111, "step": 1376 }, { "epoch": 1.5, - "learning_rate": 0.0005, - "loss": 1.0645, + "learning_rate": 0.0002, + "loss": 1.0976, "step": 1377 }, { "epoch": 1.5, - "learning_rate": 0.0005, - "loss": 1.0484, + "learning_rate": 0.0002, + "loss": 1.0706, "step": 1378 }, { "epoch": 1.5, - "learning_rate": 0.0005, - "loss": 1.0969, + "learning_rate": 0.0002, + "loss": 1.12, "step": 1379 }, { "epoch": 1.5, - "learning_rate": 0.0005, - "loss": 1.0761, + "learning_rate": 0.0002, + "loss": 1.1036, "step": 1380 }, { "epoch": 1.5, - "learning_rate": 0.0005, - "loss": 1.054, + "learning_rate": 0.0002, + "loss": 1.0773, "step": 1381 }, { "epoch": 1.5, - "learning_rate": 0.0005, - "loss": 1.0735, + "learning_rate": 0.0002, + "loss": 1.0967, "step": 1382 }, { "epoch": 1.5, - "learning_rate": 0.0005, - "loss": 1.0479, + "learning_rate": 0.0002, + "loss": 1.0688, "step": 1383 }, { "epoch": 1.51, - "learning_rate": 0.0005, - "loss": 1.0652, + "learning_rate": 0.0002, + "loss": 1.0899, "step": 1384 }, { "epoch": 1.51, - "learning_rate": 0.0005, - "loss": 1.1014, + "learning_rate": 0.0002, + "loss": 1.1195, "step": 1385 }, { "epoch": 1.51, - "learning_rate": 0.0005, - "loss": 1.093, + "learning_rate": 0.0002, + "loss": 1.1196, "step": 1386 }, { "epoch": 1.51, - "learning_rate": 0.0005, - "loss": 1.0487, + "learning_rate": 0.0002, + "loss": 1.0629, "step": 1387 }, { "epoch": 1.51, - "learning_rate": 0.0005, - "loss": 1.0555, + "learning_rate": 0.0002, + "loss": 1.0785, "step": 1388 }, { "epoch": 1.51, - "learning_rate": 0.0005, - "loss": 1.0877, + "learning_rate": 0.0002, + "loss": 1.1044, "step": 1389 }, { "epoch": 1.51, - "learning_rate": 0.0005, - "loss": 1.0894, + "learning_rate": 0.0002, + "loss": 1.1184, "step": 1390 }, { "epoch": 1.51, - "learning_rate": 0.0005, - "loss": 1.0383, + "learning_rate": 0.0002, + "loss": 1.0595, "step": 1391 }, { "epoch": 1.51, - "learning_rate": 0.0005, - "loss": 1.0867, + "learning_rate": 0.0002, + "loss": 1.1187, "step": 1392 }, { "epoch": 1.52, - "learning_rate": 0.0005, - "loss": 1.0449, + "learning_rate": 0.0002, + "loss": 1.0613, "step": 1393 }, { "epoch": 1.52, - "learning_rate": 0.0005, - "loss": 1.1213, + "learning_rate": 0.0002, + "loss": 1.1442, "step": 1394 }, { "epoch": 1.52, - "learning_rate": 0.0005, - "loss": 1.031, + "learning_rate": 0.0002, + "loss": 1.0562, "step": 1395 }, { "epoch": 1.52, - "learning_rate": 0.0005, - "loss": 1.0555, + "learning_rate": 0.0002, + "loss": 1.0712, "step": 1396 }, { "epoch": 1.52, - "learning_rate": 0.0005, - "loss": 1.0238, + "learning_rate": 0.0002, + "loss": 1.0562, "step": 1397 }, { "epoch": 1.52, - "learning_rate": 0.0005, - "loss": 1.0594, + "learning_rate": 0.0002, + "loss": 1.0844, "step": 1398 }, { "epoch": 1.52, - "learning_rate": 0.0005, - "loss": 1.008, + "learning_rate": 0.0002, + "loss": 1.0275, "step": 1399 }, { "epoch": 1.52, - "learning_rate": 0.0005, - "loss": 1.0221, + "learning_rate": 0.0002, + "loss": 1.0413, "step": 1400 }, { "epoch": 1.52, - "learning_rate": 0.0005, - "loss": 1.0769, + "learning_rate": 0.0002, + "loss": 1.0961, "step": 1401 }, { "epoch": 1.53, - "learning_rate": 0.0005, - "loss": 1.0465, + "learning_rate": 0.0002, + "loss": 1.0783, "step": 1402 }, { "epoch": 1.53, - "learning_rate": 0.0005, - "loss": 1.0238, + "learning_rate": 0.0002, + "loss": 1.0428, "step": 1403 }, { "epoch": 1.53, - "learning_rate": 0.0005, - "loss": 1.0269, + "learning_rate": 0.0002, + "loss": 1.0473, "step": 1404 }, { "epoch": 1.53, - "learning_rate": 0.0005, - "loss": 1.05, + "learning_rate": 0.0002, + "loss": 1.0741, "step": 1405 }, { "epoch": 1.53, - "learning_rate": 0.0005, - "loss": 1.0621, + "learning_rate": 0.0002, + "loss": 1.0804, "step": 1406 }, { "epoch": 1.53, - "learning_rate": 0.0005, - "loss": 1.041, + "learning_rate": 0.0002, + "loss": 1.0615, "step": 1407 }, { "epoch": 1.53, - "learning_rate": 0.0005, - "loss": 1.0424, + "learning_rate": 0.0002, + "loss": 1.0589, "step": 1408 }, { "epoch": 1.53, - "learning_rate": 0.0005, - "loss": 1.0524, + "learning_rate": 0.0002, + "loss": 1.0758, "step": 1409 }, { "epoch": 1.53, - "learning_rate": 0.0005, - "loss": 1.0777, + "learning_rate": 0.0002, + "loss": 1.0939, "step": 1410 }, { "epoch": 1.54, - "learning_rate": 0.0005, - "loss": 1.0459, + "learning_rate": 0.0002, + "loss": 1.0716, "step": 1411 }, { "epoch": 1.54, - "learning_rate": 0.0005, - "loss": 1.0855, + "learning_rate": 0.0002, + "loss": 1.1127, "step": 1412 }, { "epoch": 1.54, - "learning_rate": 0.0005, - "loss": 1.0464, + "learning_rate": 0.0002, + "loss": 1.0724, "step": 1413 }, { "epoch": 1.54, - "learning_rate": 0.0005, - "loss": 1.0096, + "learning_rate": 0.0002, + "loss": 1.0233, "step": 1414 }, { "epoch": 1.54, - "learning_rate": 0.0005, - "loss": 1.0734, + "learning_rate": 0.0002, + "loss": 1.0939, "step": 1415 }, { "epoch": 1.54, - "learning_rate": 0.0005, - "loss": 1.0408, + "learning_rate": 0.0002, + "loss": 1.0651, "step": 1416 }, { "epoch": 1.54, - "learning_rate": 0.0005, - "loss": 1.0555, + "learning_rate": 0.0002, + "loss": 1.0797, "step": 1417 }, { "epoch": 1.54, - "learning_rate": 0.0005, - "loss": 1.0799, + "learning_rate": 0.0002, + "loss": 1.0984, "step": 1418 }, { "epoch": 1.54, - "learning_rate": 0.0005, - "loss": 1.0996, + "learning_rate": 0.0002, + "loss": 1.1252, "step": 1419 }, { "epoch": 1.55, - "learning_rate": 0.0005, - "loss": 1.0901, + "learning_rate": 0.0002, + "loss": 1.1016, "step": 1420 }, { "epoch": 1.55, - "learning_rate": 0.0005, - "loss": 1.0697, + "learning_rate": 0.0002, + "loss": 1.0858, "step": 1421 }, { "epoch": 1.55, - "learning_rate": 0.0005, - "loss": 1.023, + "learning_rate": 0.0002, + "loss": 1.0416, "step": 1422 }, { "epoch": 1.55, - "learning_rate": 0.0005, - "loss": 1.0804, + "learning_rate": 0.0002, + "loss": 1.1086, "step": 1423 }, { "epoch": 1.55, - "learning_rate": 0.0005, - "loss": 1.0364, + "learning_rate": 0.0002, + "loss": 1.0551, "step": 1424 }, { "epoch": 1.55, - "learning_rate": 0.0005, - "loss": 1.0329, + "learning_rate": 0.0002, + "loss": 1.0499, "step": 1425 }, { "epoch": 1.55, - "learning_rate": 0.0005, - "loss": 1.0691, + "learning_rate": 0.0002, + "loss": 1.0857, "step": 1426 }, { "epoch": 1.55, - "learning_rate": 0.0005, - "loss": 1.0515, + "learning_rate": 0.0002, + "loss": 1.0647, "step": 1427 }, { "epoch": 1.55, - "learning_rate": 0.0005, - "loss": 1.0411, + "learning_rate": 0.0002, + "loss": 1.0612, "step": 1428 }, { "epoch": 1.55, - "learning_rate": 0.0005, - "loss": 1.0218, + "learning_rate": 0.0002, + "loss": 1.0371, "step": 1429 }, { "epoch": 1.56, - "learning_rate": 0.0005, - "loss": 1.0413, + "learning_rate": 0.0002, + "loss": 1.0583, "step": 1430 }, { "epoch": 1.56, - "learning_rate": 0.0005, - "loss": 1.0559, + "learning_rate": 0.0002, + "loss": 1.0777, "step": 1431 }, { "epoch": 1.56, - "learning_rate": 0.0005, - "loss": 1.0644, + "learning_rate": 0.0002, + "loss": 1.0853, "step": 1432 }, { "epoch": 1.56, - "learning_rate": 0.0005, - "loss": 1.0772, + "learning_rate": 0.0002, + "loss": 1.0942, "step": 1433 }, { "epoch": 1.56, - "learning_rate": 0.0005, - "loss": 1.0682, + "learning_rate": 0.0002, + "loss": 1.0959, "step": 1434 }, { "epoch": 1.56, - "learning_rate": 0.0005, - "loss": 1.1045, + "learning_rate": 0.0002, + "loss": 1.1249, "step": 1435 }, { "epoch": 1.56, - "learning_rate": 0.0005, - "loss": 1.0519, + "learning_rate": 0.0002, + "loss": 1.0788, "step": 1436 }, { "epoch": 1.56, - "learning_rate": 0.0005, - "loss": 1.0432, + "learning_rate": 0.0002, + "loss": 1.0595, "step": 1437 }, { "epoch": 1.56, - "learning_rate": 0.0005, - "loss": 1.0755, + "learning_rate": 0.0002, + "loss": 1.1027, "step": 1438 }, { "epoch": 1.57, - "learning_rate": 0.0005, - "loss": 1.0313, + "learning_rate": 0.0002, + "loss": 1.0475, "step": 1439 }, { "epoch": 1.57, - "learning_rate": 0.0005, - "loss": 1.0871, + "learning_rate": 0.0002, + "loss": 1.1073, "step": 1440 }, { "epoch": 1.57, - "learning_rate": 0.0005, - "loss": 1.0084, + "learning_rate": 0.0002, + "loss": 1.0272, "step": 1441 }, { "epoch": 1.57, - "learning_rate": 0.0005, - "loss": 1.0676, + "learning_rate": 0.0002, + "loss": 1.0761, "step": 1442 }, { "epoch": 1.57, - "learning_rate": 0.0005, - "loss": 1.0713, + "learning_rate": 0.0002, + "loss": 1.0892, "step": 1443 }, { "epoch": 1.57, - "learning_rate": 0.0005, - "loss": 1.0216, + "learning_rate": 0.0002, + "loss": 1.0405, "step": 1444 }, { "epoch": 1.57, - "learning_rate": 0.0005, - "loss": 0.9717, + "learning_rate": 0.0002, + "loss": 0.9884, "step": 1445 }, { "epoch": 1.57, - "learning_rate": 0.0005, - "loss": 1.0681, + "learning_rate": 0.0002, + "loss": 1.0954, "step": 1446 }, { "epoch": 1.57, - "learning_rate": 0.0005, - "loss": 1.0643, + "learning_rate": 0.0002, + "loss": 1.0938, "step": 1447 }, { "epoch": 1.58, - "learning_rate": 0.0005, - "loss": 