| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 546, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5.882352941176471e-05, | |
| "loss": 6.4017, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00011764705882352942, | |
| "loss": 6.3327, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00017647058823529413, | |
| "loss": 5.2711, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00023529411764705883, | |
| "loss": 4.6181, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00029411764705882356, | |
| "loss": 4.2619, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00035294117647058826, | |
| "loss": 3.9972, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004117647058823529, | |
| "loss": 3.8753, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00047058823529411766, | |
| "loss": 3.7717, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0005294117647058824, | |
| "loss": 3.6707, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0005882352941176471, | |
| "loss": 3.7333, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0006470588235294118, | |
| "loss": 3.6306, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0007058823529411765, | |
| "loss": 3.6041, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0007647058823529411, | |
| "loss": 3.638, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0008235294117647058, | |
| "loss": 3.6611, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0008823529411764706, | |
| "loss": 3.6303, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0009411764705882353, | |
| "loss": 3.6187, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.001, | |
| "loss": 3.5779, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0009999911828722436, | |
| "loss": 3.5743, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0009999647317999417, | |
| "loss": 3.5574, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0009999206477159838, | |
| "loss": 3.5504, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0009998589321751502, | |
| "loss": 3.5583, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0009997795873540561, | |
| "loss": 3.5215, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000999682616051075, | |
| "loss": 3.5178, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0009995680216862406, | |
| "loss": 3.4778, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0009994358083011254, | |
| "loss": 3.5473, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0009992859805586986, | |
| "loss": 3.4959, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0009991185437431618, | |
| "loss": 3.4824, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000998933503759762, | |
| "loss": 3.4804, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0009987308671345837, | |
| "loss": 3.4757, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0009985106410143197, | |
| "loss": 3.5102, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0009982728331660163, | |
| "loss": 3.5077, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0009980174519768031, | |
| "loss": 3.4793, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0009977445064535938, | |
| "loss": 3.4845, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0009974540062227708, | |
| "loss": 3.452, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0009971459615298447, | |
| "loss": 3.4258, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0009968203832390935, | |
| "loss": 3.4768, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000996477282833178, | |
| "loss": 3.4286, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0009961166724127392, | |
| "loss": 3.4281, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0009957385646959696, | |
| "loss": 3.4384, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0009953429730181654, | |
| "loss": 3.4185, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000994929911331256, | |
| "loss": 3.333, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0009944993942033117, | |
| "loss": 3.3709, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000994051436818031, | |
| "loss": 3.4166, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0009935860549742039, | |
| "loss": 3.3857, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000993103265085155, | |
| "loss": 3.3686, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0009926030841781648, | |
| "loss": 3.3839, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0009920855298938692, | |
| "loss": 3.3552, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0009915506204856367, | |
| "loss": 3.3498, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0009909983748189264, | |
| "loss": 3.3729, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0009904288123706204, | |
| "loss": 3.3199, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000989841953228338, | |
| "loss": 3.2798, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0009892378180897277, | |
| "loss": 3.3704, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0009886164282617359, | |
| "loss": 3.3314, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0009879778056598567, | |
| "loss": 3.369, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0009873219728073585, | |
| "loss": 3.3055, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0009866489528344896, | |
| "loss": 3.321, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000985958769477662, | |
| "loss": 3.3018, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0009852514470786154, | |
| "loss": 3.2918, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0009845270105835573, | |
| "loss": 3.238, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0009837854855422844, | |
| "loss": 3.259, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0009830268981072805, | |
| "loss": 3.2597, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0009822512750327953, | |
| "loss": 3.2607, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0009814586436738997, | |
| "loss": 3.2, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0009806490319855213, | |
| "loss": 3.2634, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.000979822468521459, | |
| "loss": 3.2436, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0009789789824333757, | |
| "loss": 3.272, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0009781186034697692, | |
| "loss": 3.2526, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0009772413619749249, | |
| "loss": 3.2513, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.000976347288887844, | |
