diff --git "a/checkpoint-919/trainer_state.json" "b/checkpoint-919/trainer_state.json" --- "a/checkpoint-919/trainer_state.json" +++ "b/checkpoint-919/trainer_state.json" @@ -10,5516 +10,5516 @@ "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, + "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 } ],