1.0511, + "learning_rate": 0.0002, + "loss": 1.0741, "step": 1448 }, { "epoch": 1.58, - "learning_rate": 0.0005, - "loss": 1.0455, + "learning_rate": 0.0002, + "loss": 1.0703, "step": 1449 }, { "epoch": 1.58, - "learning_rate": 0.0005, - "loss": 1.0534, + "learning_rate": 0.0002, + "loss": 1.075, "step": 1450 }, { "epoch": 1.58, - "learning_rate": 0.0005, - "loss": 1.0444, + "learning_rate": 0.0002, + "loss": 1.0654, "step": 1451 }, { "epoch": 1.58, - "learning_rate": 0.0005, - "loss": 1.0665, + "learning_rate": 0.0002, + "loss": 1.0869, "step": 1452 }, { "epoch": 1.58, - "learning_rate": 0.0005, - "loss": 1.0509, + "learning_rate": 0.0002, + "loss": 1.0741, "step": 1453 }, { "epoch": 1.58, - "learning_rate": 0.0005, - "loss": 1.0475, + "learning_rate": 0.0002, + "loss": 1.0646, "step": 1454 }, { "epoch": 1.58, - "learning_rate": 0.0005, - "loss": 1.0491, + "learning_rate": 0.0002, + "loss": 1.0674, "step": 1455 }, { "epoch": 1.58, - "learning_rate": 0.0005, - "loss": 1.0397, + "learning_rate": 0.0002, + "loss": 1.057, "step": 1456 }, { "epoch": 1.59, - "learning_rate": 0.0005, - "loss": 1.0087, + "learning_rate": 0.0002, + "loss": 1.0379, "step": 1457 }, { "epoch": 1.59, - "learning_rate": 0.0005, - "loss": 1.0409, + "learning_rate": 0.0002, + "loss": 1.0757, "step": 1458 }, { "epoch": 1.59, - "learning_rate": 0.0005, - "loss": 1.0441, + "learning_rate": 0.0002, + "loss": 1.0625, "step": 1459 }, { "epoch": 1.59, - "learning_rate": 0.0005, - "loss": 1.0472, + "learning_rate": 0.0002, + "loss": 1.0663, "step": 1460 }, { "epoch": 1.59, - "learning_rate": 0.0005, - "loss": 1.0883, + "learning_rate": 0.0002, + "loss": 1.1121, "step": 1461 }, { "epoch": 1.59, - "learning_rate": 0.0005, - "loss": 1.0527, + "learning_rate": 0.0002, + "loss": 1.0783, "step": 1462 }, { "epoch": 1.59, - "learning_rate": 0.0005, - "loss": 1.0485, + "learning_rate": 0.0002, + "loss": 1.0723, "step": 1463 }, { "epoch": 1.59, - "learning_rate": 0.0005, - "loss": 1.0616, + "learning_rate": 0.0002, + "loss": 1.077, "step": 1464 }, { "epoch": 1.59, - "learning_rate": 0.0005, - "loss": 1.0715, + "learning_rate": 0.0002, + "loss": 1.1015, "step": 1465 }, { "epoch": 1.6, - "learning_rate": 0.0005, - "loss": 1.0728, + "learning_rate": 0.0002, + "loss": 1.0944, "step": 1466 }, { "epoch": 1.6, - "learning_rate": 0.0005, - "loss": 1.0453, + "learning_rate": 0.0002, + "loss": 1.0653, "step": 1467 }, { "epoch": 1.6, - "learning_rate": 0.0005, - "loss": 1.0599, + "learning_rate": 0.0002, + "loss": 1.0815, "step": 1468 }, { "epoch": 1.6, - "learning_rate": 0.0005, - "loss": 1.0541, + "learning_rate": 0.0002, + "loss": 1.0705, "step": 1469 }, { "epoch": 1.6, - "learning_rate": 0.0005, - "loss": 1.0333, + "learning_rate": 0.0002, + "loss": 1.0548, "step": 1470 }, { "epoch": 1.6, - "learning_rate": 0.0005, - "loss": 1.0876, + "learning_rate": 0.0002, + "loss": 1.1045, "step": 1471 }, { "epoch": 1.6, - "learning_rate": 0.0005, - "loss": 1.0434, + "learning_rate": 0.0002, + "loss": 1.0639, "step": 1472 }, { "epoch": 1.6, - "learning_rate": 0.0005, - "loss": 1.111, + "learning_rate": 0.0002, + "loss": 1.1274, "step": 1473 }, { "epoch": 1.6, - "learning_rate": 0.0005, - "loss": 1.0601, + "learning_rate": 0.0002, + "loss": 1.0825, "step": 1474 }, { "epoch": 1.61, - "learning_rate": 0.0005, - "loss": 1.041, + "learning_rate": 0.0002, + "loss": 1.0616, "step": 1475 }, { "epoch": 1.61, - "learning_rate": 0.0005, - "loss": 1.1053, + "learning_rate": 0.0002, + "loss": 1.1221, "step": 1476 }, { "epoch": 1.61, - "learning_rate": 0.0005, - "loss": 1.049, + "learning_rate": 0.0002, + "loss": 1.0787, "step": 1477 }, { "epoch": 1.61, - "learning_rate": 0.0005, - "loss": 1.0727, + "learning_rate": 0.0002, + "loss": 1.0934, "step": 1478 }, { "epoch": 1.61, - "learning_rate": 0.0005, - "loss": 1.0751, + "learning_rate": 0.0002, + "loss": 1.1104, "step": 1479 }, { "epoch": 1.61, - "learning_rate": 0.0005, - "loss": 1.0059, + "learning_rate": 0.0002, + "loss": 1.0261, "step": 1480 }, { "epoch": 1.61, - "learning_rate": 0.0005, - "loss": 1.068, + "learning_rate": 0.0002, + "loss": 1.0798, "step": 1481 }, { "epoch": 1.61, - "learning_rate": 0.0005, - "loss": 1.0948, + "learning_rate": 0.0002, + "loss": 1.107, "step": 1482 }, { "epoch": 1.61, - "learning_rate": 0.0005, - "loss": 1.0801, + "learning_rate": 0.0002, + "loss": 1.0961, "step": 1483 }, { "epoch": 1.61, - "learning_rate": 0.0005, - "loss": 1.034, + "learning_rate": 0.0002, + "loss": 1.0589, "step": 1484 }, { "epoch": 1.62, - "learning_rate": 0.0005, - "loss": 1.0364, + "learning_rate": 0.0002, + "loss": 1.051, "step": 1485 }, { "epoch": 1.62, - "learning_rate": 0.0005, - "loss": 1.0269, + "learning_rate": 0.0002, + "loss": 1.047, "step": 1486 }, { "epoch": 1.62, - "learning_rate": 0.0005, - "loss": 1.0391, + "learning_rate": 0.0002, + "loss": 1.0658, "step": 1487 }, { "epoch": 1.62, - "learning_rate": 0.0005, - "loss": 1.062, + "learning_rate": 0.0002, + "loss": 1.0817, "step": 1488 }, { "epoch": 1.62, - "learning_rate": 0.0005, - "loss": 1.0436, + "learning_rate": 0.0002, + "loss": 1.0648, "step": 1489 }, { "epoch": 1.62, - "learning_rate": 0.0005, - "loss": 1.0428, + "learning_rate": 0.0002, + "loss": 1.0676, "step": 1490 }, { "epoch": 1.62, - "learning_rate": 0.0005, - "loss": 1.0371, + "learning_rate": 0.0002, + "loss": 1.0571, "step": 1491 }, { "epoch": 1.62, - "learning_rate": 0.0005, - "loss": 1.0382, + "learning_rate": 0.0002, + "loss": 1.0579, "step": 1492 }, { "epoch": 1.62, - "learning_rate": 0.0005, - "loss": 1.0679, + "learning_rate": 0.0002, + "loss": 1.0978, "step": 1493 }, { "epoch": 1.63, - "learning_rate": 0.0005, - "loss": 1.0381, + "learning_rate": 0.0002, + "loss": 1.0658, "step": 1494 }, { "epoch": 1.63, - "learning_rate": 0.0005, - "loss": 1.0361, + "learning_rate": 0.0002, + "loss": 1.0736, "step": 1495 }, { "epoch": 1.63, - "learning_rate": 0.0005, - "loss": 1.0637, + "learning_rate": 0.0002, + "loss": 1.0855, "step": 1496 }, { "epoch": 1.63, - "learning_rate": 0.0005, - "loss": 1.0545, + "learning_rate": 0.0002, + "loss": 1.0744, "step": 1497 }, { "epoch": 1.63, - "learning_rate": 0.0005, - "loss": 1.0506, + "learning_rate": 0.0002, + "loss": 1.0736, "step": 1498 }, { "epoch": 1.63, - "learning_rate": 0.0005, - "loss": 1.1028, + "learning_rate": 0.0002, + "loss": 1.1298, "step": 1499 }, { "epoch": 1.63, - "learning_rate": 0.0005, - "loss": 1.0567, + "learning_rate": 0.0002, + "loss": 1.074, "step": 1500 }, { "epoch": 1.63, - "learning_rate": 0.0005, - "loss": 1.0552, + "learning_rate": 0.0002, + "loss": 1.0771, "step": 1501 }, { "epoch": 1.63, - "learning_rate": 0.0005, - "loss": 1.0634, + "learning_rate": 0.0002, + "loss": 1.0861, "step": 1502 }, { "epoch": 1.64, - "learning_rate": 0.0005, - "loss": 1.0471, + "learning_rate": 0.0002, + "loss": 1.0754, "step": 1503 }, { "epoch": 1.64, - "learning_rate": 0.0005, - "loss": 1.0658, + "learning_rate": 0.0002, + "loss": 1.0948, "step": 1504 }, { "epoch": 1.64, - "learning_rate": 0.0005, - "loss": 1.0451, + "learning_rate": 0.0002, + "loss": 1.0653, "step": 1505 }, { "epoch": 1.64, - "learning_rate": 0.0005, - "loss": 1.05, + "learning_rate": 0.0002, + "loss": 1.0844, "step": 1506 }, { "epoch": 1.64, - "learning_rate": 0.0005, - "loss": 1.0216, + "learning_rate": 0.0002, + "loss": 1.0469, "step": 1507 }, { "epoch": 1.64, - "learning_rate": 0.0005, - "loss": 1.0876, + "learning_rate": 0.0002, + "loss": 1.1155, "step": 1508 }, { "epoch": 1.64, - "learning_rate": 0.0005, - "loss": 1.03, + "learning_rate": 0.0002, + "loss": 1.0494, "step": 1509 }, { "epoch": 1.64, - "learning_rate": 0.0005, - "loss": 1.0522, + "learning_rate": 0.0002, + "loss": 1.0676, "step": 1510 }, { "epoch": 1.64, - "learning_rate": 0.0005, - "loss": 1.0873, + "learning_rate": 0.0002, + "loss": 1.1099, "step": 1511 }, { "epoch": 1.65, - "learning_rate": 0.0005, - "loss": 1.1165, + "learning_rate": 0.0002, + "loss": 1.1311, "step": 1512 }, { "epoch": 1.65, - "learning_rate": 0.0005, - "loss": 1.0514, + "learning_rate": 0.0002, + "loss": 1.0881, "step": 1513 }, { "epoch": 1.65, - "learning_rate": 0.0005, - "loss": 1.0429, + "learning_rate": 0.0002, + "loss": 1.063, "step": 1514 }, { "epoch": 1.65, - "learning_rate": 0.0005, - "loss": 1.0677, + "learning_rate": 0.0002, + "loss": 1.0868, "step": 1515 }, { "epoch": 1.65, - "learning_rate": 0.0005, - "loss": 1.0844, + "learning_rate": 0.0002, + "loss": 1.1051, "step": 1516 }, { "epoch": 1.65, - "learning_rate": 0.0005, - "loss": 1.0521, + "learning_rate": 0.0002, + "loss": 1.073, "step": 1517 }, { "epoch": 1.65, - "learning_rate": 0.0005, - "loss": 1.0227, + "learning_rate": 0.0002, + "loss": 1.0494, "step": 1518 }, { "epoch": 1.65, - "learning_rate": 0.0005, - "loss": 1.1024, + "learning_rate": 0.0002, + "loss": 1.1218, "step": 1519 }, { "epoch": 1.65, - "learning_rate": 0.0005, - "loss": 1.0323, + "learning_rate": 0.0002, + "loss": 1.0463, "step": 1520 }, { "epoch": 1.66, - "learning_rate": 0.0005, - "loss": 1.066, + "learning_rate": 0.0002, + "loss": 1.0869, "step": 1521 }, { "epoch": 1.66, - "learning_rate": 0.0005, - "loss": 1.0564, + "learning_rate": 0.0002, + "loss": 1.0788, "step": 1522 }, { "epoch": 1.66, - "learning_rate": 0.0005, - "loss": 1.0404, + "learning_rate": 0.0002, + "loss": 1.0626, "step": 1523 }, { "epoch": 1.66, - "learning_rate": 0.0005, - "loss": 1.0389, + "learning_rate": 0.0002, + "loss": 1.0659, "step": 1524 }, { "epoch": 1.66, - "learning_rate": 0.0005, - "loss": 1.0369, + "learning_rate": 0.0002, + "loss": 1.062, "step": 1525 }, { "epoch": 1.66, - "learning_rate": 0.0005, - "loss": 1.0604, + "learning_rate": 0.0002, + "loss": 1.0722, "step": 1526 }, { "epoch": 1.66, - "learning_rate": 0.0005, - "loss": 1.0855, + "learning_rate": 0.0002, + "loss": 1.1057, "step": 1527 }, { "epoch": 1.66, - "learning_rate": 0.0005, - "loss": 1.0372, + "learning_rate": 0.0002, + "loss": 1.0573, "step": 1528 }, { "epoch": 1.66, - "learning_rate": 0.0005, - "loss": 1.0636, + "learning_rate": 0.0002, + "loss": 1.0827, "step": 1529 }, { "epoch": 1.66, - "learning_rate": 0.0005, - "loss": 1.013, + "learning_rate": 0.0002, + "loss": 1.0381, "step": 1530 }, { "epoch": 1.67, - "learning_rate": 0.0005, - "loss": 1.0526, + "learning_rate": 0.0002, + "loss": 1.0799, "step": 1531 }, { "epoch": 1.67, - "learning_rate": 0.0005, - "loss": 1.0638, + "learning_rate": 0.0002, + "loss": 1.0826, "step": 1532 }, { "epoch": 1.67, - "learning_rate": 0.0005, - "loss": 1.0826, + "learning_rate": 0.0002, + "loss": 1.0993, "step": 1533 }, { "epoch": 1.67, - "learning_rate": 0.0005, - "loss": 1.0094, + "learning_rate": 0.0002, + "loss": 1.0298, "step": 1534 }, { "epoch": 1.67, - "learning_rate": 0.0005, - "loss": 1.0694, + "learning_rate": 0.0002, + "loss": 1.0835, "step": 1535 }, { "epoch": 1.67, - "learning_rate": 0.0005, - "loss": 1.0065, + "learning_rate": 0.0002, + "loss": 1.0274, "step": 1536 }, { "epoch": 1.67, - "learning_rate": 0.0005, - "loss": 1.054, + "learning_rate": 0.0002, + "loss": 1.0776, "step": 1537 }, { "epoch": 1.67, - "learning_rate": 0.0005, - "loss": 1.0815, + "learning_rate": 0.0002, + "loss": 1.1, "step": 1538 }, { "epoch": 1.67, - "learning_rate": 0.0005, - "loss": 0.9896, + "learning_rate": 0.0002, + "loss": 1.0104, "step": 1539 }, { "epoch": 1.68, - "learning_rate": 0.0005, - "loss": 1.0985, + "learning_rate": 0.0002, + "loss": 1.1204, "step": 1540 }, { "epoch": 1.68, - "learning_rate": 0.0005, - "loss": 1.0575, + "learning_rate": 0.0002, + "loss": 1.0818, "step": 1541 }, { "epoch": 1.68, - "learning_rate": 0.0005, - "loss": 1.0955, + "learning_rate": 0.0002, + "loss": 1.125, "step": 1542 }, { "epoch": 1.68, - "learning_rate": 0.0005, - "loss": 1.0141, + "learning_rate": 0.0002, + "loss": 1.0321, "step": 1543 }, { "epoch": 1.68, - "learning_rate": 0.0005, - "loss": 1.0469, + "learning_rate": 0.0002, + "loss": 1.0722, "step": 1544 }, { "epoch": 1.68, - "learning_rate": 0.0005, - "loss": 1.0503, + "learning_rate": 0.0002, + "loss": 1.0745, "step": 1545 }, { "epoch": 1.68, - "learning_rate": 0.0005, - "loss": 1.1026, + "learning_rate": 0.0002, + "loss": 1.1312, "step": 1546 }, { "epoch": 1.68, - "learning_rate": 0.0005, - "loss": 1.051, + "learning_rate": 0.0002, + "loss": 1.0644, "step": 1547 }, { "epoch": 1.68, - "learning_rate": 0.0005, - "loss": 1.0255, + "learning_rate": 0.0002, + "loss": 1.0502, "step": 1548 }, { "epoch": 1.69, - "learning_rate": 0.0005, - "loss": 1.0825, + "learning_rate": 0.0002, + "loss": 1.1128, "step": 1549 }, { "epoch": 1.69, - "learning_rate": 0.0005, - "loss": 1.0121, + "learning_rate": 0.0002, + "loss": 1.029, "step": 1550 }, { "epoch": 1.69, - "learning_rate": 0.0005, - "loss": 1.0154, + "learning_rate": 0.0002, + "loss": 1.0396, "step": 1551 }, { "epoch": 1.69, - "learning_rate": 0.0005, - "loss": 1.0271, + "learning_rate": 0.0002, + "loss": 1.0529, "step": 1552 }, { "epoch": 1.69, - "learning_rate": 0.0005, - "loss": 1.0879, + "learning_rate": 0.0002, + "loss": 1.1074, "step": 1553 }, { "epoch": 1.69, - "learning_rate": 0.0005, - "loss": 1.0949, + "learning_rate": 0.0002, + "loss": 1.122, "step": 1554 }, { "epoch": 1.69, - "learning_rate": 0.0005, - "loss": 1.0793, + "learning_rate": 0.0002, + "loss": 1.1002, "step": 1555 }, { "epoch": 1.69, - "learning_rate": 0.0005, - "loss": 1.0399, + "learning_rate": 0.0002, + "loss": 1.0541, "step": 1556 }, { "epoch": 1.69, - "learning_rate": 0.0005, - "loss": 1.0222, + "learning_rate": 0.0002, + "loss": 1.047, "step": 1557 }, { "epoch": 1.7, - "learning_rate": 0.0005, - "loss": 1.0266, + "learning_rate": 0.0002, + "loss": 1.0485, "step": 1558 }, { "epoch": 1.7, - "learning_rate": 0.0005, - "loss": 1.019, + "learning_rate": 0.0002, + "loss": 1.0388, "step": 1559 }, { "epoch": 1.7, - "learning_rate": 0.0005, - "loss": 1.059, + "learning_rate": 0.0002, + "loss": 1.0874, "step": 1560 }, { "epoch": 1.7, - "learning_rate": 0.0005, - "loss": 1.0532, + "learning_rate": 0.0002, + "loss": 1.0799, "step": 1561 }, { "epoch": 1.7, - "learning_rate": 0.0005, - "loss": 1.0871, + "learning_rate": 0.0002, + "loss": 1.1196, "step": 1562 }, { "epoch": 1.7, - "learning_rate": 0.0005, - "loss": 1.0687, + "learning_rate": 0.0002, + "loss": 1.0896, "step": 1563 }, { "epoch": 1.7, - "learning_rate": 0.0005, - "loss": 1.0467, + "learning_rate": 0.0002, + "loss": 1.0646, "step": 1564 }, { "epoch": 1.7, - "learning_rate": 0.0005, - "loss": 1.0666, + "learning_rate": 0.0002, + "loss": 1.0866, "step": 1565 }, { "epoch": 1.7, - "learning_rate": 0.0005, - "loss": 1.0591, + "learning_rate": 0.0002, + "loss": 1.0834, "step": 1566 }, { "epoch": 1.71, - "learning_rate": 0.0005, - "loss": 1.0371, + "learning_rate": 0.0002, + "loss": 1.0551, "step": 1567 }, { "epoch": 1.71, - "learning_rate": 0.0005, - "loss": 1.0487, + "learning_rate": 0.0002, + "loss": 1.0613, "step": 1568 }, { "epoch": 1.71, - "learning_rate": 0.0005, - "loss": 1.0789, + "learning_rate": 0.0002, + "loss": 1.1021, "step": 1569 }, { "epoch": 1.71, - "learning_rate": 0.0005, - "loss": 1.1071, + "learning_rate": 0.0002, + "loss": 1.1186, "step": 1570 }, { "epoch": 1.71, - "learning_rate": 0.0005, - "loss": 1.0851, + "learning_rate": 0.0002, + "loss": 1.1015, "step": 1571 }, { "epoch": 1.71, - "learning_rate": 0.0005, - "loss": 1.0322, + "learning_rate": 0.0002, + "loss": 1.0444, "step": 1572 }, { "epoch": 1.71, - "learning_rate": 0.0005, - "loss": 1.0492, + "learning_rate": 0.0002, + "loss": 1.0747, "step": 1573 }, { "epoch": 1.71, - "learning_rate": 0.0005, - "loss": 1.0946, + "learning_rate": 0.0002, + "loss": 1.1101, "step": 1574 }, { "epoch": 1.71, - "learning_rate": 0.0005, - "loss": 1.0483, + "learning_rate": 0.0002, + "loss": 1.0759, "step": 1575 }, { "epoch": 1.71, - "learning_rate": 0.0005, - "loss": 1.0763, + "learning_rate": 0.0002, + "loss": 1.107, "step": 1576 }, { "epoch": 1.72, - "learning_rate": 0.0005, - "loss": 1.0598, + "learning_rate": 0.0002, + "loss": 1.0908, "step": 1577 }, { "epoch": 1.72, - "learning_rate": 0.0005, - "loss": 1.037, + "learning_rate": 0.0002, + "loss": 1.0534, "step": 1578 }, { "epoch": 1.72, - "learning_rate": 0.0005, - "loss": 1.0783, + "learning_rate": 0.0002, + "loss": 1.0991, "step": 1579 }, { "epoch": 1.72, - "learning_rate": 0.0005, - "loss": 1.0141, + "learning_rate": 0.0002, + "loss": 1.0355, "step": 1580 }, { "epoch": 1.72, - "learning_rate": 0.0005, - "loss": 1.0454, + "learning_rate": 0.0002, + "loss": 1.0799, "step": 1581 }, { "epoch": 1.72, - "learning_rate": 0.0005, - "loss": 1.08, + "learning_rate": 0.0002, + "loss": 1.1018, "step": 1582 }, { "epoch": 1.72, - "learning_rate": 0.0005, - "loss": 1.045, + "learning_rate": 0.0002, + "loss": 1.0639, "step": 1583 }, { "epoch": 1.72, - "learning_rate": 0.0005, - "loss": 1.0488, + "learning_rate": 0.0002, + "loss": 1.0608, "step": 1584 }, { "epoch": 1.72, - "learning_rate": 0.0005, - "loss": 1.0689, + "learning_rate": 0.0002, + "loss": 1.0994, "step": 1585 }, { "epoch": 1.73, - "learning_rate": 0.0005, - "loss": 1.0618, + "learning_rate": 0.0002, + "loss": 1.0808, "step": 1586 }, { "epoch": 1.73, - "learning_rate": 0.0005, - "loss": 1.0797, + "learning_rate": 0.0002, + "loss": 1.1061, "step": 1587 }, { "epoch": 1.73, - "learning_rate": 0.0005, - "loss": 1.0686, + "learning_rate": 0.0002, + "loss": 1.0902, "step": 1588 }, { "epoch": 1.73, - "learning_rate": 0.0005, - "loss": 1.0606, + "learning_rate": 0.0002, + "loss": 1.0744, "step": 1589 }, { "epoch": 1.73, - "learning_rate": 0.0005, - "loss": 1.0873, + "learning_rate": 0.0002, + "loss": 1.1096, "step": 1590 }, { "epoch": 1.73, - "learning_rate": 0.0005, - "loss": 1.0286, + "learning_rate": 0.0002, + "loss": 1.0486, "step": 1591 }, { "epoch": 1.73, - "learning_rate": 0.0005, - "loss": 1.0942, + "learning_rate": 0.0002, + "loss": 1.1084, "step": 1592 }, { "epoch": 1.73, - "learning_rate": 0.0005, - "loss": 1.031, + "learning_rate": 0.0002, + "loss": 1.0461, "step": 1593 }, { "epoch": 1.73, - "learning_rate": 0.0005, - "loss": 1.0085, + "learning_rate": 0.0002, + "loss": 1.027, "step": 1594 }, { "epoch": 1.74, - "learning_rate": 0.0005, - "loss": 1.0701, + "learning_rate": 0.0002, + "loss": 1.093, "step": 1595 }, { "epoch": 1.74, - "learning_rate": 0.0005, - "loss": 1.0224, + "learning_rate": 0.0002, + "loss": 1.0407, "step": 1596 }, { "epoch": 1.74, - "learning_rate": 0.0005, - "loss": 1.0493, + "learning_rate": 0.0002, + "loss": 1.0786, "step": 1597 }, { "epoch": 1.74, - "learning_rate": 0.0005, - "loss": 1.0616, + "learning_rate": 0.0002, + "loss": 1.0902, "step": 1598 }, { "epoch": 1.74, - "learning_rate": 0.0005, - "loss": 1.0605, + "learning_rate": 0.0002, + "loss": 1.0807, "step": 1599 }, { "epoch": 1.74, - "learning_rate": 0.0005, - "loss": 1.0817, + "learning_rate": 0.0002, + "loss": 1.1003, "step": 1600 }, { "epoch": 1.74, - "learning_rate": 0.0005, - "loss": 1.0588, + "learning_rate": 0.0002, + "loss": 1.0689, "step": 1601 }, { "epoch": 1.74, - "learning_rate": 0.0005, - "loss": 1.0897, + "learning_rate": 0.0002, + "loss": 1.1069, "step": 1602 }, { "epoch": 1.74, - "learning_rate": 0.0005, - "loss": 1.0765, + "learning_rate": 0.0002, + "loss": 1.0947, "step": 1603 }, { "epoch": 1.75, - "learning_rate": 0.0005, - "loss": 1.0534, + "learning_rate": 0.0002, + "loss": 1.078, "step": 1604 }, { "epoch": 1.75, - "learning_rate": 0.0005, - "loss": 1.081, + "learning_rate": 0.0002, + "loss": 1.1039, "step": 1605 }, { "epoch": 1.75, - "learning_rate": 0.0005, - "loss": 1.0509, + "learning_rate": 0.0002, + "loss": 1.0692, "step": 1606 }, { "epoch": 1.75, - "learning_rate": 0.0005, - "loss": 1.043, + "learning_rate": 0.0002, + "loss": 1.0702, "step": 1607 }, { "epoch": 1.75, - "learning_rate": 0.0005, - "loss": 1.0886, + "learning_rate": 0.0002, + "loss": 1.121, "step": 1608 }, { "epoch": 1.75, - "learning_rate": 0.0005, - "loss": 1.0095, + "learning_rate": 0.0002, + "loss": 1.033, "step": 1609 }, { "epoch": 1.75, - "learning_rate": 0.0005, - "loss": 1.0585, + "learning_rate": 0.0002, + "loss": 1.0817, "step": 1610 }, { "epoch": 1.75, - "learning_rate": 0.0005, - "loss": 0.9999, + "learning_rate": 0.0002, + "loss": 1.0254, "step": 1611 }, { "epoch": 1.75, - "learning_rate": 0.0005, - "loss": 1.0506, + "learning_rate": 0.0002, + "loss": 1.0745, "step": 1612 }, { "epoch": 1.76, - "learning_rate": 0.0005, - "loss": 1.0372, + "learning_rate": 0.0002, + "loss": 1.0544, "step": 1613 }, { "epoch": 1.76, - "learning_rate": 0.0005, - "loss": 1.0096, + "learning_rate": 0.0002, + "loss": 1.0297, "step": 1614 }, { "epoch": 1.76, - "learning_rate": 0.0005, - "loss": 1.0323, + "learning_rate": 0.0002, + "loss": 1.0468, "step": 1615 }, { "epoch": 1.76, - "learning_rate": 0.0005, - "loss": 1.0694, + "learning_rate": 0.0002, + "loss": 1.0927, "step": 1616 }, { "epoch": 1.76, - "learning_rate": 0.0005, - "loss": 1.0695, + "learning_rate": 0.0002, + "loss": 1.0909, "step": 1617 }, { "epoch": 1.76, - "learning_rate": 0.0005, - "loss": 1.0331, + "learning_rate": 0.0002, + "loss": 1.0542, "step": 1618 }, { "epoch": 1.76, - "learning_rate": 0.0005, - "loss": 1.0589, + "learning_rate": 0.0002, + "loss": 1.0849, "step": 1619 }, { "epoch": 1.76, - "learning_rate": 0.0005, - "loss": 1.0534, + "learning_rate": 0.0002, + "loss": 1.0773, "step": 1620 }, { "epoch": 1.76, - "learning_rate": 0.0005, - "loss": 1.0897, + "learning_rate": 0.0002, + "loss": 1.1168, "step": 1621 }, { "epoch": 1.76, - "learning_rate": 0.0005, - "loss": 1.0548, + "learning_rate": 0.0002, + "loss": 1.0771, "step": 1622 }, { "epoch": 1.77, - "learning_rate": 0.0005, - "loss": 1.0638, + "learning_rate": 0.0002, + "loss": 1.0857, "step": 1623 }, { "epoch": 1.77, - "learning_rate": 0.0005, - "loss": 1.0603, + "learning_rate": 0.0002, + "loss": 1.081, "step": 1624 }, { "epoch": 1.77, - "learning_rate": 0.0005, - "loss": 1.0661, + "learning_rate": 0.0002, + "loss": 1.0877, "step": 1625 }, { "epoch": 1.77, - "learning_rate": 0.0005, - "loss": 1.0148, + "learning_rate": 0.0002, + "loss": 1.0391, "step": 1626 }, { "epoch": 1.77, - "learning_rate": 0.0005, - "loss": 1.0618, + "learning_rate": 0.0002, + "loss": 1.0706, "step": 1627 }, { "epoch": 1.77, - "learning_rate": 0.0005, - "loss": 1.0741, + "learning_rate": 0.0002, + "loss": 1.103, "step": 1628 }, { "epoch": 1.77, - "learning_rate": 0.0005, - "loss": 1.0523, + "learning_rate": 0.0002, + "loss": 1.0743, "step": 1629 }, { "epoch": 1.77, - "learning_rate": 0.0005, - "loss": 1.0663, + "learning_rate": 0.0002, + "loss": 1.0881, "step": 1630 }, { "epoch": 1.77, - "learning_rate": 0.0005, - "loss": 1.0473, + "learning_rate": 0.0002, + "loss": 1.0619, "step": 1631 }, { "epoch": 1.78, - "learning_rate": 0.0005, - "loss": 1.049, + "learning_rate": 0.0002, + "loss": 1.06, "step": 1632 }, { "epoch": 1.78, - "learning_rate": 0.0005, - "loss": 1.0786, + "learning_rate": 0.0002, + "loss": 1.0998, "step": 1633 }, { "epoch": 1.78, - "learning_rate": 0.0005, - "loss": 1.0845, + "learning_rate": 0.0002, + "loss": 1.1068, "step": 1634 }, { "epoch": 1.78, - "learning_rate": 0.0005, - "loss": 1.027, + "learning_rate": 0.0002, + "loss": 1.0474, "step": 1635 }, { "epoch": 1.78, - "learning_rate": 0.0005, - "loss": 1.069, + "learning_rate": 0.0002, + "loss": 1.0785, "step": 1636 }, { "epoch": 1.78, - "learning_rate": 0.0005, - "loss": 1.1047, + "learning_rate": 0.0002, + "loss": 1.13, "step": 1637 }, { "epoch": 1.78, - "learning_rate": 0.0005, - "loss": 1.0683, + "learning_rate": 0.0002, + "loss": 1.1035, "step": 1638 }, { "epoch": 1.78, - "learning_rate": 0.0005, - "loss": 1.0723, + "learning_rate": 0.0002, + "loss": 1.0922, "step": 1639 }, { "epoch": 1.78, - "learning_rate": 0.0005, - "loss": 1.039, + "learning_rate": 0.0002, + "loss": 1.0578, "step": 1640 }, { "epoch": 1.79, - "learning_rate": 0.0005, - "loss": 1.0421, + "learning_rate": 0.0002, + "loss": 1.0659, "step": 1641 }, { "epoch": 1.79, - "learning_rate": 0.0005, - "loss": 1.0553, + "learning_rate": 0.0002, + "loss": 1.0816, "step": 1642 }, { "epoch": 1.79, - "learning_rate": 0.0005, - "loss": 1.0307, + "learning_rate": 0.0002, + "loss": 1.0447, "step": 1643 }, { "epoch": 1.79, - "learning_rate": 0.0005, - "loss": 1.1121, + "learning_rate": 0.0002, + "loss": 1.1344, "step": 1644 }, { "epoch": 1.79, - "learning_rate": 0.0005, - "loss": 1.0692, + "learning_rate": 0.0002, + "loss": 1.0885, "step": 1645 }, { "epoch": 1.79, - "learning_rate": 0.0005, - "loss": 1.0291, + "learning_rate": 0.0002, + "loss": 1.0491, "step": 1646 }, { "epoch": 1.79, - "learning_rate": 0.0005, - "loss": 1.0513, + "learning_rate": 0.0002, + "loss": 1.0697, "step": 1647 }, { "epoch": 1.79, - "learning_rate": 0.0005, - "loss": 1.0492, + "learning_rate": 0.0002, + "loss": 1.0644, "step": 1648 }, { "epoch": 1.79, - "learning_rate": 0.0005, - "loss": 1.0738, + "learning_rate": 0.0002, + "loss": 1.0906, "step": 1649 }, { "epoch": 1.8, - "learning_rate": 0.0005, - "loss": 1.044, + "learning_rate": 0.0002, + "loss": 1.0698, "step": 1650 }, { "epoch": 1.8, - "learning_rate": 0.0005, - "loss": 1.0605, + "learning_rate": 0.0002, + "loss": 1.0775, "step": 1651 }, { "epoch": 1.8, - "learning_rate": 0.0005, - "loss": 1.0652, + "learning_rate": 0.0002, + "loss": 1.0866, "step": 1652 }, { "epoch": 1.8, - "learning_rate": 0.0005, - "loss": 1.0406, + "learning_rate": 0.0002, + "loss": 1.0604, "step": 1653 }, { "epoch": 1.8, - "learning_rate": 0.0005, - "loss": 1.0465, + "learning_rate": 0.0002, + "loss": 1.065, "step": 1654 }, { "epoch": 1.8, - "learning_rate": 0.0005, - "loss": 1.0462, + "learning_rate": 0.0002, + "loss": 1.0728, "step": 1655 }, { "epoch": 1.8, - "learning_rate": 0.0005, - "loss": 1.0285, + "learning_rate": 0.0002, + "loss": 1.0464, "step": 1656 }, { "epoch": 1.8, - "learning_rate": 0.0005, - "loss": 1.0488, + "learning_rate": 0.0002, + "loss": 1.0785, "step": 1657 }, { "epoch": 1.8, - "learning_rate": 0.0005, - "loss": 1.0414, + "learning_rate": 0.0002, + "loss": 1.0604, "step": 1658 }, { "epoch": 1.81, - "learning_rate": 0.0005, - "loss": 1.0455, + "learning_rate": 0.0002, + "loss": 1.0695, "step": 1659 }, { "epoch": 1.81, - "learning_rate": 0.0005, - "loss": 1.0859, + "learning_rate": 0.0002, + "loss": 1.1152, "step": 1660 }, { "epoch": 1.81, - "learning_rate": 0.0005, - "loss": 1.0607, + "learning_rate": 0.0002, + "loss": 1.0792, "step": 1661 }, { "epoch": 1.81, - "learning_rate": 0.0005, - "loss": 0.9988, + "learning_rate": 0.0002, + "loss": 1.0215, "step": 1662 }, { "epoch": 1.81, - "learning_rate": 0.0005, - "loss": 1.0939, + "learning_rate": 0.0002, + "loss": 1.1137, "step": 1663 }, { "epoch": 1.81, - "learning_rate": 0.0005, - "loss": 1.0538, + "learning_rate": 0.0002, + "loss": 1.067, "step": 1664 }, { "epoch": 1.81, - "learning_rate": 0.0005, - "loss": 0.9807, + "learning_rate": 0.0002, + "loss": 1.0039, "step": 1665 }, { "epoch": 1.81, - "learning_rate": 0.0005, - "loss": 1.045, + "learning_rate": 0.0002, + "loss": 1.0679, "step": 1666 }, { "epoch": 1.81, - "learning_rate": 0.0005, - "loss": 1.0496, + "learning_rate": 0.0002, + "loss": 1.0668, "step": 1667 }, { "epoch": 1.82, - "learning_rate": 0.0005, - "loss": 1.0269, + "learning_rate": 0.0002, + "loss": 1.0544, "step": 1668 }, { "epoch": 1.82, - "learning_rate": 0.0005, - "loss": 0.9985, + "learning_rate": 0.0002, + "loss": 1.024, "step": 1669 }, { "epoch": 1.82, - "learning_rate": 0.0005, - "loss": 1.0291, + "learning_rate": 0.0002, + "loss": 1.0508, "step": 1670 }, { "epoch": 1.82, - "learning_rate": 0.0005, - "loss": 1.0389, + "learning_rate": 0.0002, + "loss": 1.0575, "step": 1671 }, { "epoch": 1.82, - "learning_rate": 0.0005, - "loss": 1.0275, + "learning_rate": 0.0002, + "loss": 1.0486, "step": 1672 }, { "epoch": 1.82, - "learning_rate": 0.0005, - "loss": 1.0361, + "learning_rate": 0.0002, + "loss": 1.055, "step": 1673 }, { "epoch": 1.82, - "learning_rate": 0.0005, - "loss": 1.067, + "learning_rate": 0.0002, + "loss": 1.0975, "step": 1674 }, { "epoch": 1.82, - "learning_rate": 0.0005, - "loss": 1.044, + "learning_rate": 0.0002, + "loss": 1.0743, "step": 1675 }, { "epoch": 1.82, - "learning_rate": 0.0005, - "loss": 1.0223, + "learning_rate": 0.0002, + "loss": 1.0436, "step": 1676 }, { "epoch": 1.82, - "learning_rate": 0.0005, - "loss": 1.0423, + "learning_rate": 0.0002, + "loss": 1.0542, "step": 1677 }, { "epoch": 1.83, - "learning_rate": 0.0005, - "loss": 1.0932, + "learning_rate": 0.0002, + "loss": 1.1132, "step": 1678 }, { "epoch": 1.83, - "learning_rate": 0.0005, - "loss": 1.0338, + "learning_rate": 0.0002, + "loss": 1.0539, "step": 1679 }, { "epoch": 1.83, - "learning_rate": 0.0005, - "loss": 1.0288, + "learning_rate": 0.0002, + "loss": 1.0507, "step": 1680 }, { "epoch": 1.83, - "learning_rate": 0.0005, - "loss": 1.0274, + "learning_rate": 0.0002, + "loss": 1.0468, "step": 1681 }, { "epoch": 1.83, - "learning_rate": 0.0005, - "loss": 1.0538, + "learning_rate": 0.0002, + "loss": 1.0698, "step": 1682 }, { "epoch": 1.83, - "learning_rate": 0.0005, - "loss": 1.0579, + "learning_rate": 0.0002, + "loss": 1.0788, "step": 1683 }, { "epoch": 1.83, - "learning_rate": 0.0005, - "loss": 1.0328, + "learning_rate": 0.0002, + "loss": 1.0609, "step": 1684 }, { "epoch": 1.83, - "learning_rate": 0.0005, - "loss": 1.0817, + "learning_rate": 0.0002, + "loss": 1.1079, "step": 1685 }, { "epoch": 1.83, - "learning_rate": 0.0005, - "loss": 1.0844, + "learning_rate": 0.0002, + "loss": 1.1017, "step": 1686 }, { "epoch": 1.84, - "learning_rate": 0.0005, - "loss": 1.0413, + "learning_rate": 0.0002, + "loss": 1.0641, "step": 1687 }, { "epoch": 1.84, - "learning_rate": 0.0005, - "loss": 1.0315, + "learning_rate": 0.0002, + "loss": 1.0499, "step": 1688 }, { "epoch": 1.84, - "learning_rate": 0.0005, - "loss": 1.0591, + "learning_rate": 0.0002, + "loss": 1.0798, "step": 1689 }, { "epoch": 1.84, - "learning_rate": 0.0005, - "loss": 1.0144, + "learning_rate": 0.0002, + "loss": 1.0303, "step": 1690 }, { "epoch": 1.84, - "learning_rate": 0.0005, - "loss": 1.0756, + "learning_rate": 0.0002, + "loss": 1.0989, "step": 1691 }, { "epoch": 1.84, - "learning_rate": 0.0005, - "loss": 1.0792, + "learning_rate": 0.0002, + "loss": 1.0973, "step": 1692 }, { "epoch": 1.84, - "learning_rate": 0.0005, - "loss": 1.0843, + "learning_rate": 0.0002, + "loss": 1.113, "step": 1693 }, { "epoch": 1.84, - "learning_rate": 0.0005, - "loss": 1.0622, + "learning_rate": 0.0002, + "loss": 1.079, "step": 1694 }, { "epoch": 1.84, - "learning_rate": 0.0005, - "loss": 1.0291, + "learning_rate": 0.0002, + "loss": 1.0472, "step": 1695 }, { "epoch": 1.85, - "learning_rate": 0.0005, - "loss": 1.0901, + "learning_rate": 0.0002, + "loss": 1.1059, "step": 1696 }, { "epoch": 1.85, - "learning_rate": 0.0005, - "loss": 1.0376, + "learning_rate": 0.0002, + "loss": 1.0465, "step": 1697 }, { "epoch": 1.85, - "learning_rate": 0.0005, - "loss": 1.0554, + "learning_rate": 0.0002, + "loss": 1.0775, "step": 1698 }, { "epoch": 1.85, - "learning_rate": 0.0005, - "loss": 1.019, + "learning_rate": 0.0002, + "loss": 1.0323, "step": 1699 }, { "epoch": 1.85, - "learning_rate": 0.0005, - "loss": 1.027, + "learning_rate": 0.0002, + "loss": 1.0546, "step": 1700 }, { "epoch": 1.85, - "learning_rate": 0.0005, - "loss": 1.0869, + "learning_rate": 0.0002, + "loss": 1.104, "step": 1701 }, { "epoch": 1.85, - "learning_rate": 0.0005, - "loss": 1.0448, + "learning_rate": 0.0002, + "loss": 1.0757, "step": 1702 }, { "epoch": 1.85, - "learning_rate": 0.0005, - "loss": 1.02, + "learning_rate": 0.0002, + "loss": 1.0416, "step": 1703 }, { "epoch": 1.85, - "learning_rate": 0.0005, - "loss": 1.03, + "learning_rate": 0.0002, + "loss": 1.0463, "step": 1704 }, { "epoch": 1.86, - "learning_rate": 0.0005, - "loss": 1.0786, + "learning_rate": 0.0002, + "loss": 1.1015, "step": 1705 }, { "epoch": 1.86, - "learning_rate": 0.0005, - "loss": 1.026, + "learning_rate": 0.0002, + "loss": 1.0471, "step": 1706 }, { "epoch": 1.86, - "learning_rate": 0.0005, - "loss": 1.0517, + "learning_rate": 0.0002, + "loss": 1.0696, "step": 1707 }, { "epoch": 1.86, - "learning_rate": 0.0005, - "loss": 1.0357, + "learning_rate": 0.0002, + "loss": 1.0473, "step": 1708 }, { "epoch": 1.86, - "learning_rate": 0.0005, - "loss": 1.0442, + "learning_rate": 0.0002, + "loss": 1.0713, "step": 1709 }, { "epoch": 1.86, - "learning_rate": 0.0005, - "loss": 1.0792, + "learning_rate": 0.0002, + "loss": 1.1059, "step": 1710 }, { "epoch": 1.86, - "learning_rate": 0.0005, - "loss": 1.0251, + "learning_rate": 0.0002, + "loss": 1.0414, "step": 1711 }, { "epoch": 1.86, - "learning_rate": 0.0005, - "loss": 1.0272, + "learning_rate": 0.0002, + "loss": 1.0475, "step": 1712 }, { "epoch": 1.86, - "learning_rate": 0.0005, - "loss": 1.0133, + "learning_rate": 0.0002, + "loss": 1.0325, "step": 1713 }, { "epoch": 1.87, - "learning_rate": 0.0005, - "loss": 1.0723, + "learning_rate": 0.0002, + "loss": 1.0937, "step": 1714 }, { "epoch": 1.87, - "learning_rate": 0.0005, - "loss": 1.085, + "learning_rate": 0.0002, + "loss": 1.1109, "step": 1715 }, { "epoch": 1.87, - "learning_rate": 0.0005, - "loss": 1.0818, + "learning_rate": 0.0002, + "loss": 1.0972, "step": 1716 }, { "epoch": 1.87, - "learning_rate": 0.0005, - "loss": 1.0687, + "learning_rate": 0.0002, + "loss": 1.0964, "step": 1717 }, { "epoch": 1.87, - "learning_rate": 0.0005, - "loss": 1.0826, + "learning_rate": 0.0002, + "loss": 1.104, "step": 1718 }, { "epoch": 1.87, - "learning_rate": 0.0005, - "loss": 1.0578, + "learning_rate": 0.0002, + "loss": 1.083, "step": 1719 }, { "epoch": 1.87, - "learning_rate": 0.0005, - "loss": 1.0442, + "learning_rate": 0.0002, + "loss": 1.0547, "step": 1720 }, { "epoch": 1.87, - "learning_rate": 0.0005, - "loss": 1.1147, + "learning_rate": 0.0002, + "loss": 1.1362, "step": 1721 }, { "epoch": 1.87, - "learning_rate": 0.0005, - "loss": 1.0998, + "learning_rate": 0.0002, + "loss": 1.1169, "step": 1722 }, { "epoch": 1.87, - "learning_rate": 0.0005, - "loss": 1.0336, + "learning_rate": 0.0002, + "loss": 1.0536, "step": 1723 }, { "epoch": 1.88, - "learning_rate": 0.0005, - "loss": 1.0643, + "learning_rate": 0.0002, + "loss": 1.0828, "step": 1724 }, { "epoch": 1.88, - "learning_rate": 0.0005, - "loss": 1.0711, + "learning_rate": 0.0002, + "loss": 1.09, "step": 1725 }, { "epoch": 1.88, - "learning_rate": 0.0005, - "loss": 1.0225, + "learning_rate": 0.0002, + "loss": 1.041, "step": 1726 }, { "epoch": 1.88, - "learning_rate": 0.0005, - "loss": 1.0709, + "learning_rate": 0.0002, + "loss": 1.086, "step": 1727 }, { "epoch": 1.88, - "learning_rate": 0.0005, - "loss": 1.0644, + "learning_rate": 0.0002, + "loss": 1.0824, "step": 1728 }, { "epoch": 1.88, - "learning_rate": 0.0005, - "loss": 1.031, + "learning_rate": 0.0002, + "loss": 1.05, "step": 1729 }, { "epoch": 1.88, - "learning_rate": 0.0005, - "loss": 1.015, + "learning_rate": 0.0002, + "loss": 1.0284, "step": 1730 }, { "epoch": 1.88, - "learning_rate": 0.0005, - "loss": 1.0506, + "learning_rate": 0.0002, + "loss": 1.0706, "step": 1731 }, { "epoch": 1.88, - "learning_rate": 0.0005, - "loss": 1.0409, + "learning_rate": 0.0002, + "loss": 1.0557, "step": 1732 }, { "epoch": 1.89, - "learning_rate": 0.0005, - "loss": 1.0748, + "learning_rate": 0.0002, + "loss": 1.0896, "step": 1733 }, { "epoch": 1.89, - "learning_rate": 0.0005, - "loss": 1.0497, + "learning_rate": 0.0002, + "loss": 1.0717, "step": 1734 }, { "epoch": 1.89, - "learning_rate": 0.0005, - "loss": 1.038, + "learning_rate": 0.0002, + "loss": 1.0614, "step": 1735 }, { "epoch": 1.89, - "learning_rate": 0.0005, - "loss": 1.04, + "learning_rate": 0.0002, + "loss": 1.057, "step": 1736 }, { "epoch": 1.89, - "learning_rate": 0.0005, - "loss": 1.0916, + "learning_rate": 0.0002, + "loss": 1.1028, "step": 1737 }, { "epoch": 1.89, - "learning_rate": 0.0005, - "loss": 1.0497, + "learning_rate": 0.0002, + "loss": 1.0683, "step": 1738 }, { "epoch": 1.89, - "learning_rate": 0.0005, - "loss": 1.0334, + "learning_rate": 0.0002, + "loss": 1.0581, "step": 1739 }, { "epoch": 1.89, - "learning_rate": 0.0005, - "loss": 1.0842, + "learning_rate": 0.0002, + "loss": 1.1042, "step": 1740 }, { "epoch": 1.89, - "learning_rate": 0.0005, - "loss": 1.0469, + "learning_rate": 0.0002, + "loss": 1.0602, "step": 1741 }, { "epoch": 1.9, - "learning_rate": 0.0005, - "loss": 1.0582, + "learning_rate": 0.0002, + "loss": 1.0717, "step": 1742 }, { "epoch": 1.9, - "learning_rate": 0.0005, - "loss": 0.9828, + "learning_rate": 0.0002, + "loss": 1.0029, "step": 1743 }, { "epoch": 1.9, - "learning_rate": 0.0005, - "loss": 0.9824, + "learning_rate": 0.0002, + "loss": 1.0084, "step": 1744 }, { "epoch": 1.9, - "learning_rate": 0.0005, - "loss": 1.0304, + "learning_rate": 0.0002, + "loss": 1.0559, "step": 1745 }, { "epoch": 1.9, - "learning_rate": 0.0005, - "loss": 1.0482, + "learning_rate": 0.0002, + "loss": 1.0647, "step": 1746 }, { "epoch": 1.9, - "learning_rate": 0.0005, - "loss": 1.0289, + "learning_rate": 0.0002, + "loss": 1.0551, "step": 1747 }, { "epoch": 1.9, - "learning_rate": 0.0005, - "loss": 1.0495, + "learning_rate": 0.0002, + "loss": 1.0712, "step": 1748 }, { "epoch": 1.9, - "learning_rate": 0.0005, - "loss": 1.0416, + "learning_rate": 0.0002, + "loss": 1.0631, "step": 1749 }, { "epoch": 1.9, - "learning_rate": 0.0005, - "loss": 1.0249, + "learning_rate": 0.0002, + "loss": 1.0455, "step": 1750 }, { "epoch": 1.91, - "learning_rate": 0.0005, - "loss": 1.0869, + "learning_rate": 0.0002, + "loss": 1.1175, "step": 1751 }, { "epoch": 1.91, - "learning_rate": 0.0005, - "loss": 1.0724, + "learning_rate": 0.0002, + "loss": 1.0987, "step": 1752 }, { "epoch": 1.91, - "learning_rate": 0.0005, - "loss": 1.0462, + "learning_rate": 0.0002, + "loss": 1.067, "step": 1753 }, { "epoch": 1.91, - "learning_rate": 0.0005, - "loss": 1.0432, + "learning_rate": 0.0002, + "loss": 1.0602, "step": 1754 }, { "epoch": 1.91, - "learning_rate": 0.0005, - "loss": 0.5345, + "learning_rate": 0.0002, + "loss": 1.0867, "step": 1755 }, { "epoch": 1.91, - "learning_rate": 0.0005, - "loss": 1.0058, + "learning_rate": 0.0002, + "loss": 1.0233, "step": 1756 }, { "epoch": 1.91, - "learning_rate": 0.0005, - "loss": 1.0624, + "learning_rate": 0.0002, + "loss": 1.0884, "step": 1757 }, { "epoch": 1.91, - "learning_rate": 0.0005, - "loss": 0.5172, + "learning_rate": 0.0002, + "loss": 1.0588, "step": 1758 }, { "epoch": 1.91, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1055, "step": 1759 }, { "epoch": 1.92, - "learning_rate": 0.0005, - "loss": 1.0687, + "learning_rate": 0.0002, + "loss": 1.1175, "step": 1760 }, { "epoch": 1.92, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0853, "step": 1761 }, { "epoch": 1.92, - "learning_rate": 0.0005, - "loss": 1.0012, + "learning_rate": 0.0002, + "loss": 1.0336, "step": 1762 }, { "epoch": 1.92, - "learning_rate": 0.0005, - "loss": 0.5264, + "learning_rate": 0.0002, + "loss": 1.0736, "step": 1763 }, { "epoch": 1.92, - "learning_rate": 0.0005, - "loss": 0.5395, + "learning_rate": 0.0002, + "loss": 1.0722, "step": 1764 }, { "epoch": 1.92, - "learning_rate": 0.0005, - "loss": 1.078, + "learning_rate": 0.0002, + "loss": 1.1204, "step": 1765 }, { "epoch": 1.92, - "learning_rate": 0.0005, - "loss": 0.5728, + "learning_rate": 0.0002, + "loss": 1.1288, "step": 1766 }, { "epoch": 1.92, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0919, "step": 1767 }, { "epoch": 1.92, - "learning_rate": 0.0005, - "loss": 0.5216, + "learning_rate": 0.0002, + "loss": 1.0804, "step": 1768 }, { "epoch": 1.92, - "learning_rate": 0.0005, - "loss": 0.5086, + "learning_rate": 0.0002, + "loss": 1.0562, "step": 1769 }, { "epoch": 1.93, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0683, "step": 1770 }, { "epoch": 1.93, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1118, "step": 1771 }, { "epoch": 1.93, - "learning_rate": 0.0005, - "loss": 0.544, + "learning_rate": 0.0002, + "loss": 1.1094, "step": 1772 }, { "epoch": 1.93, - "learning_rate": 0.0005, - "loss": 0.5307, + "learning_rate": 0.0002, + "loss": 1.098, "step": 1773 }, { "epoch": 1.93, - "learning_rate": 0.0005, - "loss": 1.0104, + "learning_rate": 0.0002, + "loss": 1.0964, "step": 1774 }, { "epoch": 1.93, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0628, "step": 1775 }, { "epoch": 1.93, - "learning_rate": 0.0005, - "loss": 1.0986, + "learning_rate": 0.0002, + "loss": 1.086, "step": 1776 }, { "epoch": 1.93, - "learning_rate": 0.0005, - "loss": 1.0599, + "learning_rate": 0.0002, + "loss": 1.0353, "step": 1777 }, { "epoch": 1.93, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1118, "step": 1778 }, { "epoch": 1.94, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0363, "step": 1779 }, { "epoch": 1.94, - "learning_rate": 0.0005, - "loss": 0.98, + "learning_rate": 0.0002, + "loss": 1.0424, "step": 1780 }, { "epoch": 1.94, - "learning_rate": 0.0005, - "loss": 1.0231, + "learning_rate": 0.0002, + "loss": 1.0478, "step": 1781 }, { "epoch": 1.94, - "learning_rate": 0.0005, - "loss": 1.0341, + "learning_rate": 0.0002, + "loss": 1.0435, "step": 1782 }, { "epoch": 1.94, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.051, "step": 1783 }, { "epoch": 1.94, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0812, "step": 1784 }, { "epoch": 1.94, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0355, "step": 1785 }, { "epoch": 1.94, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0882, "step": 1786 }, { "epoch": 1.94, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0528, "step": 1787 }, { "epoch": 1.95, - "learning_rate": 0.0005, - "loss": 1.081, + "learning_rate": 0.0002, + "loss": 1.1006, "step": 1788 }, { "epoch": 1.95, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0759, "step": 1789 }, { "epoch": 1.95, - "learning_rate": 0.0005, - "loss": 0.5163, + "learning_rate": 0.0002, + "loss": 1.0502, "step": 1790 }, { "epoch": 1.95, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1549, "step": 1791 }, { "epoch": 1.95, - "learning_rate": 0.0005, - "loss": 1.0459, + "learning_rate": 0.0002, + "loss": 1.0636, "step": 1792 }, { "epoch": 1.95, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0806, "step": 1793 }, { "epoch": 1.95, - "learning_rate": 0.0005, - "loss": 0.5214, + "learning_rate": 0.0002, + "loss": 1.0902, "step": 1794 }, { "epoch": 1.95, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0368, "step": 1795 }, { "epoch": 1.95, - "learning_rate": 0.0005, - "loss": 1.0589, + "learning_rate": 0.0002, + "loss": 1.0716, "step": 1796 }, { "epoch": 1.96, - "learning_rate": 0.0005, - "loss": 1.0322, + "learning_rate": 0.0002, + "loss": 1.0585, "step": 1797 }, { "epoch": 1.96, - "learning_rate": 0.0005, - "loss": 1.0322, + "learning_rate": 0.0002, + "loss": 1.068, "step": 1798 }, { "epoch": 1.96, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0274, "step": 1799 }, { "epoch": 1.96, - "learning_rate": 0.0005, - "loss": 1.0572, + "learning_rate": 0.0002, + "loss": 1.0769, "step": 1800 }, { "epoch": 1.96, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0562, "step": 1801 }, { "epoch": 1.96, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1172, "step": 1802 }, { "epoch": 1.96, - "learning_rate": 0.0005, - "loss": 0.5612, + "learning_rate": 0.0002, + "loss": 1.0914, "step": 1803 }, { "epoch": 1.96, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0649, "step": 1804 }, { "epoch": 1.96, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0557, "step": 1805 }, { "epoch": 1.97, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0552, "step": 1806 }, { "epoch": 1.97, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0364, "step": 1807 }, { "epoch": 1.97, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.041, "step": 1808 }, { "epoch": 1.97, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0634, "step": 1809 }, { "epoch": 1.97, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0679, "step": 1810 }, { "epoch": 1.97, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1204, "step": 1811 }, { "epoch": 1.97, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1018, "step": 1812 }, { "epoch": 1.97, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0635, "step": 1813 }, { "epoch": 1.97, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0622, "step": 1814 }, { "epoch": 1.97, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0994, "step": 1815 }, { "epoch": 1.98, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0679, "step": 1816 }, { "epoch": 1.98, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0537, "step": 1817 }, { "epoch": 1.98, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0729, "step": 1818 }, { "epoch": 1.98, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0631, "step": 1819 }, { "epoch": 1.98, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0619, "step": 1820 }, { "epoch": 1.98, - "learning_rate": 0.0005, - "loss": 0.9987, + "learning_rate": 0.0002, + "loss": 1.0248, "step": 1821 }, { "epoch": 1.98, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.07, "step": 1822 }, { "epoch": 1.98, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0362, "step": 1823 }, { "epoch": 1.98, - "learning_rate": 0.0005, - "loss": 0.545, + "learning_rate": 0.0002, + "loss": 1.0755, "step": 1824 }, { "epoch": 1.99, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0747, "step": 1825 }, { "epoch": 1.99, - "learning_rate": 0.0005, - "loss": 0.5142, + "learning_rate": 0.0002, + "loss": 1.0823, "step": 1826 }, { "epoch": 1.99, - "learning_rate": 0.0005, - "loss": 0.5121, + "learning_rate": 0.0002, + "loss": 1.0641, "step": 1827 }, { "epoch": 1.99, - "learning_rate": 0.0005, - "loss": 1.0752, + "learning_rate": 0.0002, + "loss": 1.0937, "step": 1828 }, { "epoch": 1.99, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1267, "step": 1829 }, { "epoch": 1.99, - "learning_rate": 0.0005, - "loss": 1.0843, + "learning_rate": 0.0002, + "loss": 1.0562, "step": 1830 }, { "epoch": 1.99, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0454, "step": 1831 }, { "epoch": 1.99, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1055, "step": 1832 }, { "epoch": 1.99, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0666, "step": 1833 }, { "epoch": 2.0, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0456, "step": 1834 }, { "epoch": 2.0, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0449, "step": 1835 }, { "epoch": 2.0, - "learning_rate": 0.0005, - "loss": 0.521, + "learning_rate": 0.0002, + "loss": 1.0725, "step": 1836 }, { "epoch": 2.0, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0931, "step": 1837 }, { "epoch": 2.0, - "learning_rate": 0.0005, - "loss": 1.0824, + "learning_rate": 0.0002, + "loss": 1.1026, "step": 1838 }, { "epoch": 2.0, - "learning_rate": 0.0005, - "loss": 0.4879, + "learning_rate": 0.0002, + "loss": 0.9938, "step": 1839 }, { "epoch": 2.0, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1183, "step": 1840 }, { "epoch": 2.0, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0488, "step": 1841 }, { "epoch": 2.0, - "learning_rate": 0.0005, - "loss": 0.5068, + "learning_rate": 0.0002, + "loss": 1.0333, "step": 1842 }, { "epoch": 2.01, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1488, "step": 1843 }, { "epoch": 2.01, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0562, "step": 1844 }, { "epoch": 2.01, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0662, "step": 1845 }, { "epoch": 2.01, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0566, "step": 1846 }, { "epoch": 2.01, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0706, "step": 1847 }, { "epoch": 2.01, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.062, "step": 1848 }, { "epoch": 2.01, - "learning_rate": 0.0005, - "loss": 0.9848, + "learning_rate": 0.0002, + "loss": 1.0285, "step": 1849 }, { "epoch": 2.01, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0642, "step": 1850 }, { "epoch": 2.01, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0624, "step": 1851 }, { "epoch": 2.02, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0577, "step": 1852 }, { "epoch": 2.02, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0784, "step": 1853 }, { "epoch": 2.02, - "learning_rate": 0.0005, - "loss": 0.5532, + "learning_rate": 0.0002, + "loss": 1.101, "step": 1854 }, { "epoch": 2.02, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0111, "step": 1855 }, { "epoch": 2.02, - "learning_rate": 0.0005, - "loss": 0.995, + "learning_rate": 0.0002, + "loss": 1.0305, "step": 1856 }, { "epoch": 2.02, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0546, "step": 1857 }, { "epoch": 2.02, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0536, "step": 1858 }, { "epoch": 2.02, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0999, "step": 1859 }, { "epoch": 2.02, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0773, "step": 1860 }, { "epoch": 2.03, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0338, "step": 1861 }, { "epoch": 2.03, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0466, "step": 1862 }, { "epoch": 2.03, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0144, "step": 1863 }, { "epoch": 2.03, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0472, "step": 1864 }, { "epoch": 2.03, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0678, "step": 1865 }, { "epoch": 2.03, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0662, "step": 1866 }, { "epoch": 2.03, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0472, "step": 1867 }, { "epoch": 2.03, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0552, "step": 1868 }, { "epoch": 2.03, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0668, "step": 1869 }, { "epoch": 2.03, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.065, "step": 1870 }, { "epoch": 2.04, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0376, "step": 1871 }, { "epoch": 2.04, - "learning_rate": 0.0005, - "loss": 0.5513, + "learning_rate": 0.0002, + "loss": 1.0669, "step": 1872 }, { "epoch": 2.04, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0324, "step": 1873 }, { "epoch": 2.04, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0821, "step": 1874 }, { "epoch": 2.04, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0652, "step": 1875 }, { "epoch": 2.04, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0768, "step": 1876 }, { "epoch": 2.04, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0483, "step": 1877 }, { "epoch": 2.04, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0473, "step": 1878 }, { "epoch": 2.04, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0736, "step": 1879 }, { "epoch": 2.05, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0263, "step": 1880 }, { "epoch": 2.05, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0845, "step": 1881 }, { "epoch": 2.05, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1175, "step": 1882 }, { "epoch": 2.05, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0506, "step": 1883 }, { "epoch": 2.05, - "learning_rate": 0.0005, - "loss": 0.518, + "learning_rate": 0.0002, + "loss": 1.0415, "step": 1884 }, { "epoch": 2.05, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0417, "step": 1885 }, { "epoch": 2.05, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0319, "step": 1886 }, { "epoch": 2.05, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0258, "step": 1887 }, { "epoch": 2.05, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0813, "step": 1888 }, { "epoch": 2.06, - "learning_rate": 0.0005, - "loss": 0.977, + "learning_rate": 0.0002, + "loss": 1.0251, "step": 1889 }, { "epoch": 2.06, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0687, "step": 1890 }, { "epoch": 2.06, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.027, "step": 1891 }, { "epoch": 2.06, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0514, "step": 1892 }, { "epoch": 2.06, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0545, "step": 1893 }, { "epoch": 2.06, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0606, "step": 1894 }, { "epoch": 2.06, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0414, "step": 1895 }, { "epoch": 2.06, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0475, "step": 1896 }, { "epoch": 2.06, - "learning_rate": 0.0005, - "loss": 0.5218, + "learning_rate": 0.0002, + "loss": 1.0565, "step": 1897 }, { "epoch": 2.07, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0519, "step": 1898 }, { "epoch": 2.07, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0351, "step": 1899 }, { "epoch": 2.07, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0306, "step": 1900 }, { "epoch": 2.07, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0556, "step": 1901 }, { "epoch": 2.07, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0397, "step": 1902 }, { "epoch": 2.07, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0369, "step": 1903 }, { "epoch": 2.07, - "learning_rate": 0.0005, - "loss": 0.5137, + "learning_rate": 0.0002, + "loss": 1.0472, "step": 1904 }, { "epoch": 2.07, - "learning_rate": 0.0005, - "loss": 0.9674, + "learning_rate": 0.0002, + "loss": 1.0055, "step": 1905 }, { "epoch": 2.07, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0234, "step": 1906 }, { "epoch": 2.08, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0214, "step": 1907 }, { "epoch": 2.08, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.048, "step": 1908 }, { "epoch": 2.08, - "learning_rate": 0.0005, - "loss": 0.538, + "learning_rate": 0.0002, + "loss": 1.0917, "step": 1909 }, { "epoch": 2.08, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0781, "step": 1910 }, { "epoch": 2.08, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0675, "step": 1911 }, { "epoch": 2.08, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1008, "step": 1912 }, { "epoch": 2.08, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0367, "step": 1913 }, { "epoch": 2.08, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0173, "step": 1914 }, { "epoch": 2.08, - "learning_rate": 0.0005, - "loss": 0.9889, + "learning_rate": 0.0002, + "loss": 1.0151, "step": 1915 }, { "epoch": 2.08, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0217, "step": 1916 }, { "epoch": 2.09, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0771, "step": 1917 }, { "epoch": 2.09, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0868, "step": 1918 }, { "epoch": 2.09, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0684, "step": 1919 }, { "epoch": 2.09, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0417, "step": 1920 }, { "epoch": 2.09, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0743, "step": 1921 }, { "epoch": 2.09, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0476, "step": 1922 }, { "epoch": 2.09, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0472, "step": 1923 }, { "epoch": 2.09, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0432, "step": 1924 }, { "epoch": 2.09, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1152, "step": 1925 }, { "epoch": 2.1, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0392, "step": 1926 }, { "epoch": 2.1, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0186, "step": 1927 }, { "epoch": 2.1, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0613, "step": 1928 }, { "epoch": 2.1, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 0.9918, "step": 1929 }, { "epoch": 2.1, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0507, "step": 1930 }, { "epoch": 2.1, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.043, "step": 1931 }, { "epoch": 2.1, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0853, "step": 1932 }, { "epoch": 2.1, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0326, "step": 1933 }, { "epoch": 2.1, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0125, "step": 1934 }, { "epoch": 2.11, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0715, "step": 1935 }, { "epoch": 2.11, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0208, "step": 1936 }, { "epoch": 2.11, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0526, "step": 1937 }, { "epoch": 2.11, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0724, "step": 1938 }, { "epoch": 2.11, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0486, "step": 1939 }, { "epoch": 2.11, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0379, "step": 1940 }, { "epoch": 2.11, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1085, "step": 1941 }, { "epoch": 2.11, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0958, "step": 1942 }, { "epoch": 2.11, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1236, "step": 1943 }, { "epoch": 2.12, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0669, "step": 1944 }, { "epoch": 2.12, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.072, "step": 1945 }, { "epoch": 2.12, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0219, "step": 1946 }, { "epoch": 2.12, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0094, "step": 1947 }, { "epoch": 2.12, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0285, "step": 1948 }, { "epoch": 2.12, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0575, "step": 1949 }, { "epoch": 2.12, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.028, "step": 1950 }, { "epoch": 2.12, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0739, "step": 1951 }, { "epoch": 2.12, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0703, "step": 1952 }, { "epoch": 2.13, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0468, "step": 1953 }, { "epoch": 2.13, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0595, "step": 1954 }, { "epoch": 2.13, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0145, "step": 1955 }, { "epoch": 2.13, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1069, "step": 1956 }, { "epoch": 2.13, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.048, "step": 1957 }, { "epoch": 2.13, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0306, "step": 1958 }, { "epoch": 2.13, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.091, "step": 1959 }, { "epoch": 2.13, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0388, "step": 1960 }, { "epoch": 2.13, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0289, "step": 1961 }, { "epoch": 2.13, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0692, "step": 1962 }, { "epoch": 2.14, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0455, "step": 1963 }, { "epoch": 2.14, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0679, "step": 1964 }, { "epoch": 2.14, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0297, "step": 1965 }, { "epoch": 2.14, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0691, "step": 1966 }, { "epoch": 2.14, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0311, "step": 1967 }, { "epoch": 2.14, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0125, "step": 1968 }, { "epoch": 2.14, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0262, "step": 1969 }, { "epoch": 2.14, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0795, "step": 1970 }, { "epoch": 2.14, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0346, "step": 1971 }, { "epoch": 2.15, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0367, "step": 1972 }, { "epoch": 2.15, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0487, "step": 1973 }, { "epoch": 2.15, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0483, "step": 1974 }, { "epoch": 2.15, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0605, "step": 1975 }, { "epoch": 2.15, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0512, "step": 1976 }, { "epoch": 2.15, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0718, "step": 1977 }, { "epoch": 2.15, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.068, "step": 1978 }, { "epoch": 2.15, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0576, "step": 1979 }, { "epoch": 2.15, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0652, "step": 1980 }, { "epoch": 2.16, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0789, "step": 1981 }, { "epoch": 2.16, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0694, "step": 1982 }, { "epoch": 2.16, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0474, "step": 1983 }, { "epoch": 2.16, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0317, "step": 1984 }, { "epoch": 2.16, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0702, "step": 1985 }, { "epoch": 2.16, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0375, "step": 1986 }, { "epoch": 2.16, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0297, "step": 1987 }, { "epoch": 2.16, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0173, "step": 1988 }, { "epoch": 2.16, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0629, "step": 1989 }, { "epoch": 2.17, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0679, "step": 1990 }, { "epoch": 2.17, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0324, "step": 1991 }, { "epoch": 2.17, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0704, "step": 1992 }, { "epoch": 2.17, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0283, "step": 1993 }, { "epoch": 2.17, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0651, "step": 1994 }, { "epoch": 2.17, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0659, "step": 1995 }, { "epoch": 2.17, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.1034, "step": 1996 }, { "epoch": 2.17, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0223, "step": 1997 }, { "epoch": 2.17, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0217, "step": 1998 }, { "epoch": 2.18, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0691, "step": 1999 }, { "epoch": 2.18, - "learning_rate": 0.0005, - "loss": 0.0, + "learning_rate": 0.0002, + "loss": 1.0562, "step": 2000 } ],