| "loss": 3.1943, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0009754364157411531, | |
| "loss": 3.2495, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0009745087746599917, | |
| "loss": 3.2128, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0009735643983608797, | |
| "loss": 3.2157, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0009726033201505626, | |
| "loss": 3.216, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0009716255739248384, | |
| "loss": 3.2119, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0009706311941673604, | |
| "loss": 3.2183, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0009696202159484219, | |
| "loss": 3.1749, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0009685926749237195, | |
| "loss": 3.1502, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0009675486073330952, | |
| "loss": 3.1698, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0009664880499992584, | |
| "loss": 3.1696, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0009654110403264868, | |
| "loss": 3.1845, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0009643176162993082, | |
| "loss": 3.1319, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0009632078164811597, | |
| "loss": 3.2268, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0009620816800130289, | |
| "loss": 3.2153, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0009609392466120716, | |
| "loss": 3.174, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0009597805565702134, | |
| "loss": 3.1585, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0009586056507527265, | |
| "loss": 3.1292, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0009574145705967898, | |
| "loss": 3.1926, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0009562073581100267, | |
| "loss": 3.1655, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0009549840558690242, | |
| "loss": 3.2035, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0009537447070178311, | |
| "loss": 3.1358, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0009524893552664358, | |
| "loss": 3.115, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0009512180448892254, | |
| "loss": 3.1452, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0009499308207234239, | |
| "loss": 3.1578, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.000948627728167511, | |
| "loss": 3.1772, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0009473088131796211, | |
| "loss": 3.1346, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0009459741222759217, | |
| "loss": 3.1355, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0009446237025289739, | |
| "loss": 3.128, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0009432576015660713, | |
| "loss": 3.1205, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0009418758675675608, | |
| "loss": 3.1561, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.000940478549265143, | |
| "loss": 3.1419, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0009390656959401542, | |
| "loss": 3.1328, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0009376373574218271, | |
| "loss": 3.1146, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0009361935840855348, | |
| "loss": 3.1025, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.000934734426851013, | |
| "loss": 3.1287, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0009332599371805646, | |
| "loss": 3.1505, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0009317701670772449, | |
| "loss": 3.0887, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009302651690830272, | |
| "loss": 3.1247, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009287449962769498, | |
| "loss": 3.0935, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009272097022732444, | |
| "loss": 3.0858, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.000925659341219444, | |
| "loss": 3.1154, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009240939677944747, | |
| "loss": 3.0897, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0009225136372067266, | |
| "loss": 3.1088, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0009209184051921063, | |
| "loss": 3.0495, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0009193083280120717, | |
| "loss": 3.0517, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0009176834624516475, | |
| "loss": 3.0698, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0009160438658174227, | |
| "loss": 3.0736, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0009143895959355293, | |
| "loss": 3.0256, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.000912720711149603, | |
| "loss": 3.0601, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0009110372703187248, | |
| "loss": 3.0582, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0009093393328153468, | |
| "loss": 3.0477, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0009076269585231962, | |
| "loss": 3.0483, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0009059002078351648, | |
| "loss": 3.064, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0009041591416511781, | |
| "loss": 3.0498, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0009024038213760479, | |
| "loss": 3.0868, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.000900634308917307, | |
| "loss": 3.0302, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.000898850666683025, | |
| "loss": 3.0642, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0008970529575796074, | |
| "loss": 3.0781, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0008952412450095778, | |
| "loss": 3.064, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.000893415592869341, | |
| "loss": 3.0727, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0008915760655469294, | |
| "loss": 3.0727, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0008897227279197329, | |
| "loss": 3.0632, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0008878556453522099, | |
| "loss": 3.0792, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0008859748836935826, | |
| "loss": 3.0212, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0008840805092755142, | |
| "loss": 3.0428, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0008821725889097697, | |
| "loss": 3.0668, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0008802511898858597, | |
| "loss": 3.0443, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0008783163799686669, | |
| "loss": 2.9944, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.000876368227396056, | |
| "loss": 3.0289, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0008744068008764676, | |
| "loss": 3.0435, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0008724321695864944, | |
| "loss": 3.1157, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0008704444031684419, | |
| "loss": 3.0375, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0008684435717278723, | |
| "loss": 3.0802, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0008664297458311307, | |
| "loss": 2.997, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0008644029965028583, | |
| "loss": 3.0545, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0008623633952234862, | |
| "loss": 3.0168, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0008603110139267143, | |
| "loss": 3.0375, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0008582459249969752, | |
| "loss": 3.019, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0008561682012668806, | |
| "loss": 2.9904, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0008540779160146527, | |
| "loss": 3.0144, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0008519751429615399, | |
| "loss": 3.0571, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.000849859956269217, | |
| "loss": 3.0058, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0008477324305371692, | |
| "loss": 3.0146, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.000845592640800061, | |
| "loss": 3.0147, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0008434406625250904, | |
| "loss": 3.0215, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0008412765716093271, | |
| "loss": 2.9594, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0008391004443770351, | |
| "loss": 2.9985, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0008369123575769821, | |
| "loss": 3.0027, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0008347123883797312, | |
| "loss": 3.0424, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0008325006143749205, | |
| "loss": 2.9944, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0008302771135685256, | |
| "loss": 2.9764, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0008280419643801096, | |
| "loss": 2.9551, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.000825795245640056, | |
| "loss": 3.0003, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0008235370365867893, | |
| "loss": 2.9775, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0008212674168639809, | |
| "loss": 3.0199, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0008189864665177387, | |
| "loss": 2.9773, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.000816694265993785, | |
| "loss": 2.9877, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0008143908961346196, | |
| "loss": 2.9711, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0008120764381766677, | |
| "loss": 2.9637, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0008097509737474149, | |
| "loss": 2.9912, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0008074145848625296, | |
| "loss": 3.0025, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0008050673539229688, | |
| "loss": 3.0033, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0008027093637120724, | |
| "loss": 3.0264, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0008003406973926446, | |
| "loss": 2.9833, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0007979614385040195, | |
| "loss": 3.003, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0007955716709591152, | |
| "loss": 3.0152, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0007931714790414751, | |
| "loss": 2.9918, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0007907609474022938, | |
| "loss": 2.951, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0007883401610574337, | |
| "loss": 2.9806, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.000785909205384424, | |
| "loss": 2.9911, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.000783468166119452, | |
| "loss": 2.9673, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0007810171293543378, | |
| "loss": 2.9803, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.000778556181533499, | |
| "loss": 2.9586, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0007760854094509002, | |
| "loss": 2.9739, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0007736049002469944, | |
| "loss": 2.9526, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0007711147414056476, | |
| "loss": 2.9644, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0007686150207510545, | |
| "loss": 3.0029, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0007661058264446404, | |
| "loss": 2.9536, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0007635872469819525, | |
| "loss": 3.0044, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0007610593711895379, | |
| "loss": 2.9566, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0007585222882218126, | |
| "loss": 2.9516, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0007559760875579148, | |
| "loss": 3.004, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0007534208589985508, | |
| "loss": 2.976, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0007508566926628269, | |
| "loss": 2.936, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0007482836789850722, | |
| "loss": 3.0205, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0007457019087116478, | |
| "loss": 2.9077, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.000743111472897747, | |
| "loss": 2.9691, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0007405124629041839, | |
| "loss": 2.9072, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.000737904970394171, | |
| "loss": 2.9523, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0007352890873300869, | |
| "loss": 2.9824, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0007326649059702323, | |
| "loss": 2.9105, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0007300325188655761, | |
| "loss": 2.9675, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0007273920188564921, | |
| "loss": 2.9078, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.000724743499069484, | |
| "loss": 3.001, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0007220870529139012, | |
| "loss": 2.9109, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.000719422774078644, | |
| "loss": 2.9444, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0007167507565288599, | |
| "loss": 2.9225, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0007140710945026295, | |
| "loss": 2.9126, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0007113838825076424, | |
| "loss": 2.9031, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0007086892153178645, | |
| "loss": 2.9498, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0007059871879701953, | |
| "loss": 2.9369, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0007032778957611161, | |
| "loss": 2.9506, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0007005614342433291, | |
| "loss": 2.9562, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0006978378992223873, | |
| "loss": 2.8985, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0006951073867533153, | |
| "loss": 2.9216, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0006923699931372225, | |
| "loss": 2.865, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0006896258149179058, | |
| "loss": 2.8559, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0006868749488784451, | |
| "loss": 2.9316, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0006841174920377892, | |
| "loss": 2.9122, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0006813535416473359, | |
| "loss": 2.9551, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0006785831951874993, | |
| "loss": 2.9377, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.000675806550364274, | |
| "loss": 2.9158, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0006730237051057892, | |
| "loss": 2.9042, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.000670234757558853, | |
| "loss": 2.9164, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0006674398060854931, | |
| "loss": 2.9185, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0006646389492594862, | |
| "loss": 2.9421, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.000661832285862882, | |
| "loss": 2.8759, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0006590199148825196, | |
| "loss": 2.9076, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0006562019355065361, | |
| "loss": 2.881, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0006533784471208678, | |
| "loss": 2.9091, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0006505495493057462, | |
| "loss": 2.9327, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.000647715341832185, | |
| "loss": 2.8912, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0006448759246584619, | |
| "loss": 2.9345, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0006420313979265925, | |
| "loss": 2.9289, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0006391818619587996, | |
| "loss": 2.9058, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0006363274172539737, | |
| "loss": 2.8844, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0006334681644841291, | |
| "loss": 2.9056, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0006306042044908539, | |
| "loss": 2.8273, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0006277356382817528, | |
| "loss": 2.9127, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0006248625670268844, | |
| "loss": 2.8822, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0006219850920551943, | |
| "loss": 2.9106, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0006191033148509402, | |
| "loss": 2.936, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0006162173370501129, | |
| "loss": 2.9503, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0006133272604368527, | |
| "loss": 2.8914, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0006104331869398583, | |
| "loss": 2.8883, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0006075352186287927, | |
| "loss": 2.8818, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0006046334577106832, | |
| "loss": 2.9071, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.000601728006526317, | |
| "loss": 2.8626, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0005988189675466309, | |
| "loss": 2.9369, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0005959064433690983, | |
| "loss": 2.8901, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0005929905367141104, | |
| "loss": 2.8558, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0005900713504213531, | |
| "loss": 2.9102, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0005871489874461805, | |
| "loss": 2.8631, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0005842235508559834, | |
| "loss": 2.9386, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0005812951438265541, | |
| "loss": 2.8792, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0005783638696384485, | |
| "loss": 2.8817, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0005754298316733426, | |
| "loss": 2.8791, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0005724931334103867, | |
| "loss": 2.8771, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0005695538784225558, | |
| "loss": 2.8269, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0005666121703729969, | |
| "loss": 2.8714, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0005636681130113729, | |
| "loss": 2.8871, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0005607218101702031, | |
| "loss": 2.909, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0005577733657612018, | |
| "loss": 2.8903, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0005548228837716132, | |
| "loss": 2.8944, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.000551870468260544, | |
| "loss": 2.8879, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0005489162233552929, | |
| "loss": 2.8583, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0005459602532476791, | |
| "loss": 2.8721, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0005430026621903669, | |
| "loss": 2.8898, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0005400435544931892, | |
| "loss": 2.8596, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0005370830345194683, | |
| "loss": 2.8078, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0005341212066823356, | |
| "loss": 2.8633, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0005311581754410488, | |
| "loss": 2.858, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.000528194045297308, | |
| "loss": 2.8688, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0005252289207915697, | |
| "loss": 2.8223, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0005222629064993603, | |
| "loss": 2.798, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0005192961070275876, | |
| "loss": 2.8835, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0005163286270108514, | |
| "loss": 2.8502, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0005133605711077536, | |
| "loss": 2.8509, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0005103920439972062, | |
| "loss": 2.857, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0005074231503747406, | |
| "loss": 2.877, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0005044539949488142, | |
| "loss": 2.8476, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0005014846824371177, | |
| "loss": 2.8581, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0004985153175628823, | |
| "loss": 2.8393, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0004955460050511858, | |
| "loss": 2.8563, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0004925768496252594, | |
| "loss": 2.8638, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.000489607956002794, | |
| "loss": 2.9115, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00048663942889224667, | |
| "loss": 2.8549, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004836713729891487, | |
| "loss": 2.8928, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004807038929724125, | |
| "loss": 2.879, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00047773709350063984, | |
| "loss": 2.8869, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004747710792084304, | |
| "loss": 2.8378, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004718059547026923, | |
| "loss": 2.8266, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00046884182455895134, | |
| "loss": 2.8382, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00046587879331766457, | |
| "loss": 2.8281, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00046291696548053185, | |
| "loss": 2.8413, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00045995644550681094, | |
| "loss": 2.8444, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00045699733780963313, | |
| "loss": 2.8248, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.000454039746752321, | |
| "loss": 2.8827, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0004510837766447072, | |
| "loss": 2.8601, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00044812953173945614, | |
| "loss": 2.8712, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00044517711622838683, | |
| "loss": 2.8435, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0004422266342387982, | |
| "loss": 2.8583, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0004392781898297969, | |
| "loss": 2.8663, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00043633188698862706, | |
| "loss": 2.8726, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0004333878296270032, | |
| "loss": 2.8933, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00043044612157744437, | |
| "loss": 2.8376, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00042750686658961346, | |
| "loss": 2.8484, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0004245701683266575, | |
| "loss": 2.8715, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00042163613036155144, | |
| "loss": 2.868, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0004187048561734459, | |
| "loss": 2.8782, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00041577644914401687, | |
| "loss": 2.8383, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00041285101255381963, | |
| "loss": 2.8603, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00040992864957864694, | |
| "loss": 2.8798, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0004070094632858897, | |
| "loss": 2.8899, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00040409355663090175, | |
| "loss": 2.8318, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00040118103245336917, | |
| "loss": 2.8441, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0003982719934736832, | |
| "loss": 2.8266, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0003953665422893168, | |
| "loss": 2.8586, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0003924647813712074, | |
| "loss": 2.8647, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0003895668130601417, | |
| "loss": 2.8254, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0003866727395631473, | |
| "loss": 2.8828, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.000383782662949887, | |
| "loss": 2.8311, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0003808966851490599, | |
| "loss": 2.8324, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0003780149079448057, | |
| "loss": 2.7876, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00037513743297311553, | |
| "loss": 2.8545, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0003722643617182473, | |
| "loss": 2.8848, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00036939579550914603, | |
| "loss": 2.818, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00036653183551587086, | |
| "loss": 2.8281, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00036367258274602656, | |
| "loss": 2.8624, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0003608181380412005, | |
| "loss": 2.824, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00035796860207340747, | |
| "loss": 2.8443, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0003551240753415382, | |
| "loss": 2.8436, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00035228465816781497, | |
| "loss": 2.8098, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00034945045069425385, | |
| "loss": 2.8407, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0003466215528791323, | |
| "loss": 2.8486, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00034379806449346415, | |
| "loss": 2.85, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00034098008511748045, | |
| "loss": 2.8669, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.0003381677141371181, | |
| "loss": 2.8006, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00033536105074051395, | |
| "loss": 2.7903, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00033256019391450693, | |
| "loss": 2.8569, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.000329765242441147, | |
| "loss": 2.7598, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.0003269762948942109, | |
| "loss": 2.8112, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.000324193449635726, | |
| "loss": 2.8643, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.0003214168048125009, | |
| "loss": 2.7787, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00031864645835266427, | |
| "loss": 2.847, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.0003158825079622106, | |
| "loss": 2.822, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.000313125051121555, | |
| "loss": 2.8276, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.0003103741850820944, | |
| "loss": 2.8283, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.0003076300068627775, | |
| "loss": 2.8423, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00030489261324668485, | |
| "loss": 2.8221, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00030216210077761284, | |
| "loss": 2.8427, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.0002994385657566709, | |
| "loss": 2.8179, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00029672210423888377, | |
| "loss": 2.8383, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0002940128120298048, | |
| "loss": 2.7978, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.00029131078468213543, | |
| "loss": 2.789, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.00028861611749235764, | |
| "loss": 2.8005, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0002859289054973705, | |
| "loss": 2.8807, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0002832492434711401, | |
| "loss": 2.8634, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.000280577225921356, | |
| "loss": 2.7821, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0002779129470860991, | |
| "loss": 2.8128, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.00027525650093051607, | |
| "loss": 2.823, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.00027260798114350787, | |
| "loss": 2.817, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.00026996748113442394, | |
| "loss": 2.8003, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0002673350940297677, | |
| "loss": 2.8937, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.000264710912669913, | |
| "loss": 2.8219, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.00026209502960582897, | |
| "loss": 2.7964, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0002594875370958163, | |
| "loss": 2.7854, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.00025688852710225316, | |
| "loss": 2.786, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.00025429809128835244, | |
| "loss": 2.8407, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0002517163210149279, | |
| "loss": 2.8282, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.00024914330733717324, | |
| "loss": 2.8158, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.00024657914100144953, | |
| "loss": 2.7842, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.00024402391244208527, | |
| "loss": 2.8361, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0002414777117781875, | |
| "loss": 2.8115, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.00023894062881046202, | |
| "loss": 2.8681, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00023641275301804776, | |
| "loss": 2.7963, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00023389417355535963, | |
| "loss": 2.8183, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00023138497924894558, | |
| "loss": 2.8023, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0002288852585943525, | |
| "loss": 2.8147, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00022639509975300566, | |
| "loss": 2.7992, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0002239145905490999, | |
| "loss": 2.8513, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00022144381846650113, | |
| "loss": 2.8424, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00021898287064566214, | |
| "loss": 2.7755, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00021653183388054809, | |
| "loss": 2.8059, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00021409079461557622, | |
| "loss": 2.8222, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.00021165983894256646, | |
| "loss": 2.7634, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.00020923905259770625, | |
| "loss": 2.8553, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.00020682852095852506, | |
| "loss": 2.8081, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0002044283290408847, | |
| "loss": 2.82, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0002020385614959806, | |
| "loss": 2.8751, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0001996593026073555, | |
| "loss": 2.8136, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.0001972906362879277, | |
| "loss": 2.8415, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00019493264607703138, | |
| "loss": 2.8335, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00019258541513747046, | |
| "loss": 2.8777, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.000190249026252585, | |
| "loss": 2.8731, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00018792356182333248, | |
| "loss": 2.8439, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00018560910386538048, | |
| "loss": 2.7752, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00018330573400621492, | |
| "loss": 2.8387, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.0001810135334822614, | |
| "loss": 2.8013, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.0001787325831360191, | |
| "loss": 2.8364, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00017646296341321068, | |
| "loss": 2.8712, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00017420475435994405, | |
| "loss": 2.8237, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.0001719580356198905, | |
| "loss": 2.7693, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00016972288643147445, | |
| "loss": 2.8298, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00016749938562507955, | |
| "loss": 2.8057, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00016528761162026885, | |
| "loss": 2.8495, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00016308764242301788, | |
| "loss": 2.8454, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0001608995556229648, | |
| "loss": 2.8052, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.00015872342839067305, | |
| "loss": 2.7927, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.00015655933747490974, | |
| "loss": 2.7948, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0001544073591999391, | |
| "loss": 2.7802, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.000152267569462831, | |
| "loss": 2.8129, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.00015014004373078293, | |
| "loss": 2.7935, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00014802485703846013, | |
| "loss": 2.8237, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00014592208398534745, | |
| "loss": 2.7628, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00014383179873311947, | |
| "loss": 2.8115, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.0001417540750030249, | |
| "loss": 2.7946, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00013968898607328572, | |
| "loss": 2.8265, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00013763660477651395, | |
| "loss": 2.7865, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00013559700349714165, | |
| "loss": 2.8576, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0001335702541688693, | |
| "loss": 2.8085, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00013155642827212787, | |
| "loss": 2.824, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.000129555596831558, | |
| "loss": 2.814, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00012756783041350568, | |
| "loss": 2.8217, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0001255931991235325, | |
| "loss": 2.7571, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.00012363177260394414, | |
| "loss": 2.8073, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.00012168362003133315, | |
| "loss": 2.8593, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.00011974881011414046, | |
| "loss": 2.7239, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.00011782741109023037, | |
| "loss": 2.7763, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00011591949072448588, | |
| "loss": 2.798, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00011402511630641749, | |
| "loss": 2.8086, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00011214435464779005, | |
| "loss": 2.778, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0001102772720802671, | |
| "loss": 2.7896, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00010842393445307064, | |
| "loss": 2.8086, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00010658440713065915, | |
| "loss": 2.8341, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0001047587549904222, | |
| "loss": 2.7641, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.00010294704242039271, | |
| "loss": 2.7572, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.00010114933331697513, | |
| "loss": 2.8211, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.936569108269306e-05, | |
| "loss": 2.7909, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.759617862395221e-05, | |
| "loss": 2.8325, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.584085834882206e-05, | |
| "loss": 2.8339, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.409979216483539e-05, | |
| "loss": 2.7849, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.237304147680375e-05, | |
| "loss": 2.8141, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.066066718465322e-05, | |
| "loss": 2.7879, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.896272968127506e-05, | |
| "loss": 2.8245, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.727928885039715e-05, | |
| "loss": 2.784, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.561040406447074e-05, | |
| "loss": 2.768, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.39561341825773e-05, | |
| "loss": 2.7894, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.231653754835261e-05, | |
| "loss": 2.7973, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.069167198792837e-05, | |
| "loss": 2.8101, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.908159480789379e-05, | |
| "loss": 2.821, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.748636279327347e-05, | |
| "loss": 2.7684, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.590603220552538e-05, | |
| "loss": 2.8303, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.434065878055613e-05, | |
| "loss": 2.857, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.279029772675571e-05, | |
| "loss": 2.8013, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.125500372305016e-05, | |
| "loss": 2.7653, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.973483091697281e-05, | |
| "loss": 2.8133, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.822983292275525e-05, | |
| "loss": 2.7852, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.674006281943551e-05, | |
| "loss": 2.8088, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.526557314898718e-05, | |
| "loss": 2.8105, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.380641591446523e-05, | |
| "loss": 2.8434, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.236264257817292e-05, | |
| "loss": 2.8382, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 6.093430405984585e-05, | |
| "loss": 2.8044, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.9521450734856985e-05, | |
| "loss": 2.7798, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.812413243243936e-05, | |
| "loss": 2.7843, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.6742398433928753e-05, | |
| "loss": 2.8073, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.5376297471026124e-05, | |
| "loss": 2.7854, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5.4025877724078245e-05, | |
| "loss": 2.7764, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5.269118682037893e-05, | |
| "loss": 2.8102, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5.1372271832488934e-05, | |
| "loss": 2.8031, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5.0069179276576174e-05, | |
| "loss": 2.799, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.8781955110774764e-05, | |
| "loss": 2.7793, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.751064473356431e-05, | |
| "loss": 2.8841, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.625529298216896e-05, | |
| "loss": 2.8155, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.5015944130975706e-05, | |
| "loss": 2.8485, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.379264188997334e-05, | |
| "loss": 2.8263, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.258542940321031e-05, | |
| "loss": 2.8356, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.139434924727359e-05, | |
| "loss": 2.8628, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.0219443429786565e-05, | |
| "loss": 2.7872, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.906075338792836e-05, | |
| "loss": 2.8715, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.7918319986971204e-05, | |
| "loss": 2.8155, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.679218351884017e-05, | |
| "loss": 2.8038, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.5682383700691935e-05, | |
| "loss": 2.8042, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.458895967351328e-05, | |
| "loss": 2.8087, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.3511950000741754e-05, | |
| "loss": 2.7797, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.2451392666904774e-05, | |
| "loss": 2.7939, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.140732507628047e-05, | |
| "loss": 2.7947, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.0379784051578096e-05, | |
| "loss": 2.7489, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.936880583263968e-05, | |
| "loss": 2.7877, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.837442607516161e-05, | |
| "loss": 2.7938, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.7396679849437332e-05, | |
| "loss": 2.8457, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.6435601639120454e-05, | |
| "loss": 2.8401, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.5491225340008308e-05, | |
| "loss": 2.813, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.4563584258846973e-05, | |
| "loss": 2.7653, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.3652711112156112e-05, | |
| "loss": 2.8367, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.2758638025075163e-05, | |
| "loss": 2.7804, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.188139653023091e-05, | |
| "loss": 2.8422, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.1021017566624446e-05, | |
| "loss": 2.7783, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.0177531478540956e-05, | |
| "loss": 2.79, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.9350968014478664e-05, | |
| "loss": 2.8591, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.8541356326100432e-05, | |
| "loss": 2.7813, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.77487249672047e-05, | |
| "loss": 2.8119, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.6973101892719554e-05, | |
| "loss": 2.7676, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.6214514457715768e-05, | |
| "loss": 2.7541, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.5472989416442873e-05, | |
| "loss": 2.7966, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.4748552921384717e-05, | |
| "loss": 2.8278, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.4041230522337911e-05, | |
| "loss": 2.8074, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.3351047165510445e-05, | |
| "loss": 2.7564, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.2678027192641384e-05, | |
| "loss": 2.7928, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.2022194340143288e-05, | |
| "loss": 2.7933, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.1383571738264153e-05, | |
| "loss": 2.7906, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.0762181910272395e-05, | |
| "loss": 2.7661, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.0158046771661878e-05, | |
| "loss": 2.7956, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 9.571187629379585e-06, | |
| "loss": 2.8278, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 9.001625181073503e-06, | |
| "loss": 2.7835, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.449379514363142e-06, | |
| "loss": 2.7671, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 7.914470106130944e-06, | |
| "loss": 2.7933, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 7.396915821835215e-06, | |
| "loss": 2.7539, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.896734914844993e-06, | |
| "loss": 2.8291, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.413945025796064e-06, | |
| "loss": 2.8066, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.948563181968903e-06, | |
| "loss": 2.8095, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.500605796688263e-06, | |
| "loss": 2.8644, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 5.070088668744144e-06, | |
| "loss": 2.7881, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.657026981834622e-06, | |
| "loss": 2.768, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.261435304030281e-06, | |
| "loss": 2.8363, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.883327587260732e-06, | |
| "loss": 2.7849, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.522717166821898e-06, | |
| "loss": 2.8405, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.179616760906612e-06, | |
| "loss": 2.7715, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.85403847015514e-06, | |
| "loss": 2.8489, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.545993777229194e-06, | |
| "loss": 2.7385, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.255493546406251e-06, | |
| "loss": 2.8282, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.9825480231970282e-06, | |
| "loss": 2.7387, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.7271668339836621e-06, | |
| "loss": 2.7942, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.489358985680478e-06, | |
| "loss": 2.8073, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.2691328654161339e-06, | |
| "loss": 2.8458, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.066496240238024e-06, | |
| "loss": 2.7932, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.814562568382201e-07, | |
| "loss": 2.8132, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 7.14019441301339e-07, | |
| "loss": 2.8292, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.641916988746166e-07, | |
| "loss": 2.7654, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.319783137594624e-07, | |
| "loss": 2.7792, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.173839489250541e-07, | |
| "loss": 2.7379, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.2041264594402366e-07, | |
| "loss": 2.8301, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.410678248498498e-07, | |
| "loss": 2.804, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 7.935228401623196e-08, | |
| "loss": 2.7446, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.5268200058391486e-08, | |
| "loss": 2.7603, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.817127756355347e-09, | |
| "loss": 2.8247, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0, | |
| "loss": 2.8084, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 546, | |
| "total_flos": 1.463562078907269e+17, | |
| "train_loss": 2.9841683015718563, | |
| "train_runtime": 1457.6636, | |
| "train_samples_per_second": 382.892, | |
| "train_steps_per_second": 0.375 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 546, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 24000, | |
| "total_flos": 1.463562078907269e+17, | |
| "train_batch_size": 256, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |