| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 700, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.523809523809525e-07, | |
| "loss": 0.9682, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.904761904761905e-06, | |
| "loss": 1.0391, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.8571428571428573e-06, | |
| "loss": 1.0184, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.80952380952381e-06, | |
| "loss": 0.9074, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.761904761904762e-06, | |
| "loss": 0.7204, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5.7142857142857145e-06, | |
| "loss": 0.3424, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.2966, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.61904761904762e-06, | |
| "loss": 0.2079, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.571428571428571e-06, | |
| "loss": 0.1443, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.523809523809525e-06, | |
| "loss": 0.1984, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.0476190476190477e-05, | |
| "loss": 0.2289, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.1428571428571429e-05, | |
| "loss": 0.1791, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.2380952380952383e-05, | |
| "loss": 0.1422, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.1695, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.4285714285714287e-05, | |
| "loss": 0.1504, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.523809523809524e-05, | |
| "loss": 0.1498, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.6190476190476193e-05, | |
| "loss": 0.1533, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.7142857142857142e-05, | |
| "loss": 0.1885, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.8095238095238097e-05, | |
| "loss": 0.1423, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.904761904761905e-05, | |
| "loss": 0.1335, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2e-05, | |
| "loss": 0.1319, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9999892964109543e-05, | |
| "loss": 0.1405, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.999957185872951e-05, | |
| "loss": 0.1292, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9999036690733865e-05, | |
| "loss": 0.1346, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9998287471579036e-05, | |
| "loss": 0.1221, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.99973242173037e-05, | |
| "loss": 0.1244, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.99961469485284e-05, | |
| "loss": 0.1248, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9994755690455154e-05, | |
| "loss": 0.1207, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9993150472866856e-05, | |
| "loss": 0.1265, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9991331330126698e-05, | |
| "loss": 0.1241, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9989298301177387e-05, | |
| "loss": 0.123, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9987051429540335e-05, | |
| "loss": 0.12, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9984590763314722e-05, | |
| "loss": 0.1177, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9981916355176473e-05, | |
| "loss": 0.12, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.997902826237712e-05, | |
| "loss": 0.1248, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9975926546742573e-05, | |
| "loss": 0.1171, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9972611274671815e-05, | |
| "loss": 0.1167, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9969082517135463e-05, | |
| "loss": 0.1177, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9965340349674262e-05, | |
| "loss": 0.12, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9961384852397456e-05, | |
| "loss": 0.1212, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.995721610998108e-05, | |
| "loss": 0.1179, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.995283421166614e-05, | |
| "loss": 0.1165, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.994823925125672e-05, | |
| "loss": 0.113, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9943431327117955e-05, | |
| "loss": 0.1194, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9938410542173937e-05, | |
| "loss": 0.114, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9933177003905494e-05, | |
| "loss": 0.119, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9927730824347915e-05, | |
| "loss": 0.1158, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9922072120088537e-05, | |
| "loss": 0.1151, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9916201012264255e-05, | |
| "loss": 0.1182, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9910117626558908e-05, | |
| "loss": 0.1149, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9903822093200625e-05, | |
| "loss": 0.1166, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9897314546959015e-05, | |
| "loss": 0.1158, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.989059512714227e-05, | |
| "loss": 0.1171, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.988366397759421e-05, | |
| "loss": 0.1168, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9876521246691183e-05, | |
| "loss": 0.1139, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9869167087338908e-05, | |
| "loss": 0.1164, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9861601656969185e-05, | |
| "loss": 0.1141, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9853825117536522e-05, | |
| "loss": 0.114, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9845837635514692e-05, | |
| "loss": 0.1151, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9837639381893137e-05, | |
| "loss": 0.1164, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9829230532173337e-05, | |
| "loss": 0.1125, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9820611266365036e-05, | |
| "loss": 0.1128, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9811781768982392e-05, | |
| "loss": 0.1156, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9802742229040027e-05, | |
| "loss": 0.1144, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.979349284004898e-05, | |
| "loss": 0.1161, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.978403380001257e-05, | |
| "loss": 0.1154, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.9774365311422155e-05, | |
| "loss": 0.1111, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.9764487581252787e-05, | |
| "loss": 0.1173, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.9754400820958798e-05, | |
| "loss": 0.1142, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.9744105246469264e-05, | |
| "loss": 0.1145, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.9733601078183377e-05, | |
| "loss": 0.1149, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.9722888540965738e-05, | |
| "loss": 0.1142, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.9711967864141542e-05, | |
| "loss": 0.1135, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.970083928149166e-05, | |
| "loss": 0.1149, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.968950303124765e-05, | |
| "loss": 0.1093, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.967795935608663e-05, | |
| "loss": 0.1142, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.9666208503126115e-05, | |
| "loss": 0.1141, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.9654250723918706e-05, | |
| "loss": 0.1118, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.9642086274446713e-05, | |
| "loss": 0.1121, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.962971541511667e-05, | |
| "loss": 0.1136, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.961713841075377e-05, | |
| "loss": 0.1182, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.9604355530596182e-05, | |
| "loss": 0.1138, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.9591367048289297e-05, | |
| "loss": 0.114, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.957817324187987e-05, | |
| "loss": 0.1149, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.9564774393810068e-05, | |
| "loss": 0.1121, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.9551170790911415e-05, | |
| "loss": 0.1152, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.953736272439866e-05, | |
| "loss": 0.1113, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.9523350489863545e-05, | |
| "loss": 0.1092, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.9509134387268463e-05, | |
| "loss": 0.1131, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.9494714720940065e-05, | |
| "loss": 0.1138, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.9480091799562706e-05, | |
| "loss": 0.1169, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.9465265936171875e-05, | |
| "loss": 0.1173, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.9450237448147463e-05, | |
| "loss": 0.1154, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.9435006657206998e-05, | |
| "loss": 0.1151, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.941957388939873e-05, | |
| "loss": 0.1166, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.9403939475094664e-05, | |
| "loss": 0.1153, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.9388103748983496e-05, | |
| "loss": 0.1157, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.937206705006344e-05, | |
| "loss": 0.1134, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.9355829721634952e-05, | |
| "loss": 0.112, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.9339392111293425e-05, | |
| "loss": 0.1107, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.9322754570921713e-05, | |
| "loss": 0.1134, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.930591745668259e-05, | |
| "loss": 0.108, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.9288881129011177e-05, | |
| "loss": 0.1107, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.9271645952607164e-05, | |
| "loss": 0.1059, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.9254212296427043e-05, | |
| "loss": 0.1062, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.92365805336762e-05, | |
| "loss": 0.1045, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.921875104180092e-05, | |
| "loss": 0.1073, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.9200724202480305e-05, | |
| "loss": 0.1071, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.9182500401618122e-05, | |
| "loss": 0.111, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.9164080029334517e-05, | |
| "loss": 0.1053, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.914546347995768e-05, | |
| "loss": 0.1068, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.9126651152015404e-05, | |
| "loss": 0.1051, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.9107643448226536e-05, | |
| "loss": 0.1014, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.9088440775492386e-05, | |
| "loss": 0.0971, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.9069043544887978e-05, | |
| "loss": 0.1099, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.904945217165329e-05, | |
| "loss": 0.1003, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.9029667075184333e-05, | |
| "loss": 0.1041, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.900968867902419e-05, | |
| "loss": 0.1097, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.8989517410853956e-05, | |
| "loss": 0.1041, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.896915370248355e-05, | |
| "loss": 0.0966, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.8948597989842504e-05, | |
| "loss": 0.1023, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.8927850712970626e-05, | |
| "loss": 0.1037, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.890691231600856e-05, | |
| "loss": 0.1021, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.8885783247188305e-05, | |
| "loss": 0.1063, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.8864463958823598e-05, | |
| "loss": 0.1032, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.8842954907300236e-05, | |
| "loss": 0.0982, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.8821256553066326e-05, | |
| "loss": 0.103, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.8799369360622394e-05, | |
| "loss": 0.0964, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.8777293798511466e-05, | |
| "loss": 0.1024, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.8755030339309037e-05, | |
| "loss": 0.1063, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.8732579459612935e-05, | |
| "loss": 0.0967, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.8709941640033153e-05, | |
| "loss": 0.0943, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.8687117365181514e-05, | |
| "loss": 0.0932, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.8664107123661337e-05, | |
| "loss": 0.0993, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.8640911408056963e-05, | |
| "loss": 0.0995, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.861753071492321e-05, | |
| "loss": 0.0981, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.859396554477473e-05, | |
| "loss": 0.1041, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.8570216402075326e-05, | |
| "loss": 0.0976, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.854628379522712e-05, | |
| "loss": 0.0956, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.8522168236559693e-05, | |
| "loss": 0.0985, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.8497870242319103e-05, | |
| "loss": 0.0995, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.847339033265684e-05, | |
| "loss": 0.1016, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.8448729031618687e-05, | |
| "loss": 0.0901, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.842388686713351e-05, | |
| "loss": 0.0951, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.8398864371001945e-05, | |
| "loss": 0.1018, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.8373662078885028e-05, | |
| "loss": 0.0982, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.8348280530292712e-05, | |
| "loss": 0.0956, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.8322720268572333e-05, | |
| "loss": 0.0992, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.829698184089695e-05, | |
| "loss": 0.086, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.8271065798253688e-05, | |
| "loss": 0.103, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.8244972695431874e-05, | |
| "loss": 0.0894, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.8218703091011214e-05, | |
| "loss": 0.0954, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.8192257547349805e-05, | |
| "loss": 0.1016, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.816563663057211e-05, | |
| "loss": 0.1004, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.8138840910556836e-05, | |
| "loss": 0.102, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.8111870960924733e-05, | |
| "loss": 0.0996, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.8084727359026316e-05, | |
| "loss": 0.092, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.8057410685929505e-05, | |
| "loss": 0.0948, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.8029921526407184e-05, | |
| "loss": 0.0914, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.8002260468924696e-05, | |
| "loss": 0.0905, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.797442810562721e-05, | |
| "loss": 0.0962, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.794642503232709e-05, | |
| "loss": 0.1, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.7918251848491118e-05, | |
| "loss": 0.0966, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.7889909157227652e-05, | |
| "loss": 0.1106, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.7861397565273736e-05, | |
| "loss": 0.1, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.7832717682982092e-05, | |
| "loss": 0.0976, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.780387012430807e-05, | |
| "loss": 0.0976, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.7774855506796497e-05, | |
| "loss": 0.0995, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.7745674451568454e-05, | |
| "loss": 0.1011, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.7716327583307988e-05, | |
| "loss": 0.1059, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.768681553024874e-05, | |
| "loss": 0.0956, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.765713892416048e-05, | |
| "loss": 0.0935, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.76272984003356e-05, | |
| "loss": 0.1022, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.759729459757551e-05, | |
| "loss": 0.0948, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.7567128158176955e-05, | |
| "loss": 0.0965, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.753679972791828e-05, | |
| "loss": 0.0905, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.750630995604559e-05, | |
| "loss": 0.0964, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.7475659495258864e-05, | |
| "loss": 0.0999, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.7444849001697976e-05, | |
| "loss": 0.091, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.7413879134928644e-05, | |
| "loss": 0.097, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.7382750557928323e-05, | |
| "loss": 0.0955, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.7351463937072008e-05, | |
| "loss": 0.099, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.7320019942117954e-05, | |
| "loss": 0.0943, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.7288419246193373e-05, | |
| "loss": 0.0892, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.7256662525779984e-05, | |
| "loss": 0.0981, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.7224750460699558e-05, | |
| "loss": 0.0912, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.7192683734099355e-05, | |
| "loss": 0.0951, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.71604630324375e-05, | |
| "loss": 0.0908, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.7128089045468294e-05, | |
| "loss": 0.0934, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.709556246622744e-05, | |
| "loss": 0.0954, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.7062883991017217e-05, | |
| "loss": 0.0995, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.7030054319391563e-05, | |
| "loss": 0.1032, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.6997074154141097e-05, | |
| "loss": 0.0919, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.6963944201278097e-05, | |
| "loss": 0.0893, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.6930665170021364e-05, | |
| "loss": 0.0967, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.6897237772781046e-05, | |
| "loss": 0.0876, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.6863662725143383e-05, | |
| "loss": 0.0921, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.682994074585541e-05, | |
| "loss": 0.0921, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.6796072556809536e-05, | |
| "loss": 0.0877, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.676205888302812e-05, | |
| "loss": 0.093, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.6727900452647925e-05, | |
| "loss": 0.0902, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.6693597996904566e-05, | |
| "loss": 0.091, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.665915225011681e-05, | |
| "loss": 0.0903, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.6624563949670903e-05, | |
| "loss": 0.0935, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.6589833836004753e-05, | |
| "loss": 0.0932, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.6554962652592078e-05, | |
| "loss": 0.0875, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.6519951145926515e-05, | |
| "loss": 0.0871, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.6484800065505627e-05, | |
| "loss": 0.1, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.6449510163814854e-05, | |
| "loss": 0.0986, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.6414082196311402e-05, | |
| "loss": 0.0872, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.6378516921408077e-05, | |
| "loss": 0.0942, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.6342815100457064e-05, | |
| "loss": 0.0948, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.630697749773359e-05, | |
| "loss": 0.0916, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.6271004880419606e-05, | |
| "loss": 0.095, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.6234898018587336e-05, | |
| "loss": 0.0866, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.61986576851828e-05, | |
| "loss": 0.0876, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.6162284656009276e-05, | |
| "loss": 0.0868, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.6125779709710668e-05, | |
| "loss": 0.0978, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.6089143627754863e-05, | |
| "loss": 0.0911, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.6052377194417e-05, | |
| "loss": 0.0874, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.6015481196762658e-05, | |
| "loss": 0.0825, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.5978456424631033e-05, | |
| "loss": 0.0928, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.5941303670618018e-05, | |
| "loss": 0.0883, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.5904023730059227e-05, | |
| "loss": 0.0889, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.5866617401012995e-05, | |
| "loss": 0.0997, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.5829085484243264e-05, | |
| "loss": 0.0845, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.5791428783202465e-05, | |
| "loss": 0.0892, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.57536481040143e-05, | |
| "loss": 0.0834, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.5715744255456497e-05, | |
| "loss": 0.0832, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5677718048943495e-05, | |
| "loss": 0.0956, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5639570298509067e-05, | |
| "loss": 0.0949, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.56013018207889e-05, | |
| "loss": 0.0834, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.5562913435003113e-05, | |
| "loss": 0.0838, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.5524405962938717e-05, | |
| "loss": 0.0905, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.548578022893202e-05, | |
| "loss": 0.0887, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.5447037059851e-05, | |
| "loss": 0.0868, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.5408177285077564e-05, | |
| "loss": 0.0899, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.536920173648984e-05, | |
| "loss": 0.0889, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.5330111248444332e-05, | |
| "loss": 0.0892, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5290906657758085e-05, | |
| "loss": 0.1011, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5251588803690753e-05, | |
| "loss": 0.0958, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5212158527926635e-05, | |
| "loss": 0.0986, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5172616674556673e-05, | |
| "loss": 0.0854, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.5132964090060358e-05, | |
| "loss": 0.0885, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.5093201623287631e-05, | |
| "loss": 0.0933, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.5053330125440701e-05, | |
| "loss": 0.1046, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.5013350450055821e-05, | |
| "loss": 0.0927, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4973263452985023e-05, | |
| "loss": 0.0953, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4933069992377793e-05, | |
| "loss": 0.0979, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4892770928662699e-05, | |
| "loss": 0.0978, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.4852367124528975e-05, | |
| "loss": 0.0834, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.4811859444908053e-05, | |
| "loss": 0.1048, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.4771248756955042e-05, | |
| "loss": 0.0953, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.4730535930030172e-05, | |
| "loss": 0.0983, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.4689721835680183e-05, | |
| "loss": 0.0924, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.4648807347619662e-05, | |
| "loss": 0.0863, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.4607793341712336e-05, | |
| "loss": 0.0904, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.4566680695952333e-05, | |
| "loss": 0.0728, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.4525470290445392e-05, | |
| "loss": 0.0875, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.4484163007389991e-05, | |
| "loss": 0.0862, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.4442759731058502e-05, | |
| "loss": 0.0798, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.4401261347778235e-05, | |
| "loss": 0.0911, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.4359668745912472e-05, | |
| "loss": 0.0809, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.4317982815841442e-05, | |
| "loss": 0.0939, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.427620444994328e-05, | |
| "loss": 0.0809, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.4234334542574906e-05, | |
| "loss": 0.0831, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.4192373990052877e-05, | |
| "loss": 0.0867, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.415032369063422e-05, | |
| "loss": 0.0837, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.410818454449718e-05, | |
| "loss": 0.0868, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.4065957453721963e-05, | |
| "loss": 0.0901, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.4023643322271425e-05, | |
| "loss": 0.0928, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.3981243055971712e-05, | |
| "loss": 0.0876, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.3938757562492873e-05, | |
| "loss": 0.0864, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.3896187751329438e-05, | |
| "loss": 0.0827, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.3853534533780932e-05, | |
| "loss": 0.0784, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.381079882293238e-05, | |
| "loss": 0.0924, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.3767981533634755e-05, | |
| "loss": 0.094, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.3725083582485397e-05, | |
| "loss": 0.0956, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.368210588780838e-05, | |
| "loss": 0.0956, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.3639049369634878e-05, | |
| "loss": 0.0846, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.3595914949683433e-05, | |
| "loss": 0.0931, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.3552703551340259e-05, | |
| "loss": 0.0972, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.3509416099639456e-05, | |
| "loss": 0.0935, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.3466053521243214e-05, | |
| "loss": 0.0857, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.3422616744421966e-05, | |
| "loss": 0.0928, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.3379106699034536e-05, | |
| "loss": 0.09, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.3335524316508208e-05, | |
| "loss": 0.0892, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.3291870529818809e-05, | |
| "loss": 0.088, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.3248146273470725e-05, | |
| "loss": 0.0879, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.3204352483476898e-05, | |
| "loss": 0.0826, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.316049009733879e-05, | |
| "loss": 0.0882, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.311656005402631e-05, | |
| "loss": 0.0829, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.3072563293957725e-05, | |
| "loss": 0.0875, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.3028500758979507e-05, | |
| "loss": 0.0808, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.2984373392346195e-05, | |
| "loss": 0.0892, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.294018213870018e-05, | |
| "loss": 0.0827, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.2895927944051503e-05, | |
| "loss": 0.0873, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.2851611755757587e-05, | |
| "loss": 0.0991, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.280723452250297e-05, | |
| "loss": 0.0926, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.2762797194278984e-05, | |
| "loss": 0.0939, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.2718300722363431e-05, | |
| "loss": 0.0865, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.2673746059300209e-05, | |
| "loss": 0.0921, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.2629134158878919e-05, | |
| "loss": 0.0896, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.258446597611447e-05, | |
| "loss": 0.0846, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.25397424672266e-05, | |
| "loss": 0.0918, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.2494964589619424e-05, | |
| "loss": 0.0902, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.2450133301860951e-05, | |
| "loss": 0.0806, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.2405249563662539e-05, | |
| "loss": 0.0831, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.236031433585836e-05, | |
| "loss": 0.0902, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.2315328580384841e-05, | |
| "loss": 0.0878, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.2270293260260066e-05, | |
| "loss": 0.0897, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.2225209339563144e-05, | |
| "loss": 0.0905, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.2180077783413601e-05, | |
| "loss": 0.0864, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.21348995579507e-05, | |
| "loss": 0.0875, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.2089675630312755e-05, | |
| "loss": 0.0806, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.2044406968616433e-05, | |
| "loss": 0.0833, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.1999094541936047e-05, | |
| "loss": 0.0848, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.1953739320282778e-05, | |
| "loss": 0.0838, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.1908342274583937e-05, | |
| "loss": 0.0935, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.1862904376662167e-05, | |
| "loss": 0.0831, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.1817426599214637e-05, | |
| "loss": 0.0856, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.177190991579223e-05, | |
| "loss": 0.0855, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.1726355300778693e-05, | |
| "loss": 0.0809, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.1680763729369783e-05, | |
| "loss": 0.0893, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.163513617755239e-05, | |
| "loss": 0.0894, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.1589473622083642e-05, | |
| "loss": 0.0931, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.1543777040469994e-05, | |
| "loss": 0.0857, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.1498047410946307e-05, | |
| "loss": 0.0854, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.1452285712454905e-05, | |
| "loss": 0.0848, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.1406492924624614e-05, | |
| "loss": 0.0821, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.1360670027749806e-05, | |
| "loss": 0.0811, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.131481800276939e-05, | |
| "loss": 0.0857, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.126893783124583e-05, | |
| "loss": 0.0806, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.1223030495344127e-05, | |
| "loss": 0.0838, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.1177096977810803e-05, | |
| "loss": 0.0873, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.1131138261952845e-05, | |
| "loss": 0.0899, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.1085155331616664e-05, | |
| "loss": 0.0801, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.1039149171167046e-05, | |
| "loss": 0.0884, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.0993120765466056e-05, | |
| "loss": 0.0895, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.0947071099851972e-05, | |
| "loss": 0.0878, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.090100116011819e-05, | |
| "loss": 0.0887, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.0854911932492114e-05, | |
| "loss": 0.081, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.0808804403614044e-05, | |
| "loss": 0.0843, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.0762679560516067e-05, | |
| "loss": 0.0887, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0716538390600908e-05, | |
| "loss": 0.0902, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0670381881620814e-05, | |
| "loss": 0.0718, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0624211021656392e-05, | |
| "loss": 0.082, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0578026799095464e-05, | |
| "loss": 0.0926, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.0531830202611904e-05, | |
| "loss": 0.0852, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.0485622221144485e-05, | |
| "loss": 0.0843, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.043940384387569e-05, | |
| "loss": 0.0782, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.0393176060210556e-05, | |
| "loss": 0.081, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.0346939859755481e-05, | |
| "loss": 0.085, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.0300696232297041e-05, | |
| "loss": 0.0797, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.0254446167780803e-05, | |
| "loss": 0.0838, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.0208190656290138e-05, | |
| "loss": 0.0845, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.0161930688025018e-05, | |
| "loss": 0.0694, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.0115667253280817e-05, | |
| "loss": 0.0822, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.006940134242713e-05, | |
| "loss": 0.0697, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.0023133945886545e-05, | |
| "loss": 0.0724, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.976866054113458e-06, | |
| "loss": 0.0765, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.930598657572874e-06, | |
| "loss": 0.0775, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.884332746719186e-06, | |
| "loss": 0.079, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.838069311974986e-06, | |
| "loss": 0.0897, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.791809343709866e-06, | |
| "loss": 0.0909, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.745553832219202e-06, | |
| "loss": 0.089, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.699303767702964e-06, | |
| "loss": 0.0811, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.653060140244524e-06, | |
| "loss": 0.0792, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.606823939789449e-06, | |
| "loss": 0.0775, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.560596156124314e-06, | |
| "loss": 0.0805, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.514377778855521e-06, | |
| "loss": 0.0765, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.468169797388098e-06, | |
| "loss": 0.0835, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.421973200904538e-06, | |
| "loss": 0.0889, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.375788978343608e-06, | |
| "loss": 0.0782, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.329618118379187e-06, | |
| "loss": 0.084, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.283461609399094e-06, | |
| "loss": 0.0843, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.237320439483936e-06, | |
| "loss": 0.0841, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.19119559638596e-06, | |
| "loss": 0.0868, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.14508806750789e-06, | |
| "loss": 0.0808, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.098998839881812e-06, | |
| "loss": 0.088, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.052928900148031e-06, | |
| "loss": 0.0765, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.006879234533948e-06, | |
| "loss": 0.0744, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.960850828832958e-06, | |
| "loss": 0.0791, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.91484466838334e-06, | |
| "loss": 0.0766, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.868861738047158e-06, | |
| "loss": 0.0861, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.8229030221892e-06, | |
| "loss": 0.0679, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.776969504655875e-06, | |
| "loss": 0.0827, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.731062168754174e-06, | |
| "loss": 0.0824, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.685181997230616e-06, | |
| "loss": 0.0769, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.639329972250199e-06, | |
| "loss": 0.0799, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.593507075375388e-06, | |
| "loss": 0.0752, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.5477142875451e-06, | |
| "loss": 0.0861, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.501952589053694e-06, | |
| "loss": 0.077, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.456222959530008e-06, | |
| "loss": 0.08, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.41052637791636e-06, | |
| "loss": 0.0769, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.364863822447612e-06, | |
| "loss": 0.0784, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.319236270630218e-06, | |
| "loss": 0.0769, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 8.273644699221309e-06, | |
| "loss": 0.0784, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 8.228090084207773e-06, | |
| "loss": 0.0804, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 8.182573400785367e-06, | |
| "loss": 0.0788, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.137095623337836e-06, | |
| "loss": 0.0736, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.091657725416065e-06, | |
| "loss": 0.0788, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.046260679717225e-06, | |
| "loss": 0.0816, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.000905458063956e-06, | |
| "loss": 0.0802, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.95559303138357e-06, | |
| "loss": 0.0728, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.91032436968725e-06, | |
| "loss": 0.0861, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.865100442049302e-06, | |
| "loss": 0.0685, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 7.8199222165864e-06, | |
| "loss": 0.086, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 7.774790660436857e-06, | |
| "loss": 0.0797, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 7.729706739739939e-06, | |
| "loss": 0.0827, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 7.684671419615164e-06, | |
| "loss": 0.0799, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 7.639685664141644e-06, | |
| "loss": 0.0799, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 7.594750436337467e-06, | |
| "loss": 0.0783, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 7.549866698139048e-06, | |
| "loss": 0.0831, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.505035410380574e-06, | |
| "loss": 0.0754, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.4602575327734004e-06, | |
| "loss": 0.0849, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.415534023885529e-06, | |
| "loss": 0.091, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.37086584112108e-06, | |
| "loss": 0.0751, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 7.326253940699795e-06, | |
| "loss": 0.0795, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 7.2816992776365714e-06, | |
| "loss": 0.0743, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 7.23720280572102e-06, | |
| "loss": 0.0835, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 7.192765477497032e-06, | |
| "loss": 0.0752, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 7.148388244242414e-06, | |
| "loss": 0.0845, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 7.1040720559485e-06, | |
| "loss": 0.0711, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 7.05981786129982e-06, | |
| "loss": 0.0823, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 7.015626607653807e-06, | |
| "loss": 0.0814, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.971499241020495e-06, | |
| "loss": 0.0785, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.927436706042276e-06, | |
| "loss": 0.0684, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.883439945973692e-06, | |
| "loss": 0.0772, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.839509902661215e-06, | |
| "loss": 0.0811, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.795647516523107e-06, | |
| "loss": 0.0674, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.75185372652928e-06, | |
| "loss": 0.0766, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.708129470181197e-06, | |
| "loss": 0.0845, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.664475683491797e-06, | |
| "loss": 0.0816, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.620893300965465e-06, | |
| "loss": 0.0761, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.577383255578034e-06, | |
| "loss": 0.0774, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.533946478756789e-06, | |
| "loss": 0.0788, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.490583900360543e-06, | |
| "loss": 0.0819, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.447296448659744e-06, | |
| "loss": 0.0784, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.404085050316568e-06, | |
| "loss": 0.0796, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.360950630365126e-06, | |
| "loss": 0.0808, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.317894112191621e-06, | |
| "loss": 0.081, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 6.274916417514605e-06, | |
| "loss": 0.0755, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 6.232018466365247e-06, | |
| "loss": 0.0792, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 6.189201177067623e-06, | |
| "loss": 0.0845, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 6.146465466219072e-06, | |
| "loss": 0.0771, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.103812248670567e-06, | |
| "loss": 0.0778, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.061242437507131e-06, | |
| "loss": 0.0804, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.018756944028291e-06, | |
| "loss": 0.0804, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.976356677728578e-06, | |
| "loss": 0.0704, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.9340425462780405e-06, | |
| "loss": 0.0887, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.891815455502824e-06, | |
| "loss": 0.0749, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.849676309365786e-06, | |
| "loss": 0.0838, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 5.807626009947127e-06, | |
| "loss": 0.0795, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 5.765665457425102e-06, | |
| "loss": 0.088, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 5.723795550056721e-06, | |
| "loss": 0.082, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.682017184158562e-06, | |
| "loss": 0.0831, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.6403312540875325e-06, | |
| "loss": 0.0752, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.5987386522217645e-06, | |
| "loss": 0.0751, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.557240268941497e-06, | |
| "loss": 0.0851, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.515836992610011e-06, | |
| "loss": 0.0833, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.4745297095546125e-06, | |
| "loss": 0.0795, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.433319304047666e-06, | |
| "loss": 0.082, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.3922066582876685e-06, | |
| "loss": 0.0841, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.351192652380341e-06, | |
| "loss": 0.0721, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.310278164319816e-06, | |
| "loss": 0.074, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.26946406996983e-06, | |
| "loss": 0.0753, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.228751243044961e-06, | |
| "loss": 0.085, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.18814055509195e-06, | |
| "loss": 0.0705, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.1476328754710295e-06, | |
| "loss": 0.0852, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 5.1072290713373056e-06, | |
| "loss": 0.0762, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 5.066930007622209e-06, | |
| "loss": 0.0824, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 5.026736547014981e-06, | |
| "loss": 0.0865, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.9866495499441824e-06, | |
| "loss": 0.0691, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.946669874559301e-06, | |
| "loss": 0.0807, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.9067983767123736e-06, | |
| "loss": 0.0795, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.8670359099396466e-06, | |
| "loss": 0.0764, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.827383325443331e-06, | |
| "loss": 0.0844, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.787841472073366e-06, | |
| "loss": 0.08, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.748411196309248e-06, | |
| "loss": 0.0783, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.709093342241917e-06, | |
| "loss": 0.0733, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.669888751555669e-06, | |
| "loss": 0.0679, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.630798263510162e-06, | |
| "loss": 0.0769, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.591822714922441e-06, | |
| "loss": 0.0744, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.552962940149005e-06, | |
| "loss": 0.0823, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.51421977106798e-06, | |
| "loss": 0.0767, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.475594037061285e-06, | |
| "loss": 0.0781, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.437086564996891e-06, | |
| "loss": 0.0767, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.398698179211103e-06, | |
| "loss": 0.0842, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.360429701490935e-06, | |
| "loss": 0.0799, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.3222819510565085e-06, | |
| "loss": 0.0788, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.284255744543505e-06, | |
| "loss": 0.076, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.246351895985702e-06, | |
| "loss": 0.0823, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.208571216797539e-06, | |
| "loss": 0.0784, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.170914515756738e-06, | |
| "loss": 0.0705, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.133382598987007e-06, | |
| "loss": 0.0838, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.095976269940777e-06, | |
| "loss": 0.0758, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.058696329381987e-06, | |
| "loss": 0.0662, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.021543575368968e-06, | |
| "loss": 0.0799, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 3.984518803237343e-06, | |
| "loss": 0.0826, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 3.9476228055830015e-06, | |
| "loss": 0.0692, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 3.910856372245139e-06, | |
| "loss": 0.0684, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 3.874220290289337e-06, | |
| "loss": 0.0812, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 3.837715343990727e-06, | |
| "loss": 0.0719, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 3.801342314817201e-06, | |
| "loss": 0.0806, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 3.7651019814126656e-06, | |
| "loss": 0.0824, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 3.7289951195803954e-06, | |
| "loss": 0.0762, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 3.6930225022664136e-06, | |
| "loss": 0.0794, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 3.6571848995429404e-06, | |
| "loss": 0.0755, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 3.621483078591923e-06, | |
| "loss": 0.0853, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 3.585917803688603e-06, | |
| "loss": 0.0807, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 3.5504898361851492e-06, | |
| "loss": 0.0778, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 3.515199934494373e-06, | |
| "loss": 0.0788, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 3.4800488540734857e-06, | |
| "loss": 0.0736, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 3.4450373474079278e-06, | |
| "loss": 0.0733, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 3.4101661639952533e-06, | |
| "loss": 0.0798, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 3.3754360503290973e-06, | |
| "loss": 0.0725, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 3.3408477498831917e-06, | |
| "loss": 0.0804, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 3.306402003095438e-06, | |
| "loss": 0.0854, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 3.2720995473520766e-06, | |
| "loss": 0.0868, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 3.2379411169718844e-06, | |
| "loss": 0.0809, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 3.2039274431904643e-06, | |
| "loss": 0.0712, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 3.170059254144593e-06, | |
| "loss": 0.0645, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 3.136337274856618e-06, | |
| "loss": 0.0734, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 3.1027622272189572e-06, | |
| "loss": 0.0852, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 3.0693348299786407e-06, | |
| "loss": 0.0855, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 3.0360557987219054e-06, | |
| "loss": 0.0869, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 3.002925845858905e-06, | |
| "loss": 0.0795, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.9699456806084427e-06, | |
| "loss": 0.0722, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.9371160089827842e-06, | |
| "loss": 0.0776, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.9044375337725595e-06, | |
| "loss": 0.0808, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.8719109545317102e-06, | |
| "loss": 0.0732, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.839536967562504e-06, | |
| "loss": 0.0745, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.807316265900649e-06, | |
| "loss": 0.0833, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.7752495393004476e-06, | |
| "loss": 0.0783, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.743337474220019e-06, | |
| "loss": 0.0799, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.71158075380663e-06, | |
| "loss": 0.0751, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.679980057882049e-06, | |
| "loss": 0.0752, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.648536062927999e-06, | |
| "loss": 0.0811, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.61724944207168e-06, | |
| "loss": 0.0718, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.5861208650713586e-06, | |
| "loss": 0.073, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.555150998302026e-06, | |
| "loss": 0.0654, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.5243405047411353e-06, | |
| "loss": 0.0806, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.493690043954411e-06, | |
| "loss": 0.0721, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.4632002720817227e-06, | |
| "loss": 0.081, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.432871841823047e-06, | |
| "loss": 0.0814, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.4027054024244954e-06, | |
| "loss": 0.0707, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.3727015996644043e-06, | |
| "loss": 0.0731, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.342861075839523e-06, | |
| "loss": 0.0751, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.313184469751265e-06, | |
| "loss": 0.0907, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.2836724166920134e-06, | |
| "loss": 0.0785, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.2543255484315486e-06, | |
| "loss": 0.0796, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.2251444932035094e-06, | |
| "loss": 0.0846, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.1961298756919337e-06, | |
| "loss": 0.0764, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.16728231701791e-06, | |
| "loss": 0.0739, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.1386024347262692e-06, | |
| "loss": 0.075, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.11009084277235e-06, | |
| "loss": 0.0787, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.081748151508883e-06, | |
| "loss": 0.0712, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.0535749676729123e-06, | |
| "loss": 0.0774, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.025571894372794e-06, | |
| "loss": 0.0796, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.9977395310753087e-06, | |
| "loss": 0.0738, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.970078473592817e-06, | |
| "loss": 0.0846, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.942589314070494e-06, | |
| "loss": 0.0718, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.9152726409736834e-06, | |
| "loss": 0.0766, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.8881290390752682e-06, | |
| "loss": 0.0784, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.8611590894431653e-06, | |
| "loss": 0.0871, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.8343633694278895e-06, | |
| "loss": 0.0712, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.8077424526501964e-06, | |
| "loss": 0.0808, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.7812969089887867e-06, | |
| "loss": 0.0767, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.7550273045681243e-06, | |
| "loss": 0.0823, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.728934201746314e-06, | |
| "loss": 0.0741, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.7030181591030493e-06, | |
| "loss": 0.073, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.6772797314276712e-06, | |
| "loss": 0.0753, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.6517194697072903e-06, | |
| "loss": 0.0809, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.6263379211149733e-06, | |
| "loss": 0.0869, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.6011356289980562e-06, | |
| "loss": 0.0772, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.5761131328664946e-06, | |
| "loss": 0.0707, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.5512709683813165e-06, | |
| "loss": 0.0781, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.526609667343163e-06, | |
| "loss": 0.0761, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.5021297576809002e-06, | |
| "loss": 0.07, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.4778317634403082e-06, | |
| "loss": 0.0773, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.4537162047728804e-06, | |
| "loss": 0.0696, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.4297835979246777e-06, | |
| "loss": 0.0804, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.4060344552252702e-06, | |
| "loss": 0.0639, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.3824692850767928e-06, | |
| "loss": 0.0724, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.3590885919430385e-06, | |
| "loss": 0.0745, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.3358928763386647e-06, | |
| "loss": 0.0746, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.3128826348184886e-06, | |
| "loss": 0.0813, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.2900583599668515e-06, | |
| "loss": 0.0656, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.2674205403870643e-06, | |
| "loss": 0.0716, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.2449696606909656e-06, | |
| "loss": 0.0691, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.222706201488536e-06, | |
| "loss": 0.0727, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.200630639377609e-06, | |
| "loss": 0.0809, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.1787434469336767e-06, | |
| "loss": 0.0709, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.1570450926997657e-06, | |
| "loss": 0.0728, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.135536041176406e-06, | |
| "loss": 0.0765, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.1142167528116965e-06, | |
| "loss": 0.0714, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.0930876839914418e-06, | |
| "loss": 0.0678, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.072149287029377e-06, | |
| "loss": 0.0807, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.0514020101574974e-06, | |
| "loss": 0.0745, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.0308462975164547e-06, | |
| "loss": 0.0742, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.010482589146048e-06, | |
| "loss": 0.081, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 9.903113209758098e-07, | |
| "loss": 0.0771, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 9.703329248156712e-07, | |
| "loss": 0.0781, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 9.505478283467129e-07, | |
| "loss": 0.0672, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 9.309564551120254e-07, | |
| "loss": 0.0744, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 9.115592245076177e-07, | |
| "loss": 0.0765, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.923565517734633e-07, | |
| "loss": 0.079, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.733488479845997e-07, | |
| "loss": 0.0705, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.545365200423216e-07, | |
| "loss": 0.0766, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 8.359199706654852e-07, | |
| "loss": 0.0625, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 8.174995983818801e-07, | |
| "loss": 0.0653, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 7.992757975196974e-07, | |
| "loss": 0.0719, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 7.812489581990846e-07, | |
| "loss": 0.0766, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 7.634194663238015e-07, | |
| "loss": 0.0826, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 7.457877035729588e-07, | |
| "loss": 0.0797, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 7.283540473928386e-07, | |
| "loss": 0.0673, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 7.11118870988825e-07, | |
| "loss": 0.0831, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 6.940825433174103e-07, | |
| "loss": 0.0725, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 6.772454290782926e-07, | |
| "loss": 0.0767, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 6.606078887065748e-07, | |
| "loss": 0.0766, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 6.441702783650494e-07, | |
| "loss": 0.0652, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 6.279329499365649e-07, | |
| "loss": 0.0707, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 6.118962510165039e-07, | |
| "loss": 0.0751, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 5.960605249053386e-07, | |
| "loss": 0.0747, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 5.804261106012732e-07, | |
| "loss": 0.0688, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 5.649933427930043e-07, | |
| "loss": 0.0811, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 5.497625518525374e-07, | |
| "loss": 0.0755, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 5.347340638281273e-07, | |
| "loss": 0.0761, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 5.199082004372958e-07, | |
| "loss": 0.0843, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 5.052852790599383e-07, | |
| "loss": 0.0751, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.908656127315359e-07, | |
| "loss": 0.0658, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.7664951013645875e-07, | |
| "loss": 0.08, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.6263727560134086e-07, | |
| "loss": 0.0734, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.4882920908858595e-07, | |
| "loss": 0.075, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.352256061899329e-07, | |
| "loss": 0.0844, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.218267581201296e-07, | |
| "loss": 0.075, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.086329517107046e-07, | |
| "loss": 0.0811, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.9564446940382087e-07, | |
| "loss": 0.0645, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.8286158924623287e-07, | |
| "loss": 0.0746, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.702845848833314e-07, | |
| "loss": 0.073, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.579137255532894e-07, | |
| "loss": 0.0732, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.457492760812975e-07, | |
| "loss": 0.0753, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.3379149687388866e-07, | |
| "loss": 0.0854, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.2204064391337277e-07, | |
| "loss": 0.0822, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.1049696875235515e-07, | |
| "loss": 0.0791, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 2.9916071850834004e-07, | |
| "loss": 0.0717, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 2.8803213585846036e-07, | |
| "loss": 0.0735, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 2.7711145903426494e-07, | |
| "loss": 0.0862, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 2.6639892181662744e-07, | |
| "loss": 0.0741, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 2.5589475353073987e-07, | |
| "loss": 0.0751, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 2.4559917904120266e-07, | |
| "loss": 0.0689, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 2.3551241874721353e-07, | |
| "loss": 0.0771, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 2.256346885778482e-07, | |
| "loss": 0.0807, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 2.1596619998743075e-07, | |
| "loss": 0.0752, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 2.0650715995102266e-07, | |
| "loss": 0.0706, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.9725777095997655e-07, | |
| "loss": 0.0718, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.8821823101760949e-07, | |
| "loss": 0.0716, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.7938873363496422e-07, | |
| "loss": 0.0733, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.707694678266636e-07, | |
| "loss": 0.0739, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.6236061810686488e-07, | |
| "loss": 0.0736, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.5416236448531075e-07, | |
| "loss": 0.0745, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.4617488246348012e-07, | |
| "loss": 0.0764, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.383983430308189e-07, | |
| "loss": 0.072, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.30832912661093e-07, | |
| "loss": 0.0693, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.2347875330881886e-07, | |
| "loss": 0.0622, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.1633602240579522e-07, | |
| "loss": 0.0677, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.094048728577346e-07, | |
| "loss": 0.0741, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.026854530409882e-07, | |
| "loss": 0.0775, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 9.617790679937288e-08, | |
| "loss": 0.071, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.988237344109251e-08, | |
| "loss": 0.0736, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.379898773574924e-08, | |
| "loss": 0.0682, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 7.792787991146356e-08, | |
| "loss": 0.073, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 7.226917565208658e-08, | |
| "loss": 0.0671, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 6.682299609450993e-08, | |
| "loss": 0.0696, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 6.158945782606674e-08, | |
| "loss": 0.0723, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 5.6568672882043595e-08, | |
| "loss": 0.0814, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 5.176074874327919e-08, | |
| "loss": 0.069, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.716578833386054e-08, | |
| "loss": 0.0745, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.278389001892369e-08, | |
| "loss": 0.0698, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 3.861514760254648e-08, | |
| "loss": 0.0729, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 3.465965032574015e-08, | |
| "loss": 0.076, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 3.091748286453866e-08, | |
| "loss": 0.0838, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 2.7388725328189036e-08, | |
| "loss": 0.0723, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 2.407345325743049e-08, | |
| "loss": 0.0795, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 2.0971737622883515e-08, | |
| "loss": 0.0699, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.8083644823526645e-08, | |
| "loss": 0.074, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.5409236685277608e-08, | |
| "loss": 0.0708, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.2948570459667687e-08, | |
| "loss": 0.0722, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.070169882261496e-08, | |
| "loss": 0.0751, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 8.668669873304058e-09, | |
| "loss": 0.0698, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.849527133144795e-09, | |
| "loss": 0.0782, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 5.2443095448506674e-09, | |
| "loss": 0.0755, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 3.853051471601754e-09, | |
| "loss": 0.0715, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 2.675782696305307e-09, | |
| "loss": 0.072, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.712528420966253e-09, | |
| "loss": 0.0851, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 9.633092661376353e-10, | |
| "loss": 0.0758, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.2814127048873553e-10, | |
| "loss": 0.0763, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.070358904564639e-10, | |
| "loss": 0.0725, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.0676, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 700, | |
| "total_flos": 35619982540800.0, | |
| "train_loss": 0.0957284898098026, | |
| "train_runtime": 9340.1654, | |
| "train_samples_per_second": 8.39, | |
| "train_steps_per_second": 0.075 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 700, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 188888, | |
| "total_flos": 35619982540800.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |