| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 2.9978153749059397, |
| "eval_steps": 500, |
| "global_step": 1929, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0, |
| "loss": 1.5191, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 1.2797901240490773e-05, |
| "loss": 1.2224, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 2.0284193554110643e-05, |
| "loss": 1.2649, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 2.5595802480981545e-05, |
| "loss": 0.9563, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 2.9715806445889356e-05, |
| "loss": 1.0907, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.308209479460142e-05, |
| "loss": 0.9246, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.592825103949889e-05, |
| "loss": 0.8697, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 3.839370372147232e-05, |
| "loss": 0.8596, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.056838710822129e-05, |
| "loss": 0.8237, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.251370768638013e-05, |
| "loss": 0.8796, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.427346420355127e-05, |
| "loss": 0.8094, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.587999603509219e-05, |
| "loss": 0.8867, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.735786205915921e-05, |
| "loss": 0.7684, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.8726152279989664e-05, |
| "loss": 0.6962, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 4.9999999999999996e-05, |
| "loss": 0.687, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 5e-05, |
| "loss": 0.6645, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 4.997387669801463e-05, |
| "loss": 0.6157, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 4.994775339602926e-05, |
| "loss": 0.5482, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 4.9921630094043895e-05, |
| "loss": 0.5796, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 4.989550679205852e-05, |
| "loss": 0.5452, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 4.9869383490073145e-05, |
| "loss": 0.549, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 4.9843260188087774e-05, |
| "loss": 0.5562, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 4.98171368861024e-05, |
| "loss": 0.5443, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 4.979101358411704e-05, |
| "loss": 0.5448, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 4.9764890282131666e-05, |
| "loss": 0.5262, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 4.9738766980146295e-05, |
| "loss": 0.5465, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 4.971264367816092e-05, |
| "loss": 0.5164, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 4.968652037617555e-05, |
| "loss": 0.52, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 4.966039707419018e-05, |
| "loss": 0.5399, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 4.963427377220481e-05, |
| "loss": 0.5388, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 4.960815047021944e-05, |
| "loss": 0.5582, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 4.958202716823407e-05, |
| "loss": 0.5355, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 4.9555903866248695e-05, |
| "loss": 0.5296, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 4.9529780564263324e-05, |
| "loss": 0.5222, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 4.950365726227795e-05, |
| "loss": 0.5078, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 4.947753396029258e-05, |
| "loss": 0.5023, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 4.945141065830721e-05, |
| "loss": 0.5064, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 4.9425287356321845e-05, |
| "loss": 0.4979, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 4.9399164054336474e-05, |
| "loss": 0.4796, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 4.93730407523511e-05, |
| "loss": 0.4856, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 4.9346917450365724e-05, |
| "loss": 0.5058, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 4.932079414838036e-05, |
| "loss": 0.4753, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 4.929467084639499e-05, |
| "loss": 0.4997, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 4.9268547544409617e-05, |
| "loss": 0.4789, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 4.9242424242424245e-05, |
| "loss": 0.4778, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 4.9216300940438874e-05, |
| "loss": 0.4708, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 4.91901776384535e-05, |
| "loss": 0.465, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 4.916405433646813e-05, |
| "loss": 0.4773, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 4.913793103448276e-05, |
| "loss": 0.4614, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 4.911180773249739e-05, |
| "loss": 0.456, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 4.908568443051202e-05, |
| "loss": 0.4602, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 4.905956112852665e-05, |
| "loss": 0.4711, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 4.903343782654128e-05, |
| "loss": 0.4643, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 4.900731452455591e-05, |
| "loss": 0.4718, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 4.898119122257053e-05, |
| "loss": 0.4697, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 4.895506792058516e-05, |
| "loss": 0.4609, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 4.8928944618599795e-05, |
| "loss": 0.4705, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 4.8902821316614424e-05, |
| "loss": 0.478, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 4.887669801462905e-05, |
| "loss": 0.4746, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 4.885057471264368e-05, |
| "loss": 0.4294, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 4.882445141065831e-05, |
| "loss": 0.4555, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 4.879832810867294e-05, |
| "loss": 0.4656, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 4.8772204806687567e-05, |
| "loss": 0.4454, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 4.8746081504702195e-05, |
| "loss": 0.4695, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 4.8719958202716824e-05, |
| "loss": 0.4622, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 4.869383490073145e-05, |
| "loss": 0.4703, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 4.866771159874609e-05, |
| "loss": 0.4565, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 4.8641588296760716e-05, |
| "loss": 0.4703, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 4.861546499477534e-05, |
| "loss": 0.4594, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 4.858934169278997e-05, |
| "loss": 0.4579, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 4.85632183908046e-05, |
| "loss": 0.4663, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 4.853709508881923e-05, |
| "loss": 0.4576, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 4.851097178683386e-05, |
| "loss": 0.4435, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 4.848484848484849e-05, |
| "loss": 0.4569, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 4.8458725182863116e-05, |
| "loss": 0.4593, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 4.8432601880877745e-05, |
| "loss": 0.4454, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 4.8406478578892374e-05, |
| "loss": 0.4476, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 4.8380355276907e-05, |
| "loss": 0.4729, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 4.835423197492163e-05, |
| "loss": 0.4521, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 4.832810867293626e-05, |
| "loss": 0.4502, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 4.8301985370950895e-05, |
| "loss": 0.4537, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 4.827586206896552e-05, |
| "loss": 0.4395, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 4.8249738766980145e-05, |
| "loss": 0.4578, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 4.8223615464994774e-05, |
| "loss": 0.461, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 4.81974921630094e-05, |
| "loss": 0.4519, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 4.817136886102404e-05, |
| "loss": 0.464, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 4.8145245559038666e-05, |
| "loss": 0.4487, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 4.8119122257053295e-05, |
| "loss": 0.4802, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 4.8092998955067923e-05, |
| "loss": 0.447, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 4.806687565308255e-05, |
| "loss": 0.4393, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 4.804075235109718e-05, |
| "loss": 0.4615, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 4.801462904911181e-05, |
| "loss": 0.4604, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 4.798850574712644e-05, |
| "loss": 0.4411, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 4.7962382445141066e-05, |
| "loss": 0.4553, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 4.79362591431557e-05, |
| "loss": 0.447, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 4.791013584117033e-05, |
| "loss": 0.4653, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 4.788401253918495e-05, |
| "loss": 0.4453, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 4.785788923719958e-05, |
| "loss": 0.4568, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 4.783176593521421e-05, |
| "loss": 0.446, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 4.7805642633228845e-05, |
| "loss": 0.4412, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 4.777951933124347e-05, |
| "loss": 0.4541, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 4.77533960292581e-05, |
| "loss": 0.4364, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 4.772727272727273e-05, |
| "loss": 0.436, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 4.770114942528736e-05, |
| "loss": 0.4647, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 4.767502612330199e-05, |
| "loss": 0.4443, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 4.7648902821316616e-05, |
| "loss": 0.4612, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 4.7622779519331245e-05, |
| "loss": 0.4428, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 4.7596656217345873e-05, |
| "loss": 0.4495, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 4.75705329153605e-05, |
| "loss": 0.4376, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 4.754440961337514e-05, |
| "loss": 0.438, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 4.751828631138976e-05, |
| "loss": 0.4506, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 4.749216300940439e-05, |
| "loss": 0.4464, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 4.7466039707419016e-05, |
| "loss": 0.414, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 4.743991640543365e-05, |
| "loss": 0.4408, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 4.741379310344828e-05, |
| "loss": 0.4305, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 4.738766980146291e-05, |
| "loss": 0.4558, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 4.736154649947754e-05, |
| "loss": 0.4409, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 4.7335423197492166e-05, |
| "loss": 0.4402, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 4.7309299895506795e-05, |
| "loss": 0.451, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 4.728317659352142e-05, |
| "loss": 0.4384, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 4.725705329153605e-05, |
| "loss": 0.4367, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 4.723092998955068e-05, |
| "loss": 0.4492, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 4.720480668756531e-05, |
| "loss": 0.4347, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 4.7178683385579944e-05, |
| "loss": 0.4371, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 4.715256008359457e-05, |
| "loss": 0.4425, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 4.7126436781609195e-05, |
| "loss": 0.422, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 4.7100313479623823e-05, |
| "loss": 0.4379, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 4.707419017763845e-05, |
| "loss": 0.4439, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 4.704806687565309e-05, |
| "loss": 0.4285, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 4.7021943573667716e-05, |
| "loss": 0.4493, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 4.6995820271682345e-05, |
| "loss": 0.4426, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 4.696969696969697e-05, |
| "loss": 0.4478, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 4.69435736677116e-05, |
| "loss": 0.4388, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 4.691745036572623e-05, |
| "loss": 0.4095, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 4.689132706374086e-05, |
| "loss": 0.4134, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 4.686520376175549e-05, |
| "loss": 0.4499, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 4.6839080459770116e-05, |
| "loss": 0.4541, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 4.6812957157784745e-05, |
| "loss": 0.4325, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 4.678683385579938e-05, |
| "loss": 0.4502, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 4.6760710553814e-05, |
| "loss": 0.4291, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 4.673458725182863e-05, |
| "loss": 0.4195, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 4.670846394984326e-05, |
| "loss": 0.4357, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 4.6682340647857894e-05, |
| "loss": 0.4328, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 4.665621734587252e-05, |
| "loss": 0.4326, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 4.663009404388715e-05, |
| "loss": 0.435, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 4.660397074190178e-05, |
| "loss": 0.4347, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 4.657784743991641e-05, |
| "loss": 0.4297, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 4.655172413793104e-05, |
| "loss": 0.4505, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 4.6525600835945666e-05, |
| "loss": 0.4349, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 4.6499477533960295e-05, |
| "loss": 0.4664, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 4.647335423197492e-05, |
| "loss": 0.4495, |
| "step": 151 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 4.644723092998955e-05, |
| "loss": 0.4354, |
| "step": 152 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 4.642110762800419e-05, |
| "loss": 0.4431, |
| "step": 153 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 4.639498432601881e-05, |
| "loss": 0.4393, |
| "step": 154 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 4.636886102403344e-05, |
| "loss": 0.4441, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 4.6342737722048066e-05, |
| "loss": 0.4384, |
| "step": 156 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 4.63166144200627e-05, |
| "loss": 0.4341, |
| "step": 157 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 4.629049111807733e-05, |
| "loss": 0.4507, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 4.626436781609196e-05, |
| "loss": 0.4146, |
| "step": 159 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 4.623824451410659e-05, |
| "loss": 0.4494, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 4.621212121212121e-05, |
| "loss": 0.4566, |
| "step": 161 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 4.6185997910135844e-05, |
| "loss": 0.4103, |
| "step": 162 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 4.615987460815047e-05, |
| "loss": 0.4348, |
| "step": 163 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 4.61337513061651e-05, |
| "loss": 0.4251, |
| "step": 164 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 4.610762800417973e-05, |
| "loss": 0.4261, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 4.608150470219436e-05, |
| "loss": 0.4307, |
| "step": 166 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 4.6055381400208994e-05, |
| "loss": 0.4274, |
| "step": 167 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 4.6029258098223616e-05, |
| "loss": 0.421, |
| "step": 168 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 4.6003134796238245e-05, |
| "loss": 0.4345, |
| "step": 169 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 4.597701149425287e-05, |
| "loss": 0.424, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 4.59508881922675e-05, |
| "loss": 0.4323, |
| "step": 171 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 4.592476489028214e-05, |
| "loss": 0.4272, |
| "step": 172 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 4.5898641588296766e-05, |
| "loss": 0.4217, |
| "step": 173 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 4.5872518286311394e-05, |
| "loss": 0.445, |
| "step": 174 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 4.5846394984326016e-05, |
| "loss": 0.4087, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 4.582027168234065e-05, |
| "loss": 0.4263, |
| "step": 176 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 4.579414838035528e-05, |
| "loss": 0.4371, |
| "step": 177 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 4.576802507836991e-05, |
| "loss": 0.4436, |
| "step": 178 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 4.574190177638454e-05, |
| "loss": 0.4402, |
| "step": 179 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 4.5715778474399166e-05, |
| "loss": 0.4261, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 4.5689655172413794e-05, |
| "loss": 0.429, |
| "step": 181 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 4.566353187042842e-05, |
| "loss": 0.422, |
| "step": 182 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 4.563740856844305e-05, |
| "loss": 0.4288, |
| "step": 183 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 4.561128526645768e-05, |
| "loss": 0.4418, |
| "step": 184 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 4.558516196447231e-05, |
| "loss": 0.425, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 4.5559038662486944e-05, |
| "loss": 0.4088, |
| "step": 186 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 4.553291536050157e-05, |
| "loss": 0.4446, |
| "step": 187 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 4.55067920585162e-05, |
| "loss": 0.4325, |
| "step": 188 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 4.548066875653082e-05, |
| "loss": 0.4273, |
| "step": 189 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 4.545454545454546e-05, |
| "loss": 0.4245, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 4.542842215256009e-05, |
| "loss": 0.454, |
| "step": 191 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 4.5402298850574716e-05, |
| "loss": 0.4407, |
| "step": 192 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 4.5376175548589344e-05, |
| "loss": 0.4269, |
| "step": 193 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 4.535005224660397e-05, |
| "loss": 0.4087, |
| "step": 194 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 4.53239289446186e-05, |
| "loss": 0.4121, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 4.529780564263323e-05, |
| "loss": 0.4048, |
| "step": 196 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 4.527168234064786e-05, |
| "loss": 0.4186, |
| "step": 197 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 4.524555903866249e-05, |
| "loss": 0.4159, |
| "step": 198 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 4.5219435736677116e-05, |
| "loss": 0.4346, |
| "step": 199 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 4.519331243469175e-05, |
| "loss": 0.4155, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 4.516718913270638e-05, |
| "loss": 0.4247, |
| "step": 201 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 4.514106583072101e-05, |
| "loss": 0.4343, |
| "step": 202 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 4.511494252873563e-05, |
| "loss": 0.4205, |
| "step": 203 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 4.508881922675026e-05, |
| "loss": 0.4506, |
| "step": 204 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 4.5062695924764894e-05, |
| "loss": 0.433, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 4.503657262277952e-05, |
| "loss": 0.4327, |
| "step": 206 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 4.501044932079415e-05, |
| "loss": 0.4186, |
| "step": 207 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 4.498432601880878e-05, |
| "loss": 0.4094, |
| "step": 208 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 4.495820271682341e-05, |
| "loss": 0.4262, |
| "step": 209 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 4.493207941483804e-05, |
| "loss": 0.4091, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 4.4905956112852666e-05, |
| "loss": 0.4163, |
| "step": 211 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 4.4879832810867294e-05, |
| "loss": 0.4103, |
| "step": 212 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 4.485370950888192e-05, |
| "loss": 0.4017, |
| "step": 213 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 4.482758620689655e-05, |
| "loss": 0.4109, |
| "step": 214 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 4.480146290491119e-05, |
| "loss": 0.422, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 4.4775339602925815e-05, |
| "loss": 0.4035, |
| "step": 216 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 4.474921630094044e-05, |
| "loss": 0.4017, |
| "step": 217 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 4.4723092998955066e-05, |
| "loss": 0.427, |
| "step": 218 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 4.46969696969697e-05, |
| "loss": 0.4271, |
| "step": 219 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 4.467084639498433e-05, |
| "loss": 0.4151, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 4.464472309299896e-05, |
| "loss": 0.4046, |
| "step": 221 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 4.461859979101359e-05, |
| "loss": 0.4263, |
| "step": 222 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 4.4592476489028216e-05, |
| "loss": 0.41, |
| "step": 223 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 4.4566353187042844e-05, |
| "loss": 0.4192, |
| "step": 224 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 4.454022988505747e-05, |
| "loss": 0.4028, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 4.45141065830721e-05, |
| "loss": 0.4096, |
| "step": 226 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 4.448798328108673e-05, |
| "loss": 0.4424, |
| "step": 227 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 4.446185997910136e-05, |
| "loss": 0.4387, |
| "step": 228 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 4.4435736677115994e-05, |
| "loss": 0.4346, |
| "step": 229 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 4.440961337513062e-05, |
| "loss": 0.4337, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 4.4383490073145244e-05, |
| "loss": 0.4392, |
| "step": 231 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 4.435736677115987e-05, |
| "loss": 0.4344, |
| "step": 232 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 4.43312434691745e-05, |
| "loss": 0.4191, |
| "step": 233 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 4.430512016718914e-05, |
| "loss": 0.43, |
| "step": 234 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 4.4278996865203766e-05, |
| "loss": 0.4412, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 4.4252873563218394e-05, |
| "loss": 0.4238, |
| "step": 236 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 4.422675026123302e-05, |
| "loss": 0.4314, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 4.420062695924765e-05, |
| "loss": 0.4026, |
| "step": 238 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 4.417450365726228e-05, |
| "loss": 0.416, |
| "step": 239 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 4.414838035527691e-05, |
| "loss": 0.4136, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 4.412225705329154e-05, |
| "loss": 0.4328, |
| "step": 241 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 4.4096133751306166e-05, |
| "loss": 0.4275, |
| "step": 242 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 4.40700104493208e-05, |
| "loss": 0.3912, |
| "step": 243 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 4.404388714733543e-05, |
| "loss": 0.4268, |
| "step": 244 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 4.401776384535005e-05, |
| "loss": 0.4213, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 4.399164054336468e-05, |
| "loss": 0.4414, |
| "step": 246 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 4.396551724137931e-05, |
| "loss": 0.4148, |
| "step": 247 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 4.3939393939393944e-05, |
| "loss": 0.4056, |
| "step": 248 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 4.391327063740857e-05, |
| "loss": 0.4278, |
| "step": 249 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 4.38871473354232e-05, |
| "loss": 0.4081, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 4.386102403343783e-05, |
| "loss": 0.4098, |
| "step": 251 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 4.383490073145246e-05, |
| "loss": 0.4149, |
| "step": 252 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 4.380877742946709e-05, |
| "loss": 0.412, |
| "step": 253 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 4.3782654127481716e-05, |
| "loss": 0.4297, |
| "step": 254 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 4.3756530825496344e-05, |
| "loss": 0.4071, |
| "step": 255 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 4.373040752351097e-05, |
| "loss": 0.4183, |
| "step": 256 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 4.37042842215256e-05, |
| "loss": 0.4153, |
| "step": 257 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 4.367816091954024e-05, |
| "loss": 0.4234, |
| "step": 258 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 4.365203761755486e-05, |
| "loss": 0.4428, |
| "step": 259 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 4.362591431556949e-05, |
| "loss": 0.4028, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 4.3599791013584116e-05, |
| "loss": 0.4232, |
| "step": 261 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 4.357366771159875e-05, |
| "loss": 0.4192, |
| "step": 262 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 4.354754440961338e-05, |
| "loss": 0.4281, |
| "step": 263 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 4.352142110762801e-05, |
| "loss": 0.4258, |
| "step": 264 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 4.349529780564264e-05, |
| "loss": 0.4257, |
| "step": 265 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 4.346917450365726e-05, |
| "loss": 0.4195, |
| "step": 266 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 4.3443051201671894e-05, |
| "loss": 0.414, |
| "step": 267 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 4.341692789968652e-05, |
| "loss": 0.4229, |
| "step": 268 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 4.339080459770115e-05, |
| "loss": 0.42, |
| "step": 269 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 4.336468129571578e-05, |
| "loss": 0.4151, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 4.333855799373041e-05, |
| "loss": 0.4295, |
| "step": 271 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 4.3312434691745044e-05, |
| "loss": 0.4357, |
| "step": 272 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 4.3286311389759666e-05, |
| "loss": 0.4232, |
| "step": 273 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 4.3260188087774294e-05, |
| "loss": 0.4199, |
| "step": 274 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 4.323406478578892e-05, |
| "loss": 0.4183, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 4.320794148380355e-05, |
| "loss": 0.3994, |
| "step": 276 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 4.318181818181819e-05, |
| "loss": 0.4117, |
| "step": 277 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 4.3155694879832815e-05, |
| "loss": 0.392, |
| "step": 278 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 4.3129571577847444e-05, |
| "loss": 0.3938, |
| "step": 279 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 4.3103448275862066e-05, |
| "loss": 0.3969, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 4.30773249738767e-05, |
| "loss": 0.4147, |
| "step": 281 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 4.305120167189133e-05, |
| "loss": 0.432, |
| "step": 282 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 4.302507836990596e-05, |
| "loss": 0.4104, |
| "step": 283 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 4.299895506792059e-05, |
| "loss": 0.4035, |
| "step": 284 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 4.2972831765935215e-05, |
| "loss": 0.3853, |
| "step": 285 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 4.294670846394985e-05, |
| "loss": 0.4105, |
| "step": 286 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 4.292058516196447e-05, |
| "loss": 0.4179, |
| "step": 287 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 4.28944618599791e-05, |
| "loss": 0.4194, |
| "step": 288 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 4.286833855799373e-05, |
| "loss": 0.4225, |
| "step": 289 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 4.284221525600836e-05, |
| "loss": 0.415, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 4.2816091954022994e-05, |
| "loss": 0.4176, |
| "step": 291 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 4.278996865203762e-05, |
| "loss": 0.4073, |
| "step": 292 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 4.276384535005225e-05, |
| "loss": 0.4121, |
| "step": 293 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 4.273772204806687e-05, |
| "loss": 0.4135, |
| "step": 294 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 4.271159874608151e-05, |
| "loss": 0.4075, |
| "step": 295 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 4.268547544409614e-05, |
| "loss": 0.405, |
| "step": 296 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 4.2659352142110765e-05, |
| "loss": 0.3962, |
| "step": 297 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 4.2633228840125394e-05, |
| "loss": 0.4279, |
| "step": 298 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 4.260710553814002e-05, |
| "loss": 0.4064, |
| "step": 299 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 4.258098223615465e-05, |
| "loss": 0.4111, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 4.255485893416928e-05, |
| "loss": 0.3993, |
| "step": 301 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 4.252873563218391e-05, |
| "loss": 0.4256, |
| "step": 302 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 4.250261233019854e-05, |
| "loss": 0.3878, |
| "step": 303 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 4.2476489028213165e-05, |
| "loss": 0.3979, |
| "step": 304 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 4.24503657262278e-05, |
| "loss": 0.4282, |
| "step": 305 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 4.242424242424243e-05, |
| "loss": 0.4092, |
| "step": 306 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 4.239811912225706e-05, |
| "loss": 0.4134, |
| "step": 307 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 4.2371995820271687e-05, |
| "loss": 0.4001, |
| "step": 308 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 4.234587251828631e-05, |
| "loss": 0.4447, |
| "step": 309 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 4.2319749216300944e-05, |
| "loss": 0.4099, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 4.229362591431557e-05, |
| "loss": 0.4187, |
| "step": 311 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 4.22675026123302e-05, |
| "loss": 0.4173, |
| "step": 312 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 4.224137931034483e-05, |
| "loss": 0.408, |
| "step": 313 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 4.221525600835946e-05, |
| "loss": 0.4209, |
| "step": 314 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 4.2189132706374093e-05, |
| "loss": 0.4141, |
| "step": 315 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 4.2163009404388715e-05, |
| "loss": 0.4195, |
| "step": 316 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 4.2136886102403344e-05, |
| "loss": 0.4004, |
| "step": 317 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 4.211076280041797e-05, |
| "loss": 0.4192, |
| "step": 318 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 4.20846394984326e-05, |
| "loss": 0.4296, |
| "step": 319 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 4.2058516196447236e-05, |
| "loss": 0.412, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 4.2032392894461865e-05, |
| "loss": 0.391, |
| "step": 321 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 4.2006269592476494e-05, |
| "loss": 0.4201, |
| "step": 322 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 4.1980146290491115e-05, |
| "loss": 0.4121, |
| "step": 323 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 4.195402298850575e-05, |
| "loss": 0.4077, |
| "step": 324 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 4.192789968652038e-05, |
| "loss": 0.4202, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 4.190177638453501e-05, |
| "loss": 0.4228, |
| "step": 326 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 4.1875653082549637e-05, |
| "loss": 0.4196, |
| "step": 327 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 4.1849529780564265e-05, |
| "loss": 0.3991, |
| "step": 328 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 4.1823406478578894e-05, |
| "loss": 0.4226, |
| "step": 329 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 4.179728317659352e-05, |
| "loss": 0.4046, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 4.177115987460815e-05, |
| "loss": 0.4059, |
| "step": 331 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 4.174503657262278e-05, |
| "loss": 0.4201, |
| "step": 332 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 4.171891327063741e-05, |
| "loss": 0.4017, |
| "step": 333 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 4.1692789968652043e-05, |
| "loss": 0.3815, |
| "step": 334 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 4.166666666666667e-05, |
| "loss": 0.4223, |
| "step": 335 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 4.16405433646813e-05, |
| "loss": 0.4242, |
| "step": 336 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 4.161442006269592e-05, |
| "loss": 0.4013, |
| "step": 337 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 4.158829676071056e-05, |
| "loss": 0.4175, |
| "step": 338 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 4.1562173458725186e-05, |
| "loss": 0.4138, |
| "step": 339 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 4.1536050156739815e-05, |
| "loss": 0.4159, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 4.1509926854754444e-05, |
| "loss": 0.4141, |
| "step": 341 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 4.148380355276907e-05, |
| "loss": 0.426, |
| "step": 342 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 4.14576802507837e-05, |
| "loss": 0.4096, |
| "step": 343 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 4.143155694879833e-05, |
| "loss": 0.4132, |
| "step": 344 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 4.140543364681296e-05, |
| "loss": 0.3836, |
| "step": 345 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 4.1379310344827587e-05, |
| "loss": 0.4232, |
| "step": 346 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 4.1353187042842215e-05, |
| "loss": 0.4083, |
| "step": 347 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 4.132706374085685e-05, |
| "loss": 0.3974, |
| "step": 348 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 4.130094043887148e-05, |
| "loss": 0.4108, |
| "step": 349 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 4.127481713688611e-05, |
| "loss": 0.3954, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 4.124869383490073e-05, |
| "loss": 0.3984, |
| "step": 351 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 4.122257053291536e-05, |
| "loss": 0.3953, |
| "step": 352 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 4.1196447230929993e-05, |
| "loss": 0.4295, |
| "step": 353 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 4.117032392894462e-05, |
| "loss": 0.3987, |
| "step": 354 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 4.114420062695925e-05, |
| "loss": 0.3981, |
| "step": 355 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 4.111807732497388e-05, |
| "loss": 0.4059, |
| "step": 356 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 4.109195402298851e-05, |
| "loss": 0.3979, |
| "step": 357 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 4.1065830721003136e-05, |
| "loss": 0.4205, |
| "step": 358 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 4.1039707419017765e-05, |
| "loss": 0.3937, |
| "step": 359 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 4.1013584117032394e-05, |
| "loss": 0.3922, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 4.098746081504702e-05, |
| "loss": 0.3899, |
| "step": 361 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 4.096133751306165e-05, |
| "loss": 0.3906, |
| "step": 362 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 4.0935214211076286e-05, |
| "loss": 0.3983, |
| "step": 363 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 4.0909090909090915e-05, |
| "loss": 0.4041, |
| "step": 364 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 4.0882967607105537e-05, |
| "loss": 0.3866, |
| "step": 365 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 4.0856844305120165e-05, |
| "loss": 0.4318, |
| "step": 366 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 4.08307210031348e-05, |
| "loss": 0.4206, |
| "step": 367 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 4.080459770114943e-05, |
| "loss": 0.4092, |
| "step": 368 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 4.077847439916406e-05, |
| "loss": 0.3747, |
| "step": 369 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 4.0752351097178686e-05, |
| "loss": 0.4086, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 4.0726227795193315e-05, |
| "loss": 0.3921, |
| "step": 371 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 4.0700104493207943e-05, |
| "loss": 0.4016, |
| "step": 372 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 4.067398119122257e-05, |
| "loss": 0.3928, |
| "step": 373 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 4.06478578892372e-05, |
| "loss": 0.389, |
| "step": 374 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 4.062173458725183e-05, |
| "loss": 0.4054, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 4.059561128526646e-05, |
| "loss": 0.3906, |
| "step": 376 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 4.056948798328109e-05, |
| "loss": 0.3923, |
| "step": 377 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 4.054336468129572e-05, |
| "loss": 0.3978, |
| "step": 378 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 4.0517241379310344e-05, |
| "loss": 0.3775, |
| "step": 379 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 4.049111807732497e-05, |
| "loss": 0.3928, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 4.04649947753396e-05, |
| "loss": 0.4007, |
| "step": 381 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 4.0438871473354236e-05, |
| "loss": 0.3973, |
| "step": 382 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 4.0412748171368865e-05, |
| "loss": 0.405, |
| "step": 383 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 4.038662486938349e-05, |
| "loss": 0.4247, |
| "step": 384 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 4.036050156739812e-05, |
| "loss": 0.418, |
| "step": 385 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 4.033437826541275e-05, |
| "loss": 0.3924, |
| "step": 386 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 4.030825496342738e-05, |
| "loss": 0.3961, |
| "step": 387 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 4.028213166144201e-05, |
| "loss": 0.3983, |
| "step": 388 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 4.0256008359456636e-05, |
| "loss": 0.4047, |
| "step": 389 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 4.0229885057471265e-05, |
| "loss": 0.4149, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 4.02037617554859e-05, |
| "loss": 0.4073, |
| "step": 391 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 4.017763845350053e-05, |
| "loss": 0.4055, |
| "step": 392 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 4.015151515151515e-05, |
| "loss": 0.3951, |
| "step": 393 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 4.012539184952978e-05, |
| "loss": 0.4179, |
| "step": 394 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 4.009926854754441e-05, |
| "loss": 0.3858, |
| "step": 395 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 4.007314524555904e-05, |
| "loss": 0.3994, |
| "step": 396 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 4.004702194357367e-05, |
| "loss": 0.4001, |
| "step": 397 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 4.00208986415883e-05, |
| "loss": 0.4003, |
| "step": 398 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 3.999477533960293e-05, |
| "loss": 0.39, |
| "step": 399 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 3.996865203761756e-05, |
| "loss": 0.4044, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 3.9942528735632186e-05, |
| "loss": 0.3952, |
| "step": 401 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 3.9916405433646815e-05, |
| "loss": 0.4001, |
| "step": 402 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 3.989028213166144e-05, |
| "loss": 0.3842, |
| "step": 403 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 3.986415882967607e-05, |
| "loss": 0.384, |
| "step": 404 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 3.98380355276907e-05, |
| "loss": 0.4196, |
| "step": 405 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 3.9811912225705336e-05, |
| "loss": 0.3912, |
| "step": 406 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 3.978578892371996e-05, |
| "loss": 0.3914, |
| "step": 407 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 3.9759665621734586e-05, |
| "loss": 0.4165, |
| "step": 408 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 3.9733542319749215e-05, |
| "loss": 0.4064, |
| "step": 409 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 3.970741901776385e-05, |
| "loss": 0.4181, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 3.968129571577848e-05, |
| "loss": 0.3934, |
| "step": 411 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 3.965517241379311e-05, |
| "loss": 0.3912, |
| "step": 412 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 3.9629049111807736e-05, |
| "loss": 0.399, |
| "step": 413 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 3.960292580982236e-05, |
| "loss": 0.3849, |
| "step": 414 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 3.957680250783699e-05, |
| "loss": 0.4052, |
| "step": 415 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 3.955067920585162e-05, |
| "loss": 0.4329, |
| "step": 416 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 3.952455590386625e-05, |
| "loss": 0.4142, |
| "step": 417 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 3.949843260188088e-05, |
| "loss": 0.4152, |
| "step": 418 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 3.947230929989551e-05, |
| "loss": 0.4019, |
| "step": 419 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 3.944618599791014e-05, |
| "loss": 0.3879, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 3.9420062695924765e-05, |
| "loss": 0.4009, |
| "step": 421 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 3.939393939393939e-05, |
| "loss": 0.3969, |
| "step": 422 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 3.936781609195402e-05, |
| "loss": 0.4009, |
| "step": 423 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 3.934169278996865e-05, |
| "loss": 0.3888, |
| "step": 424 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 3.9315569487983286e-05, |
| "loss": 0.3958, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 3.9289446185997915e-05, |
| "loss": 0.3848, |
| "step": 426 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 3.926332288401254e-05, |
| "loss": 0.3984, |
| "step": 427 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 3.9237199582027165e-05, |
| "loss": 0.3879, |
| "step": 428 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 3.92110762800418e-05, |
| "loss": 0.3985, |
| "step": 429 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 3.918495297805643e-05, |
| "loss": 0.4057, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 3.915882967607106e-05, |
| "loss": 0.4007, |
| "step": 431 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 3.9132706374085686e-05, |
| "loss": 0.4301, |
| "step": 432 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 3.9106583072100315e-05, |
| "loss": 0.386, |
| "step": 433 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 3.908045977011495e-05, |
| "loss": 0.3777, |
| "step": 434 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 3.905433646812957e-05, |
| "loss": 0.4055, |
| "step": 435 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 3.90282131661442e-05, |
| "loss": 0.4114, |
| "step": 436 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 3.900208986415883e-05, |
| "loss": 0.4118, |
| "step": 437 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 3.897596656217346e-05, |
| "loss": 0.3895, |
| "step": 438 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 3.894984326018809e-05, |
| "loss": 0.4014, |
| "step": 439 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 3.892371995820272e-05, |
| "loss": 0.4024, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 3.889759665621735e-05, |
| "loss": 0.4122, |
| "step": 441 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 3.887147335423197e-05, |
| "loss": 0.4014, |
| "step": 442 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 3.884535005224661e-05, |
| "loss": 0.4123, |
| "step": 443 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 3.8819226750261236e-05, |
| "loss": 0.4266, |
| "step": 444 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 3.8793103448275865e-05, |
| "loss": 0.403, |
| "step": 445 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 3.876698014629049e-05, |
| "loss": 0.3819, |
| "step": 446 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 3.874085684430512e-05, |
| "loss": 0.3839, |
| "step": 447 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 3.871473354231975e-05, |
| "loss": 0.393, |
| "step": 448 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 3.868861024033438e-05, |
| "loss": 0.3874, |
| "step": 449 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 3.866248693834901e-05, |
| "loss": 0.4023, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 3.8636363636363636e-05, |
| "loss": 0.3799, |
| "step": 451 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 3.8610240334378265e-05, |
| "loss": 0.3886, |
| "step": 452 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 3.85841170323929e-05, |
| "loss": 0.4127, |
| "step": 453 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 3.855799373040753e-05, |
| "loss": 0.3971, |
| "step": 454 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 3.853187042842216e-05, |
| "loss": 0.3851, |
| "step": 455 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 3.850574712643678e-05, |
| "loss": 0.3939, |
| "step": 456 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 3.847962382445141e-05, |
| "loss": 0.3975, |
| "step": 457 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 3.845350052246604e-05, |
| "loss": 0.4067, |
| "step": 458 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 3.842737722048067e-05, |
| "loss": 0.3938, |
| "step": 459 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 3.84012539184953e-05, |
| "loss": 0.3952, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.837513061650993e-05, |
| "loss": 0.3966, |
| "step": 461 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.834900731452456e-05, |
| "loss": 0.374, |
| "step": 462 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.8322884012539186e-05, |
| "loss": 0.3966, |
| "step": 463 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.8296760710553815e-05, |
| "loss": 0.386, |
| "step": 464 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.827063740856844e-05, |
| "loss": 0.4226, |
| "step": 465 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.824451410658307e-05, |
| "loss": 0.3971, |
| "step": 466 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.82183908045977e-05, |
| "loss": 0.3856, |
| "step": 467 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.8192267502612336e-05, |
| "loss": 0.3989, |
| "step": 468 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.8166144200626964e-05, |
| "loss": 0.3947, |
| "step": 469 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.8140020898641586e-05, |
| "loss": 0.3979, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.8113897596656215e-05, |
| "loss": 0.4216, |
| "step": 471 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.808777429467085e-05, |
| "loss": 0.3947, |
| "step": 472 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.806165099268548e-05, |
| "loss": 0.3863, |
| "step": 473 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.803552769070011e-05, |
| "loss": 0.4041, |
| "step": 474 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.8009404388714736e-05, |
| "loss": 0.4067, |
| "step": 475 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.7983281086729364e-05, |
| "loss": 0.3855, |
| "step": 476 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.795715778474399e-05, |
| "loss": 0.4092, |
| "step": 477 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.793103448275862e-05, |
| "loss": 0.4007, |
| "step": 478 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.790491118077325e-05, |
| "loss": 0.4007, |
| "step": 479 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.787878787878788e-05, |
| "loss": 0.3998, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.785266457680251e-05, |
| "loss": 0.4042, |
| "step": 481 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.782654127481714e-05, |
| "loss": 0.3899, |
| "step": 482 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.780041797283177e-05, |
| "loss": 0.366, |
| "step": 483 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.777429467084639e-05, |
| "loss": 0.4055, |
| "step": 484 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.774817136886102e-05, |
| "loss": 0.375, |
| "step": 485 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 3.772204806687566e-05, |
| "loss": 0.3955, |
| "step": 486 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 3.7695924764890286e-05, |
| "loss": 0.4015, |
| "step": 487 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 3.7669801462904914e-05, |
| "loss": 0.3675, |
| "step": 488 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 3.764367816091954e-05, |
| "loss": 0.3962, |
| "step": 489 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 3.761755485893417e-05, |
| "loss": 0.3832, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 3.75914315569488e-05, |
| "loss": 0.4021, |
| "step": 491 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 3.756530825496343e-05, |
| "loss": 0.3808, |
| "step": 492 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 3.753918495297806e-05, |
| "loss": 0.4146, |
| "step": 493 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 3.7513061650992686e-05, |
| "loss": 0.3796, |
| "step": 494 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 3.7486938349007314e-05, |
| "loss": 0.3858, |
| "step": 495 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 3.746081504702195e-05, |
| "loss": 0.3966, |
| "step": 496 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 3.743469174503658e-05, |
| "loss": 0.3617, |
| "step": 497 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 3.740856844305121e-05, |
| "loss": 0.4047, |
| "step": 498 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 3.738244514106583e-05, |
| "loss": 0.3907, |
| "step": 499 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 3.735632183908046e-05, |
| "loss": 0.3831, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 3.733019853709509e-05, |
| "loss": 0.3896, |
| "step": 501 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 3.730407523510972e-05, |
| "loss": 0.3949, |
| "step": 502 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 3.727795193312435e-05, |
| "loss": 0.398, |
| "step": 503 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 3.725182863113898e-05, |
| "loss": 0.3747, |
| "step": 504 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 3.722570532915361e-05, |
| "loss": 0.3924, |
| "step": 505 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 3.7199582027168236e-05, |
| "loss": 0.3677, |
| "step": 506 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 3.7173458725182864e-05, |
| "loss": 0.4038, |
| "step": 507 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 3.714733542319749e-05, |
| "loss": 0.3791, |
| "step": 508 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 3.712121212121212e-05, |
| "loss": 0.3748, |
| "step": 509 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 3.709508881922675e-05, |
| "loss": 0.3975, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 3.7068965517241385e-05, |
| "loss": 0.3919, |
| "step": 511 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 3.7042842215256014e-05, |
| "loss": 0.4034, |
| "step": 512 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 3.7016718913270636e-05, |
| "loss": 0.4028, |
| "step": 513 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 3.6990595611285264e-05, |
| "loss": 0.3953, |
| "step": 514 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 3.69644723092999e-05, |
| "loss": 0.3997, |
| "step": 515 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 3.693834900731453e-05, |
| "loss": 0.3802, |
| "step": 516 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 3.691222570532916e-05, |
| "loss": 0.3852, |
| "step": 517 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 3.6886102403343786e-05, |
| "loss": 0.3997, |
| "step": 518 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 3.6859979101358414e-05, |
| "loss": 0.3892, |
| "step": 519 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 3.683385579937304e-05, |
| "loss": 0.3875, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 3.680773249738767e-05, |
| "loss": 0.4135, |
| "step": 521 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 3.67816091954023e-05, |
| "loss": 0.4028, |
| "step": 522 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 3.675548589341693e-05, |
| "loss": 0.3808, |
| "step": 523 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 3.672936259143156e-05, |
| "loss": 0.3847, |
| "step": 524 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 3.670323928944619e-05, |
| "loss": 0.3848, |
| "step": 525 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 3.667711598746082e-05, |
| "loss": 0.3956, |
| "step": 526 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 3.665099268547544e-05, |
| "loss": 0.3894, |
| "step": 527 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 3.662486938349007e-05, |
| "loss": 0.4044, |
| "step": 528 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 3.65987460815047e-05, |
| "loss": 0.4107, |
| "step": 529 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 3.6572622779519335e-05, |
| "loss": 0.3754, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 3.6546499477533964e-05, |
| "loss": 0.395, |
| "step": 531 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 3.652037617554859e-05, |
| "loss": 0.3756, |
| "step": 532 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 3.649425287356322e-05, |
| "loss": 0.398, |
| "step": 533 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 3.646812957157785e-05, |
| "loss": 0.3857, |
| "step": 534 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 3.644200626959248e-05, |
| "loss": 0.4171, |
| "step": 535 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 3.641588296760711e-05, |
| "loss": 0.3891, |
| "step": 536 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 3.6389759665621736e-05, |
| "loss": 0.3778, |
| "step": 537 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 3.6363636363636364e-05, |
| "loss": 0.393, |
| "step": 538 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 3.6337513061651e-05, |
| "loss": 0.3777, |
| "step": 539 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 3.631138975966563e-05, |
| "loss": 0.4173, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 3.628526645768025e-05, |
| "loss": 0.3964, |
| "step": 541 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 3.625914315569488e-05, |
| "loss": 0.4127, |
| "step": 542 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 3.623301985370951e-05, |
| "loss": 0.4251, |
| "step": 543 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 3.620689655172414e-05, |
| "loss": 0.3963, |
| "step": 544 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 3.618077324973877e-05, |
| "loss": 0.399, |
| "step": 545 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 3.61546499477534e-05, |
| "loss": 0.4132, |
| "step": 546 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 3.612852664576803e-05, |
| "loss": 0.393, |
| "step": 547 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 3.610240334378266e-05, |
| "loss": 0.3835, |
| "step": 548 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 3.6076280041797285e-05, |
| "loss": 0.4, |
| "step": 549 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 3.6050156739811914e-05, |
| "loss": 0.4011, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 3.602403343782654e-05, |
| "loss": 0.4007, |
| "step": 551 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 3.599791013584117e-05, |
| "loss": 0.4054, |
| "step": 552 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 3.59717868338558e-05, |
| "loss": 0.3736, |
| "step": 553 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 3.5945663531870435e-05, |
| "loss": 0.3793, |
| "step": 554 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 3.591954022988506e-05, |
| "loss": 0.3834, |
| "step": 555 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 3.5893416927899686e-05, |
| "loss": 0.3762, |
| "step": 556 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 3.5867293625914314e-05, |
| "loss": 0.3905, |
| "step": 557 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 3.584117032392895e-05, |
| "loss": 0.4097, |
| "step": 558 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 3.581504702194358e-05, |
| "loss": 0.3965, |
| "step": 559 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 3.578892371995821e-05, |
| "loss": 0.4038, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 3.5762800417972835e-05, |
| "loss": 0.3918, |
| "step": 561 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 3.573667711598746e-05, |
| "loss": 0.3782, |
| "step": 562 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 3.571055381400209e-05, |
| "loss": 0.3689, |
| "step": 563 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 3.568443051201672e-05, |
| "loss": 0.3647, |
| "step": 564 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 3.565830721003135e-05, |
| "loss": 0.3966, |
| "step": 565 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 3.563218390804598e-05, |
| "loss": 0.4129, |
| "step": 566 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 3.560606060606061e-05, |
| "loss": 0.3862, |
| "step": 567 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 3.557993730407524e-05, |
| "loss": 0.4038, |
| "step": 568 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 3.5553814002089864e-05, |
| "loss": 0.3795, |
| "step": 569 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 3.552769070010449e-05, |
| "loss": 0.3841, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 3.550156739811912e-05, |
| "loss": 0.3931, |
| "step": 571 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 3.547544409613375e-05, |
| "loss": 0.3715, |
| "step": 572 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 3.5449320794148385e-05, |
| "loss": 0.3813, |
| "step": 573 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 3.5423197492163014e-05, |
| "loss": 0.3811, |
| "step": 574 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 3.539707419017764e-05, |
| "loss": 0.3764, |
| "step": 575 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 3.5370950888192264e-05, |
| "loss": 0.3906, |
| "step": 576 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 3.53448275862069e-05, |
| "loss": 0.4007, |
| "step": 577 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 3.531870428422153e-05, |
| "loss": 0.376, |
| "step": 578 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 3.529258098223616e-05, |
| "loss": 0.3747, |
| "step": 579 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 3.5266457680250785e-05, |
| "loss": 0.396, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 3.5240334378265414e-05, |
| "loss": 0.3838, |
| "step": 581 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 3.521421107628005e-05, |
| "loss": 0.3767, |
| "step": 582 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 3.518808777429467e-05, |
| "loss": 0.3794, |
| "step": 583 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 3.51619644723093e-05, |
| "loss": 0.3919, |
| "step": 584 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 3.513584117032393e-05, |
| "loss": 0.3898, |
| "step": 585 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 3.510971786833856e-05, |
| "loss": 0.3894, |
| "step": 586 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 3.508359456635319e-05, |
| "loss": 0.3865, |
| "step": 587 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 3.505747126436782e-05, |
| "loss": 0.3864, |
| "step": 588 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 3.503134796238245e-05, |
| "loss": 0.389, |
| "step": 589 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 3.500522466039707e-05, |
| "loss": 0.4014, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 3.4979101358411707e-05, |
| "loss": 0.3954, |
| "step": 591 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 3.4952978056426335e-05, |
| "loss": 0.3803, |
| "step": 592 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 3.4926854754440964e-05, |
| "loss": 0.3604, |
| "step": 593 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 3.490073145245559e-05, |
| "loss": 0.4013, |
| "step": 594 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 3.487460815047022e-05, |
| "loss": 0.3882, |
| "step": 595 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 3.484848484848485e-05, |
| "loss": 0.3891, |
| "step": 596 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 3.482236154649948e-05, |
| "loss": 0.376, |
| "step": 597 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 3.479623824451411e-05, |
| "loss": 0.3789, |
| "step": 598 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 3.4770114942528735e-05, |
| "loss": 0.3763, |
| "step": 599 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 3.4743991640543364e-05, |
| "loss": 0.397, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 3.4717868338558e-05, |
| "loss": 0.4064, |
| "step": 601 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 3.469174503657263e-05, |
| "loss": 0.3887, |
| "step": 602 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 3.4665621734587256e-05, |
| "loss": 0.3755, |
| "step": 603 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 3.463949843260188e-05, |
| "loss": 0.3724, |
| "step": 604 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 3.461337513061651e-05, |
| "loss": 0.4033, |
| "step": 605 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 3.458725182863114e-05, |
| "loss": 0.3928, |
| "step": 606 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 3.456112852664577e-05, |
| "loss": 0.377, |
| "step": 607 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 3.45350052246604e-05, |
| "loss": 0.3609, |
| "step": 608 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 3.450888192267503e-05, |
| "loss": 0.3958, |
| "step": 609 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 3.4482758620689657e-05, |
| "loss": 0.4158, |
| "step": 610 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 3.4456635318704285e-05, |
| "loss": 0.3911, |
| "step": 611 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 3.4430512016718914e-05, |
| "loss": 0.3742, |
| "step": 612 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 3.440438871473354e-05, |
| "loss": 0.3894, |
| "step": 613 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 3.437826541274817e-05, |
| "loss": 0.4018, |
| "step": 614 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 3.43521421107628e-05, |
| "loss": 0.3743, |
| "step": 615 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 3.4326018808777435e-05, |
| "loss": 0.3985, |
| "step": 616 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 3.4299895506792064e-05, |
| "loss": 0.3817, |
| "step": 617 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 3.4273772204806685e-05, |
| "loss": 0.3989, |
| "step": 618 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 3.4247648902821314e-05, |
| "loss": 0.3793, |
| "step": 619 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 3.422152560083595e-05, |
| "loss": 0.3992, |
| "step": 620 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 3.419540229885058e-05, |
| "loss": 0.3894, |
| "step": 621 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 3.4169278996865206e-05, |
| "loss": 0.3763, |
| "step": 622 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 3.4143155694879835e-05, |
| "loss": 0.3693, |
| "step": 623 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 3.4117032392894464e-05, |
| "loss": 0.3953, |
| "step": 624 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 3.409090909090909e-05, |
| "loss": 0.3985, |
| "step": 625 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 3.406478578892372e-05, |
| "loss": 0.3701, |
| "step": 626 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 3.403866248693835e-05, |
| "loss": 0.3802, |
| "step": 627 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 3.401253918495298e-05, |
| "loss": 0.3827, |
| "step": 628 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 3.3986415882967607e-05, |
| "loss": 0.3741, |
| "step": 629 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 3.396029258098224e-05, |
| "loss": 0.3983, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 3.393416927899687e-05, |
| "loss": 0.3992, |
| "step": 631 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 3.390804597701149e-05, |
| "loss": 0.3695, |
| "step": 632 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 3.388192267502612e-05, |
| "loss": 0.393, |
| "step": 633 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 3.3855799373040756e-05, |
| "loss": 0.3699, |
| "step": 634 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 3.3829676071055385e-05, |
| "loss": 0.4006, |
| "step": 635 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 3.3803552769070014e-05, |
| "loss": 0.3955, |
| "step": 636 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 3.377742946708464e-05, |
| "loss": 0.4062, |
| "step": 637 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 3.375130616509927e-05, |
| "loss": 0.3975, |
| "step": 638 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 3.37251828631139e-05, |
| "loss": 0.3671, |
| "step": 639 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 3.369905956112853e-05, |
| "loss": 0.3919, |
| "step": 640 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.3672936259143156e-05, |
| "loss": 0.3523, |
| "step": 641 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.3646812957157785e-05, |
| "loss": 0.378, |
| "step": 642 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.3620689655172414e-05, |
| "loss": 0.3678, |
| "step": 643 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.359456635318705e-05, |
| "loss": 0.367, |
| "step": 644 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.356844305120168e-05, |
| "loss": 0.3152, |
| "step": 645 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.35423197492163e-05, |
| "loss": 0.3053, |
| "step": 646 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.351619644723093e-05, |
| "loss": 0.2968, |
| "step": 647 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.3490073145245557e-05, |
| "loss": 0.3183, |
| "step": 648 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.346394984326019e-05, |
| "loss": 0.3063, |
| "step": 649 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.343782654127482e-05, |
| "loss": 0.3038, |
| "step": 650 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.341170323928945e-05, |
| "loss": 0.315, |
| "step": 651 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.338557993730408e-05, |
| "loss": 0.3026, |
| "step": 652 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 3.3359456635318706e-05, |
| "loss": 0.31, |
| "step": 653 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.3333333333333335e-05, |
| "loss": 0.303, |
| "step": 654 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.3307210031347964e-05, |
| "loss": 0.3245, |
| "step": 655 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.328108672936259e-05, |
| "loss": 0.3064, |
| "step": 656 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.325496342737722e-05, |
| "loss": 0.3089, |
| "step": 657 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.322884012539185e-05, |
| "loss": 0.3056, |
| "step": 658 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.3202716823406485e-05, |
| "loss": 0.3041, |
| "step": 659 |
| }, |
| { |
| "epoch": 1.03, |
| "learning_rate": 3.3176593521421106e-05, |
| "loss": 0.3028, |
| "step": 660 |
| }, |
| { |
| "epoch": 1.03, |
| "learning_rate": 3.3150470219435735e-05, |
| "loss": 0.3157, |
| "step": 661 |
| }, |
| { |
| "epoch": 1.03, |
| "learning_rate": 3.3124346917450364e-05, |
| "loss": 0.2987, |
| "step": 662 |
| }, |
| { |
| "epoch": 1.03, |
| "learning_rate": 3.3098223615465e-05, |
| "loss": 0.3092, |
| "step": 663 |
| }, |
| { |
| "epoch": 1.03, |
| "learning_rate": 3.307210031347963e-05, |
| "loss": 0.3087, |
| "step": 664 |
| }, |
| { |
| "epoch": 1.03, |
| "learning_rate": 3.3045977011494256e-05, |
| "loss": 0.305, |
| "step": 665 |
| }, |
| { |
| "epoch": 1.03, |
| "learning_rate": 3.3019853709508885e-05, |
| "loss": 0.3143, |
| "step": 666 |
| }, |
| { |
| "epoch": 1.04, |
| "learning_rate": 3.299373040752351e-05, |
| "loss": 0.3041, |
| "step": 667 |
| }, |
| { |
| "epoch": 1.04, |
| "learning_rate": 3.296760710553814e-05, |
| "loss": 0.2872, |
| "step": 668 |
| }, |
| { |
| "epoch": 1.04, |
| "learning_rate": 3.294148380355277e-05, |
| "loss": 0.3206, |
| "step": 669 |
| }, |
| { |
| "epoch": 1.04, |
| "learning_rate": 3.29153605015674e-05, |
| "loss": 0.3193, |
| "step": 670 |
| }, |
| { |
| "epoch": 1.04, |
| "learning_rate": 3.288923719958203e-05, |
| "loss": 0.315, |
| "step": 671 |
| }, |
| { |
| "epoch": 1.04, |
| "learning_rate": 3.2863113897596656e-05, |
| "loss": 0.3073, |
| "step": 672 |
| }, |
| { |
| "epoch": 1.05, |
| "learning_rate": 3.283699059561129e-05, |
| "loss": 0.3163, |
| "step": 673 |
| }, |
| { |
| "epoch": 1.05, |
| "learning_rate": 3.2810867293625914e-05, |
| "loss": 0.2941, |
| "step": 674 |
| }, |
| { |
| "epoch": 1.05, |
| "learning_rate": 3.278474399164054e-05, |
| "loss": 0.319, |
| "step": 675 |
| }, |
| { |
| "epoch": 1.05, |
| "learning_rate": 3.275862068965517e-05, |
| "loss": 0.3183, |
| "step": 676 |
| }, |
| { |
| "epoch": 1.05, |
| "learning_rate": 3.27324973876698e-05, |
| "loss": 0.3048, |
| "step": 677 |
| }, |
| { |
| "epoch": 1.05, |
| "learning_rate": 3.2706374085684435e-05, |
| "loss": 0.2967, |
| "step": 678 |
| }, |
| { |
| "epoch": 1.05, |
| "learning_rate": 3.268025078369906e-05, |
| "loss": 0.322, |
| "step": 679 |
| }, |
| { |
| "epoch": 1.06, |
| "learning_rate": 3.265412748171369e-05, |
| "loss": 0.3055, |
| "step": 680 |
| }, |
| { |
| "epoch": 1.06, |
| "learning_rate": 3.2628004179728314e-05, |
| "loss": 0.3143, |
| "step": 681 |
| }, |
| { |
| "epoch": 1.06, |
| "learning_rate": 3.260188087774295e-05, |
| "loss": 0.2903, |
| "step": 682 |
| }, |
| { |
| "epoch": 1.06, |
| "learning_rate": 3.257575757575758e-05, |
| "loss": 0.3253, |
| "step": 683 |
| }, |
| { |
| "epoch": 1.06, |
| "learning_rate": 3.2549634273772206e-05, |
| "loss": 0.3241, |
| "step": 684 |
| }, |
| { |
| "epoch": 1.06, |
| "learning_rate": 3.2523510971786835e-05, |
| "loss": 0.3076, |
| "step": 685 |
| }, |
| { |
| "epoch": 1.07, |
| "learning_rate": 3.2497387669801463e-05, |
| "loss": 0.3147, |
| "step": 686 |
| }, |
| { |
| "epoch": 1.07, |
| "learning_rate": 3.24712643678161e-05, |
| "loss": 0.3108, |
| "step": 687 |
| }, |
| { |
| "epoch": 1.07, |
| "learning_rate": 3.244514106583073e-05, |
| "loss": 0.3071, |
| "step": 688 |
| }, |
| { |
| "epoch": 1.07, |
| "learning_rate": 3.241901776384535e-05, |
| "loss": 0.2992, |
| "step": 689 |
| }, |
| { |
| "epoch": 1.07, |
| "learning_rate": 3.239289446185998e-05, |
| "loss": 0.3068, |
| "step": 690 |
| }, |
| { |
| "epoch": 1.07, |
| "learning_rate": 3.2366771159874606e-05, |
| "loss": 0.2945, |
| "step": 691 |
| }, |
| { |
| "epoch": 1.08, |
| "learning_rate": 3.234064785788924e-05, |
| "loss": 0.2998, |
| "step": 692 |
| }, |
| { |
| "epoch": 1.08, |
| "learning_rate": 3.231452455590387e-05, |
| "loss": 0.3178, |
| "step": 693 |
| }, |
| { |
| "epoch": 1.08, |
| "learning_rate": 3.22884012539185e-05, |
| "loss": 0.3123, |
| "step": 694 |
| }, |
| { |
| "epoch": 1.08, |
| "learning_rate": 3.226227795193313e-05, |
| "loss": 0.3236, |
| "step": 695 |
| }, |
| { |
| "epoch": 1.08, |
| "learning_rate": 3.2236154649947756e-05, |
| "loss": 0.3132, |
| "step": 696 |
| }, |
| { |
| "epoch": 1.08, |
| "learning_rate": 3.2210031347962385e-05, |
| "loss": 0.2995, |
| "step": 697 |
| }, |
| { |
| "epoch": 1.08, |
| "learning_rate": 3.218390804597701e-05, |
| "loss": 0.3174, |
| "step": 698 |
| }, |
| { |
| "epoch": 1.09, |
| "learning_rate": 3.215778474399164e-05, |
| "loss": 0.2997, |
| "step": 699 |
| }, |
| { |
| "epoch": 1.09, |
| "learning_rate": 3.213166144200627e-05, |
| "loss": 0.3119, |
| "step": 700 |
| }, |
| { |
| "epoch": 1.09, |
| "learning_rate": 3.21055381400209e-05, |
| "loss": 0.3023, |
| "step": 701 |
| }, |
| { |
| "epoch": 1.09, |
| "learning_rate": 3.2079414838035534e-05, |
| "loss": 0.3002, |
| "step": 702 |
| }, |
| { |
| "epoch": 1.09, |
| "learning_rate": 3.2053291536050156e-05, |
| "loss": 0.2964, |
| "step": 703 |
| }, |
| { |
| "epoch": 1.09, |
| "learning_rate": 3.2027168234064785e-05, |
| "loss": 0.3023, |
| "step": 704 |
| }, |
| { |
| "epoch": 1.1, |
| "learning_rate": 3.2001044932079413e-05, |
| "loss": 0.3035, |
| "step": 705 |
| }, |
| { |
| "epoch": 1.1, |
| "learning_rate": 3.197492163009405e-05, |
| "loss": 0.3126, |
| "step": 706 |
| }, |
| { |
| "epoch": 1.1, |
| "learning_rate": 3.194879832810868e-05, |
| "loss": 0.3126, |
| "step": 707 |
| }, |
| { |
| "epoch": 1.1, |
| "learning_rate": 3.1922675026123306e-05, |
| "loss": 0.3112, |
| "step": 708 |
| }, |
| { |
| "epoch": 1.1, |
| "learning_rate": 3.1896551724137935e-05, |
| "loss": 0.3242, |
| "step": 709 |
| }, |
| { |
| "epoch": 1.1, |
| "learning_rate": 3.1870428422152556e-05, |
| "loss": 0.2953, |
| "step": 710 |
| }, |
| { |
| "epoch": 1.1, |
| "learning_rate": 3.184430512016719e-05, |
| "loss": 0.3031, |
| "step": 711 |
| }, |
| { |
| "epoch": 1.11, |
| "learning_rate": 3.181818181818182e-05, |
| "loss": 0.307, |
| "step": 712 |
| }, |
| { |
| "epoch": 1.11, |
| "learning_rate": 3.179205851619645e-05, |
| "loss": 0.3154, |
| "step": 713 |
| }, |
| { |
| "epoch": 1.11, |
| "learning_rate": 3.176593521421108e-05, |
| "loss": 0.2781, |
| "step": 714 |
| }, |
| { |
| "epoch": 1.11, |
| "learning_rate": 3.1739811912225706e-05, |
| "loss": 0.3196, |
| "step": 715 |
| }, |
| { |
| "epoch": 1.11, |
| "learning_rate": 3.171368861024034e-05, |
| "loss": 0.3043, |
| "step": 716 |
| }, |
| { |
| "epoch": 1.11, |
| "learning_rate": 3.168756530825496e-05, |
| "loss": 0.2995, |
| "step": 717 |
| }, |
| { |
| "epoch": 1.12, |
| "learning_rate": 3.166144200626959e-05, |
| "loss": 0.3214, |
| "step": 718 |
| }, |
| { |
| "epoch": 1.12, |
| "learning_rate": 3.163531870428422e-05, |
| "loss": 0.3149, |
| "step": 719 |
| }, |
| { |
| "epoch": 1.12, |
| "learning_rate": 3.160919540229885e-05, |
| "loss": 0.3081, |
| "step": 720 |
| }, |
| { |
| "epoch": 1.12, |
| "learning_rate": 3.1583072100313484e-05, |
| "loss": 0.3192, |
| "step": 721 |
| }, |
| { |
| "epoch": 1.12, |
| "learning_rate": 3.155694879832811e-05, |
| "loss": 0.2976, |
| "step": 722 |
| }, |
| { |
| "epoch": 1.12, |
| "learning_rate": 3.153082549634274e-05, |
| "loss": 0.3179, |
| "step": 723 |
| }, |
| { |
| "epoch": 1.12, |
| "learning_rate": 3.1504702194357363e-05, |
| "loss": 0.2817, |
| "step": 724 |
| }, |
| { |
| "epoch": 1.13, |
| "learning_rate": 3.1478578892372e-05, |
| "loss": 0.3099, |
| "step": 725 |
| }, |
| { |
| "epoch": 1.13, |
| "learning_rate": 3.145245559038663e-05, |
| "loss": 0.3007, |
| "step": 726 |
| }, |
| { |
| "epoch": 1.13, |
| "learning_rate": 3.1426332288401256e-05, |
| "loss": 0.3092, |
| "step": 727 |
| }, |
| { |
| "epoch": 1.13, |
| "learning_rate": 3.1400208986415885e-05, |
| "loss": 0.3045, |
| "step": 728 |
| }, |
| { |
| "epoch": 1.13, |
| "learning_rate": 3.137408568443051e-05, |
| "loss": 0.2978, |
| "step": 729 |
| }, |
| { |
| "epoch": 1.13, |
| "learning_rate": 3.134796238244515e-05, |
| "loss": 0.3092, |
| "step": 730 |
| }, |
| { |
| "epoch": 1.14, |
| "learning_rate": 3.132183908045977e-05, |
| "loss": 0.3117, |
| "step": 731 |
| }, |
| { |
| "epoch": 1.14, |
| "learning_rate": 3.12957157784744e-05, |
| "loss": 0.2887, |
| "step": 732 |
| }, |
| { |
| "epoch": 1.14, |
| "learning_rate": 3.126959247648903e-05, |
| "loss": 0.2982, |
| "step": 733 |
| }, |
| { |
| "epoch": 1.14, |
| "learning_rate": 3.1243469174503656e-05, |
| "loss": 0.2931, |
| "step": 734 |
| }, |
| { |
| "epoch": 1.14, |
| "learning_rate": 3.121734587251829e-05, |
| "loss": 0.313, |
| "step": 735 |
| }, |
| { |
| "epoch": 1.14, |
| "learning_rate": 3.119122257053292e-05, |
| "loss": 0.3134, |
| "step": 736 |
| }, |
| { |
| "epoch": 1.14, |
| "learning_rate": 3.116509926854755e-05, |
| "loss": 0.3054, |
| "step": 737 |
| }, |
| { |
| "epoch": 1.15, |
| "learning_rate": 3.113897596656217e-05, |
| "loss": 0.3029, |
| "step": 738 |
| }, |
| { |
| "epoch": 1.15, |
| "learning_rate": 3.1112852664576806e-05, |
| "loss": 0.3083, |
| "step": 739 |
| }, |
| { |
| "epoch": 1.15, |
| "learning_rate": 3.1086729362591434e-05, |
| "loss": 0.3132, |
| "step": 740 |
| }, |
| { |
| "epoch": 1.15, |
| "learning_rate": 3.106060606060606e-05, |
| "loss": 0.3251, |
| "step": 741 |
| }, |
| { |
| "epoch": 1.15, |
| "learning_rate": 3.103448275862069e-05, |
| "loss": 0.3168, |
| "step": 742 |
| }, |
| { |
| "epoch": 1.15, |
| "learning_rate": 3.100835945663532e-05, |
| "loss": 0.3033, |
| "step": 743 |
| }, |
| { |
| "epoch": 1.16, |
| "learning_rate": 3.098223615464995e-05, |
| "loss": 0.3046, |
| "step": 744 |
| }, |
| { |
| "epoch": 1.16, |
| "learning_rate": 3.095611285266458e-05, |
| "loss": 0.3128, |
| "step": 745 |
| }, |
| { |
| "epoch": 1.16, |
| "learning_rate": 3.0929989550679206e-05, |
| "loss": 0.3009, |
| "step": 746 |
| }, |
| { |
| "epoch": 1.16, |
| "learning_rate": 3.0903866248693835e-05, |
| "loss": 0.3315, |
| "step": 747 |
| }, |
| { |
| "epoch": 1.16, |
| "learning_rate": 3.087774294670846e-05, |
| "loss": 0.3169, |
| "step": 748 |
| }, |
| { |
| "epoch": 1.16, |
| "learning_rate": 3.08516196447231e-05, |
| "loss": 0.3146, |
| "step": 749 |
| }, |
| { |
| "epoch": 1.17, |
| "learning_rate": 3.082549634273773e-05, |
| "loss": 0.3108, |
| "step": 750 |
| }, |
| { |
| "epoch": 1.17, |
| "learning_rate": 3.0799373040752356e-05, |
| "loss": 0.3223, |
| "step": 751 |
| }, |
| { |
| "epoch": 1.17, |
| "learning_rate": 3.077324973876698e-05, |
| "loss": 0.322, |
| "step": 752 |
| }, |
| { |
| "epoch": 1.17, |
| "learning_rate": 3.0747126436781606e-05, |
| "loss": 0.3044, |
| "step": 753 |
| }, |
| { |
| "epoch": 1.17, |
| "learning_rate": 3.072100313479624e-05, |
| "loss": 0.286, |
| "step": 754 |
| }, |
| { |
| "epoch": 1.17, |
| "learning_rate": 3.069487983281087e-05, |
| "loss": 0.3074, |
| "step": 755 |
| }, |
| { |
| "epoch": 1.17, |
| "learning_rate": 3.06687565308255e-05, |
| "loss": 0.3027, |
| "step": 756 |
| }, |
| { |
| "epoch": 1.18, |
| "learning_rate": 3.064263322884013e-05, |
| "loss": 0.3125, |
| "step": 757 |
| }, |
| { |
| "epoch": 1.18, |
| "learning_rate": 3.0616509926854756e-05, |
| "loss": 0.3077, |
| "step": 758 |
| }, |
| { |
| "epoch": 1.18, |
| "learning_rate": 3.0590386624869384e-05, |
| "loss": 0.2965, |
| "step": 759 |
| }, |
| { |
| "epoch": 1.18, |
| "learning_rate": 3.056426332288401e-05, |
| "loss": 0.3067, |
| "step": 760 |
| }, |
| { |
| "epoch": 1.18, |
| "learning_rate": 3.053814002089864e-05, |
| "loss": 0.3165, |
| "step": 761 |
| }, |
| { |
| "epoch": 1.18, |
| "learning_rate": 3.0512016718913274e-05, |
| "loss": 0.31, |
| "step": 762 |
| }, |
| { |
| "epoch": 1.19, |
| "learning_rate": 3.0485893416927902e-05, |
| "loss": 0.3151, |
| "step": 763 |
| }, |
| { |
| "epoch": 1.19, |
| "learning_rate": 3.045977011494253e-05, |
| "loss": 0.312, |
| "step": 764 |
| }, |
| { |
| "epoch": 1.19, |
| "learning_rate": 3.0433646812957163e-05, |
| "loss": 0.2901, |
| "step": 765 |
| }, |
| { |
| "epoch": 1.19, |
| "learning_rate": 3.0407523510971785e-05, |
| "loss": 0.3114, |
| "step": 766 |
| }, |
| { |
| "epoch": 1.19, |
| "learning_rate": 3.0381400208986417e-05, |
| "loss": 0.3083, |
| "step": 767 |
| }, |
| { |
| "epoch": 1.19, |
| "learning_rate": 3.0355276907001045e-05, |
| "loss": 0.3224, |
| "step": 768 |
| }, |
| { |
| "epoch": 1.19, |
| "learning_rate": 3.0329153605015677e-05, |
| "loss": 0.3227, |
| "step": 769 |
| }, |
| { |
| "epoch": 1.2, |
| "learning_rate": 3.0303030303030306e-05, |
| "loss": 0.2994, |
| "step": 770 |
| }, |
| { |
| "epoch": 1.2, |
| "learning_rate": 3.0276907001044934e-05, |
| "loss": 0.3009, |
| "step": 771 |
| }, |
| { |
| "epoch": 1.2, |
| "learning_rate": 3.0250783699059566e-05, |
| "loss": 0.2915, |
| "step": 772 |
| }, |
| { |
| "epoch": 1.2, |
| "learning_rate": 3.0224660397074188e-05, |
| "loss": 0.2994, |
| "step": 773 |
| }, |
| { |
| "epoch": 1.2, |
| "learning_rate": 3.019853709508882e-05, |
| "loss": 0.3045, |
| "step": 774 |
| }, |
| { |
| "epoch": 1.2, |
| "learning_rate": 3.017241379310345e-05, |
| "loss": 0.3204, |
| "step": 775 |
| }, |
| { |
| "epoch": 1.21, |
| "learning_rate": 3.0146290491118077e-05, |
| "loss": 0.3244, |
| "step": 776 |
| }, |
| { |
| "epoch": 1.21, |
| "learning_rate": 3.012016718913271e-05, |
| "loss": 0.3132, |
| "step": 777 |
| }, |
| { |
| "epoch": 1.21, |
| "learning_rate": 3.0094043887147338e-05, |
| "loss": 0.312, |
| "step": 778 |
| }, |
| { |
| "epoch": 1.21, |
| "learning_rate": 3.006792058516197e-05, |
| "loss": 0.3229, |
| "step": 779 |
| }, |
| { |
| "epoch": 1.21, |
| "learning_rate": 3.004179728317659e-05, |
| "loss": 0.3218, |
| "step": 780 |
| }, |
| { |
| "epoch": 1.21, |
| "learning_rate": 3.0015673981191224e-05, |
| "loss": 0.3054, |
| "step": 781 |
| }, |
| { |
| "epoch": 1.21, |
| "learning_rate": 2.9989550679205852e-05, |
| "loss": 0.3149, |
| "step": 782 |
| }, |
| { |
| "epoch": 1.22, |
| "learning_rate": 2.996342737722048e-05, |
| "loss": 0.2986, |
| "step": 783 |
| }, |
| { |
| "epoch": 1.22, |
| "learning_rate": 2.9937304075235113e-05, |
| "loss": 0.3096, |
| "step": 784 |
| }, |
| { |
| "epoch": 1.22, |
| "learning_rate": 2.991118077324974e-05, |
| "loss": 0.2862, |
| "step": 785 |
| }, |
| { |
| "epoch": 1.22, |
| "learning_rate": 2.988505747126437e-05, |
| "loss": 0.2911, |
| "step": 786 |
| }, |
| { |
| "epoch": 1.22, |
| "learning_rate": 2.9858934169278995e-05, |
| "loss": 0.3049, |
| "step": 787 |
| }, |
| { |
| "epoch": 1.22, |
| "learning_rate": 2.9832810867293627e-05, |
| "loss": 0.3041, |
| "step": 788 |
| }, |
| { |
| "epoch": 1.23, |
| "learning_rate": 2.9806687565308256e-05, |
| "loss": 0.3006, |
| "step": 789 |
| }, |
| { |
| "epoch": 1.23, |
| "learning_rate": 2.9780564263322884e-05, |
| "loss": 0.3174, |
| "step": 790 |
| }, |
| { |
| "epoch": 1.23, |
| "learning_rate": 2.9754440961337516e-05, |
| "loss": 0.2918, |
| "step": 791 |
| }, |
| { |
| "epoch": 1.23, |
| "learning_rate": 2.9728317659352145e-05, |
| "loss": 0.3122, |
| "step": 792 |
| }, |
| { |
| "epoch": 1.23, |
| "learning_rate": 2.9702194357366773e-05, |
| "loss": 0.2856, |
| "step": 793 |
| }, |
| { |
| "epoch": 1.23, |
| "learning_rate": 2.96760710553814e-05, |
| "loss": 0.2981, |
| "step": 794 |
| }, |
| { |
| "epoch": 1.24, |
| "learning_rate": 2.964994775339603e-05, |
| "loss": 0.3022, |
| "step": 795 |
| }, |
| { |
| "epoch": 1.24, |
| "learning_rate": 2.962382445141066e-05, |
| "loss": 0.2917, |
| "step": 796 |
| }, |
| { |
| "epoch": 1.24, |
| "learning_rate": 2.9597701149425288e-05, |
| "loss": 0.2866, |
| "step": 797 |
| }, |
| { |
| "epoch": 1.24, |
| "learning_rate": 2.957157784743992e-05, |
| "loss": 0.3102, |
| "step": 798 |
| }, |
| { |
| "epoch": 1.24, |
| "learning_rate": 2.954545454545455e-05, |
| "loss": 0.3078, |
| "step": 799 |
| }, |
| { |
| "epoch": 1.24, |
| "learning_rate": 2.9519331243469177e-05, |
| "loss": 0.3086, |
| "step": 800 |
| }, |
| { |
| "epoch": 1.24, |
| "learning_rate": 2.9493207941483802e-05, |
| "loss": 0.3042, |
| "step": 801 |
| }, |
| { |
| "epoch": 1.25, |
| "learning_rate": 2.946708463949843e-05, |
| "loss": 0.3162, |
| "step": 802 |
| }, |
| { |
| "epoch": 1.25, |
| "learning_rate": 2.9440961337513063e-05, |
| "loss": 0.3237, |
| "step": 803 |
| }, |
| { |
| "epoch": 1.25, |
| "learning_rate": 2.941483803552769e-05, |
| "loss": 0.3124, |
| "step": 804 |
| }, |
| { |
| "epoch": 1.25, |
| "learning_rate": 2.9388714733542323e-05, |
| "loss": 0.301, |
| "step": 805 |
| }, |
| { |
| "epoch": 1.25, |
| "learning_rate": 2.9362591431556952e-05, |
| "loss": 0.3028, |
| "step": 806 |
| }, |
| { |
| "epoch": 1.25, |
| "learning_rate": 2.933646812957158e-05, |
| "loss": 0.3025, |
| "step": 807 |
| }, |
| { |
| "epoch": 1.26, |
| "learning_rate": 2.9310344827586206e-05, |
| "loss": 0.3114, |
| "step": 808 |
| }, |
| { |
| "epoch": 1.26, |
| "learning_rate": 2.9284221525600834e-05, |
| "loss": 0.2852, |
| "step": 809 |
| }, |
| { |
| "epoch": 1.26, |
| "learning_rate": 2.9258098223615466e-05, |
| "loss": 0.3045, |
| "step": 810 |
| }, |
| { |
| "epoch": 1.26, |
| "learning_rate": 2.9231974921630095e-05, |
| "loss": 0.3149, |
| "step": 811 |
| }, |
| { |
| "epoch": 1.26, |
| "learning_rate": 2.9205851619644723e-05, |
| "loss": 0.3062, |
| "step": 812 |
| }, |
| { |
| "epoch": 1.26, |
| "learning_rate": 2.9179728317659355e-05, |
| "loss": 0.2979, |
| "step": 813 |
| }, |
| { |
| "epoch": 1.26, |
| "learning_rate": 2.9153605015673984e-05, |
| "loss": 0.3275, |
| "step": 814 |
| }, |
| { |
| "epoch": 1.27, |
| "learning_rate": 2.912748171368861e-05, |
| "loss": 0.3051, |
| "step": 815 |
| }, |
| { |
| "epoch": 1.27, |
| "learning_rate": 2.9101358411703238e-05, |
| "loss": 0.3121, |
| "step": 816 |
| }, |
| { |
| "epoch": 1.27, |
| "learning_rate": 2.907523510971787e-05, |
| "loss": 0.3037, |
| "step": 817 |
| }, |
| { |
| "epoch": 1.27, |
| "learning_rate": 2.90491118077325e-05, |
| "loss": 0.2939, |
| "step": 818 |
| }, |
| { |
| "epoch": 1.27, |
| "learning_rate": 2.9022988505747127e-05, |
| "loss": 0.2971, |
| "step": 819 |
| }, |
| { |
| "epoch": 1.27, |
| "learning_rate": 2.899686520376176e-05, |
| "loss": 0.2944, |
| "step": 820 |
| }, |
| { |
| "epoch": 1.28, |
| "learning_rate": 2.8970741901776388e-05, |
| "loss": 0.3198, |
| "step": 821 |
| }, |
| { |
| "epoch": 1.28, |
| "learning_rate": 2.8944618599791013e-05, |
| "loss": 0.3047, |
| "step": 822 |
| }, |
| { |
| "epoch": 1.28, |
| "learning_rate": 2.891849529780564e-05, |
| "loss": 0.3077, |
| "step": 823 |
| }, |
| { |
| "epoch": 1.28, |
| "learning_rate": 2.8892371995820273e-05, |
| "loss": 0.2946, |
| "step": 824 |
| }, |
| { |
| "epoch": 1.28, |
| "learning_rate": 2.8866248693834902e-05, |
| "loss": 0.2761, |
| "step": 825 |
| }, |
| { |
| "epoch": 1.28, |
| "learning_rate": 2.884012539184953e-05, |
| "loss": 0.3179, |
| "step": 826 |
| }, |
| { |
| "epoch": 1.28, |
| "learning_rate": 2.8814002089864163e-05, |
| "loss": 0.3052, |
| "step": 827 |
| }, |
| { |
| "epoch": 1.29, |
| "learning_rate": 2.878787878787879e-05, |
| "loss": 0.3148, |
| "step": 828 |
| }, |
| { |
| "epoch": 1.29, |
| "learning_rate": 2.8761755485893416e-05, |
| "loss": 0.3034, |
| "step": 829 |
| }, |
| { |
| "epoch": 1.29, |
| "learning_rate": 2.8735632183908045e-05, |
| "loss": 0.3089, |
| "step": 830 |
| }, |
| { |
| "epoch": 1.29, |
| "learning_rate": 2.8709508881922677e-05, |
| "loss": 0.3111, |
| "step": 831 |
| }, |
| { |
| "epoch": 1.29, |
| "learning_rate": 2.8683385579937305e-05, |
| "loss": 0.2996, |
| "step": 832 |
| }, |
| { |
| "epoch": 1.29, |
| "learning_rate": 2.8657262277951934e-05, |
| "loss": 0.3041, |
| "step": 833 |
| }, |
| { |
| "epoch": 1.3, |
| "learning_rate": 2.8631138975966566e-05, |
| "loss": 0.3065, |
| "step": 834 |
| }, |
| { |
| "epoch": 1.3, |
| "learning_rate": 2.8605015673981195e-05, |
| "loss": 0.2998, |
| "step": 835 |
| }, |
| { |
| "epoch": 1.3, |
| "learning_rate": 2.857889237199582e-05, |
| "loss": 0.3187, |
| "step": 836 |
| }, |
| { |
| "epoch": 1.3, |
| "learning_rate": 2.855276907001045e-05, |
| "loss": 0.2985, |
| "step": 837 |
| }, |
| { |
| "epoch": 1.3, |
| "learning_rate": 2.8526645768025077e-05, |
| "loss": 0.3017, |
| "step": 838 |
| }, |
| { |
| "epoch": 1.3, |
| "learning_rate": 2.850052246603971e-05, |
| "loss": 0.2857, |
| "step": 839 |
| }, |
| { |
| "epoch": 1.3, |
| "learning_rate": 2.8474399164054338e-05, |
| "loss": 0.3146, |
| "step": 840 |
| }, |
| { |
| "epoch": 1.31, |
| "learning_rate": 2.844827586206897e-05, |
| "loss": 0.2899, |
| "step": 841 |
| }, |
| { |
| "epoch": 1.31, |
| "learning_rate": 2.8422152560083598e-05, |
| "loss": 0.3113, |
| "step": 842 |
| }, |
| { |
| "epoch": 1.31, |
| "learning_rate": 2.8396029258098223e-05, |
| "loss": 0.2991, |
| "step": 843 |
| }, |
| { |
| "epoch": 1.31, |
| "learning_rate": 2.8369905956112852e-05, |
| "loss": 0.2985, |
| "step": 844 |
| }, |
| { |
| "epoch": 1.31, |
| "learning_rate": 2.834378265412748e-05, |
| "loss": 0.322, |
| "step": 845 |
| }, |
| { |
| "epoch": 1.31, |
| "learning_rate": 2.8317659352142113e-05, |
| "loss": 0.3103, |
| "step": 846 |
| }, |
| { |
| "epoch": 1.32, |
| "learning_rate": 2.829153605015674e-05, |
| "loss": 0.3019, |
| "step": 847 |
| }, |
| { |
| "epoch": 1.32, |
| "learning_rate": 2.8265412748171373e-05, |
| "loss": 0.3203, |
| "step": 848 |
| }, |
| { |
| "epoch": 1.32, |
| "learning_rate": 2.8239289446186e-05, |
| "loss": 0.3025, |
| "step": 849 |
| }, |
| { |
| "epoch": 1.32, |
| "learning_rate": 2.8213166144200627e-05, |
| "loss": 0.2975, |
| "step": 850 |
| }, |
| { |
| "epoch": 1.32, |
| "learning_rate": 2.8187042842215255e-05, |
| "loss": 0.3003, |
| "step": 851 |
| }, |
| { |
| "epoch": 1.32, |
| "learning_rate": 2.8160919540229884e-05, |
| "loss": 0.3224, |
| "step": 852 |
| }, |
| { |
| "epoch": 1.33, |
| "learning_rate": 2.8134796238244516e-05, |
| "loss": 0.2909, |
| "step": 853 |
| }, |
| { |
| "epoch": 1.33, |
| "learning_rate": 2.8108672936259145e-05, |
| "loss": 0.2868, |
| "step": 854 |
| }, |
| { |
| "epoch": 1.33, |
| "learning_rate": 2.8082549634273773e-05, |
| "loss": 0.2977, |
| "step": 855 |
| }, |
| { |
| "epoch": 1.33, |
| "learning_rate": 2.8056426332288405e-05, |
| "loss": 0.3143, |
| "step": 856 |
| }, |
| { |
| "epoch": 1.33, |
| "learning_rate": 2.803030303030303e-05, |
| "loss": 0.3159, |
| "step": 857 |
| }, |
| { |
| "epoch": 1.33, |
| "learning_rate": 2.800417972831766e-05, |
| "loss": 0.3086, |
| "step": 858 |
| }, |
| { |
| "epoch": 1.33, |
| "learning_rate": 2.7978056426332288e-05, |
| "loss": 0.3146, |
| "step": 859 |
| }, |
| { |
| "epoch": 1.34, |
| "learning_rate": 2.795193312434692e-05, |
| "loss": 0.3056, |
| "step": 860 |
| }, |
| { |
| "epoch": 1.34, |
| "learning_rate": 2.7925809822361548e-05, |
| "loss": 0.3108, |
| "step": 861 |
| }, |
| { |
| "epoch": 1.34, |
| "learning_rate": 2.7899686520376177e-05, |
| "loss": 0.2958, |
| "step": 862 |
| }, |
| { |
| "epoch": 1.34, |
| "learning_rate": 2.787356321839081e-05, |
| "loss": 0.313, |
| "step": 863 |
| }, |
| { |
| "epoch": 1.34, |
| "learning_rate": 2.7847439916405434e-05, |
| "loss": 0.3069, |
| "step": 864 |
| }, |
| { |
| "epoch": 1.34, |
| "learning_rate": 2.7821316614420063e-05, |
| "loss": 0.3186, |
| "step": 865 |
| }, |
| { |
| "epoch": 1.35, |
| "learning_rate": 2.779519331243469e-05, |
| "loss": 0.3095, |
| "step": 866 |
| }, |
| { |
| "epoch": 1.35, |
| "learning_rate": 2.7769070010449323e-05, |
| "loss": 0.3146, |
| "step": 867 |
| }, |
| { |
| "epoch": 1.35, |
| "learning_rate": 2.774294670846395e-05, |
| "loss": 0.3118, |
| "step": 868 |
| }, |
| { |
| "epoch": 1.35, |
| "learning_rate": 2.771682340647858e-05, |
| "loss": 0.3064, |
| "step": 869 |
| }, |
| { |
| "epoch": 1.35, |
| "learning_rate": 2.7690700104493212e-05, |
| "loss": 0.3109, |
| "step": 870 |
| }, |
| { |
| "epoch": 1.35, |
| "learning_rate": 2.7664576802507834e-05, |
| "loss": 0.3063, |
| "step": 871 |
| }, |
| { |
| "epoch": 1.35, |
| "learning_rate": 2.7638453500522466e-05, |
| "loss": 0.3123, |
| "step": 872 |
| }, |
| { |
| "epoch": 1.36, |
| "learning_rate": 2.7612330198537095e-05, |
| "loss": 0.2943, |
| "step": 873 |
| }, |
| { |
| "epoch": 1.36, |
| "learning_rate": 2.7586206896551727e-05, |
| "loss": 0.308, |
| "step": 874 |
| }, |
| { |
| "epoch": 1.36, |
| "learning_rate": 2.7560083594566355e-05, |
| "loss": 0.3075, |
| "step": 875 |
| }, |
| { |
| "epoch": 1.36, |
| "learning_rate": 2.7533960292580984e-05, |
| "loss": 0.2947, |
| "step": 876 |
| }, |
| { |
| "epoch": 1.36, |
| "learning_rate": 2.7507836990595616e-05, |
| "loss": 0.3104, |
| "step": 877 |
| }, |
| { |
| "epoch": 1.36, |
| "learning_rate": 2.7481713688610244e-05, |
| "loss": 0.3111, |
| "step": 878 |
| }, |
| { |
| "epoch": 1.37, |
| "learning_rate": 2.745559038662487e-05, |
| "loss": 0.2942, |
| "step": 879 |
| }, |
| { |
| "epoch": 1.37, |
| "learning_rate": 2.7429467084639498e-05, |
| "loss": 0.3052, |
| "step": 880 |
| }, |
| { |
| "epoch": 1.37, |
| "learning_rate": 2.7403343782654127e-05, |
| "loss": 0.2903, |
| "step": 881 |
| }, |
| { |
| "epoch": 1.37, |
| "learning_rate": 2.737722048066876e-05, |
| "loss": 0.3045, |
| "step": 882 |
| }, |
| { |
| "epoch": 1.37, |
| "learning_rate": 2.7351097178683387e-05, |
| "loss": 0.2894, |
| "step": 883 |
| }, |
| { |
| "epoch": 1.37, |
| "learning_rate": 2.732497387669802e-05, |
| "loss": 0.3062, |
| "step": 884 |
| }, |
| { |
| "epoch": 1.37, |
| "learning_rate": 2.7298850574712648e-05, |
| "loss": 0.3037, |
| "step": 885 |
| }, |
| { |
| "epoch": 1.38, |
| "learning_rate": 2.7272727272727273e-05, |
| "loss": 0.3175, |
| "step": 886 |
| }, |
| { |
| "epoch": 1.38, |
| "learning_rate": 2.7246603970741902e-05, |
| "loss": 0.3206, |
| "step": 887 |
| }, |
| { |
| "epoch": 1.38, |
| "learning_rate": 2.722048066875653e-05, |
| "loss": 0.2861, |
| "step": 888 |
| }, |
| { |
| "epoch": 1.38, |
| "learning_rate": 2.7194357366771162e-05, |
| "loss": 0.3021, |
| "step": 889 |
| }, |
| { |
| "epoch": 1.38, |
| "learning_rate": 2.716823406478579e-05, |
| "loss": 0.304, |
| "step": 890 |
| }, |
| { |
| "epoch": 1.38, |
| "learning_rate": 2.7142110762800423e-05, |
| "loss": 0.2937, |
| "step": 891 |
| }, |
| { |
| "epoch": 1.39, |
| "learning_rate": 2.711598746081505e-05, |
| "loss": 0.3101, |
| "step": 892 |
| }, |
| { |
| "epoch": 1.39, |
| "learning_rate": 2.7089864158829677e-05, |
| "loss": 0.2926, |
| "step": 893 |
| }, |
| { |
| "epoch": 1.39, |
| "learning_rate": 2.7063740856844305e-05, |
| "loss": 0.3169, |
| "step": 894 |
| }, |
| { |
| "epoch": 1.39, |
| "learning_rate": 2.7037617554858934e-05, |
| "loss": 0.3116, |
| "step": 895 |
| }, |
| { |
| "epoch": 1.39, |
| "learning_rate": 2.7011494252873566e-05, |
| "loss": 0.3062, |
| "step": 896 |
| }, |
| { |
| "epoch": 1.39, |
| "learning_rate": 2.6985370950888194e-05, |
| "loss": 0.3049, |
| "step": 897 |
| }, |
| { |
| "epoch": 1.4, |
| "learning_rate": 2.6959247648902823e-05, |
| "loss": 0.3014, |
| "step": 898 |
| }, |
| { |
| "epoch": 1.4, |
| "learning_rate": 2.6933124346917455e-05, |
| "loss": 0.3064, |
| "step": 899 |
| }, |
| { |
| "epoch": 1.4, |
| "learning_rate": 2.690700104493208e-05, |
| "loss": 0.3266, |
| "step": 900 |
| }, |
| { |
| "epoch": 1.4, |
| "learning_rate": 2.688087774294671e-05, |
| "loss": 0.3066, |
| "step": 901 |
| }, |
| { |
| "epoch": 1.4, |
| "learning_rate": 2.6854754440961337e-05, |
| "loss": 0.2853, |
| "step": 902 |
| }, |
| { |
| "epoch": 1.4, |
| "learning_rate": 2.682863113897597e-05, |
| "loss": 0.3028, |
| "step": 903 |
| }, |
| { |
| "epoch": 1.4, |
| "learning_rate": 2.6802507836990598e-05, |
| "loss": 0.3008, |
| "step": 904 |
| }, |
| { |
| "epoch": 1.41, |
| "learning_rate": 2.6776384535005227e-05, |
| "loss": 0.3139, |
| "step": 905 |
| }, |
| { |
| "epoch": 1.41, |
| "learning_rate": 2.675026123301986e-05, |
| "loss": 0.3, |
| "step": 906 |
| }, |
| { |
| "epoch": 1.41, |
| "learning_rate": 2.672413793103448e-05, |
| "loss": 0.2998, |
| "step": 907 |
| }, |
| { |
| "epoch": 1.41, |
| "learning_rate": 2.6698014629049112e-05, |
| "loss": 0.2972, |
| "step": 908 |
| }, |
| { |
| "epoch": 1.41, |
| "learning_rate": 2.667189132706374e-05, |
| "loss": 0.307, |
| "step": 909 |
| }, |
| { |
| "epoch": 1.41, |
| "learning_rate": 2.6645768025078373e-05, |
| "loss": 0.3117, |
| "step": 910 |
| }, |
| { |
| "epoch": 1.42, |
| "learning_rate": 2.6619644723093e-05, |
| "loss": 0.3048, |
| "step": 911 |
| }, |
| { |
| "epoch": 1.42, |
| "learning_rate": 2.659352142110763e-05, |
| "loss": 0.3059, |
| "step": 912 |
| }, |
| { |
| "epoch": 1.42, |
| "learning_rate": 2.6567398119122262e-05, |
| "loss": 0.3059, |
| "step": 913 |
| }, |
| { |
| "epoch": 1.42, |
| "learning_rate": 2.6541274817136884e-05, |
| "loss": 0.3026, |
| "step": 914 |
| }, |
| { |
| "epoch": 1.42, |
| "learning_rate": 2.6515151515151516e-05, |
| "loss": 0.3032, |
| "step": 915 |
| }, |
| { |
| "epoch": 1.42, |
| "learning_rate": 2.6489028213166144e-05, |
| "loss": 0.3067, |
| "step": 916 |
| }, |
| { |
| "epoch": 1.42, |
| "learning_rate": 2.6462904911180776e-05, |
| "loss": 0.2985, |
| "step": 917 |
| }, |
| { |
| "epoch": 1.43, |
| "learning_rate": 2.6436781609195405e-05, |
| "loss": 0.2924, |
| "step": 918 |
| }, |
| { |
| "epoch": 1.43, |
| "learning_rate": 2.6410658307210034e-05, |
| "loss": 0.3203, |
| "step": 919 |
| }, |
| { |
| "epoch": 1.43, |
| "learning_rate": 2.6384535005224666e-05, |
| "loss": 0.2902, |
| "step": 920 |
| }, |
| { |
| "epoch": 1.43, |
| "learning_rate": 2.6358411703239287e-05, |
| "loss": 0.3089, |
| "step": 921 |
| }, |
| { |
| "epoch": 1.43, |
| "learning_rate": 2.633228840125392e-05, |
| "loss": 0.315, |
| "step": 922 |
| }, |
| { |
| "epoch": 1.43, |
| "learning_rate": 2.6306165099268548e-05, |
| "loss": 0.3026, |
| "step": 923 |
| }, |
| { |
| "epoch": 1.44, |
| "learning_rate": 2.6280041797283177e-05, |
| "loss": 0.319, |
| "step": 924 |
| }, |
| { |
| "epoch": 1.44, |
| "learning_rate": 2.625391849529781e-05, |
| "loss": 0.2862, |
| "step": 925 |
| }, |
| { |
| "epoch": 1.44, |
| "learning_rate": 2.6227795193312437e-05, |
| "loss": 0.2939, |
| "step": 926 |
| }, |
| { |
| "epoch": 1.44, |
| "learning_rate": 2.620167189132707e-05, |
| "loss": 0.3154, |
| "step": 927 |
| }, |
| { |
| "epoch": 1.44, |
| "learning_rate": 2.617554858934169e-05, |
| "loss": 0.2958, |
| "step": 928 |
| }, |
| { |
| "epoch": 1.44, |
| "learning_rate": 2.6149425287356323e-05, |
| "loss": 0.3058, |
| "step": 929 |
| }, |
| { |
| "epoch": 1.44, |
| "learning_rate": 2.612330198537095e-05, |
| "loss": 0.309, |
| "step": 930 |
| }, |
| { |
| "epoch": 1.45, |
| "learning_rate": 2.609717868338558e-05, |
| "loss": 0.3191, |
| "step": 931 |
| }, |
| { |
| "epoch": 1.45, |
| "learning_rate": 2.6071055381400212e-05, |
| "loss": 0.3025, |
| "step": 932 |
| }, |
| { |
| "epoch": 1.45, |
| "learning_rate": 2.604493207941484e-05, |
| "loss": 0.3159, |
| "step": 933 |
| }, |
| { |
| "epoch": 1.45, |
| "learning_rate": 2.601880877742947e-05, |
| "loss": 0.3236, |
| "step": 934 |
| }, |
| { |
| "epoch": 1.45, |
| "learning_rate": 2.5992685475444094e-05, |
| "loss": 0.3074, |
| "step": 935 |
| }, |
| { |
| "epoch": 1.45, |
| "learning_rate": 2.5966562173458726e-05, |
| "loss": 0.2888, |
| "step": 936 |
| }, |
| { |
| "epoch": 1.46, |
| "learning_rate": 2.5940438871473355e-05, |
| "loss": 0.3175, |
| "step": 937 |
| }, |
| { |
| "epoch": 1.46, |
| "learning_rate": 2.5914315569487984e-05, |
| "loss": 0.306, |
| "step": 938 |
| }, |
| { |
| "epoch": 1.46, |
| "learning_rate": 2.5888192267502616e-05, |
| "loss": 0.3047, |
| "step": 939 |
| }, |
| { |
| "epoch": 1.46, |
| "learning_rate": 2.5862068965517244e-05, |
| "loss": 0.2971, |
| "step": 940 |
| }, |
| { |
| "epoch": 1.46, |
| "learning_rate": 2.5835945663531873e-05, |
| "loss": 0.2996, |
| "step": 941 |
| }, |
| { |
| "epoch": 1.46, |
| "learning_rate": 2.5809822361546498e-05, |
| "loss": 0.3016, |
| "step": 942 |
| }, |
| { |
| "epoch": 1.46, |
| "learning_rate": 2.578369905956113e-05, |
| "loss": 0.3114, |
| "step": 943 |
| }, |
| { |
| "epoch": 1.47, |
| "learning_rate": 2.575757575757576e-05, |
| "loss": 0.2977, |
| "step": 944 |
| }, |
| { |
| "epoch": 1.47, |
| "learning_rate": 2.5731452455590387e-05, |
| "loss": 0.2951, |
| "step": 945 |
| }, |
| { |
| "epoch": 1.47, |
| "learning_rate": 2.570532915360502e-05, |
| "loss": 0.2838, |
| "step": 946 |
| }, |
| { |
| "epoch": 1.47, |
| "learning_rate": 2.5679205851619648e-05, |
| "loss": 0.3033, |
| "step": 947 |
| }, |
| { |
| "epoch": 1.47, |
| "learning_rate": 2.5653082549634276e-05, |
| "loss": 0.2904, |
| "step": 948 |
| }, |
| { |
| "epoch": 1.47, |
| "learning_rate": 2.56269592476489e-05, |
| "loss": 0.3079, |
| "step": 949 |
| }, |
| { |
| "epoch": 1.48, |
| "learning_rate": 2.560083594566353e-05, |
| "loss": 0.2992, |
| "step": 950 |
| }, |
| { |
| "epoch": 1.48, |
| "learning_rate": 2.5574712643678162e-05, |
| "loss": 0.2943, |
| "step": 951 |
| }, |
| { |
| "epoch": 1.48, |
| "learning_rate": 2.554858934169279e-05, |
| "loss": 0.307, |
| "step": 952 |
| }, |
| { |
| "epoch": 1.48, |
| "learning_rate": 2.5522466039707423e-05, |
| "loss": 0.2934, |
| "step": 953 |
| }, |
| { |
| "epoch": 1.48, |
| "learning_rate": 2.549634273772205e-05, |
| "loss": 0.3107, |
| "step": 954 |
| }, |
| { |
| "epoch": 1.48, |
| "learning_rate": 2.547021943573668e-05, |
| "loss": 0.3132, |
| "step": 955 |
| }, |
| { |
| "epoch": 1.49, |
| "learning_rate": 2.5444096133751305e-05, |
| "loss": 0.2967, |
| "step": 956 |
| }, |
| { |
| "epoch": 1.49, |
| "learning_rate": 2.5417972831765934e-05, |
| "loss": 0.2918, |
| "step": 957 |
| }, |
| { |
| "epoch": 1.49, |
| "learning_rate": 2.5391849529780566e-05, |
| "loss": 0.2969, |
| "step": 958 |
| }, |
| { |
| "epoch": 1.49, |
| "learning_rate": 2.5365726227795194e-05, |
| "loss": 0.3009, |
| "step": 959 |
| }, |
| { |
| "epoch": 1.49, |
| "learning_rate": 2.5339602925809823e-05, |
| "loss": 0.2952, |
| "step": 960 |
| }, |
| { |
| "epoch": 1.49, |
| "learning_rate": 2.5313479623824455e-05, |
| "loss": 0.3068, |
| "step": 961 |
| }, |
| { |
| "epoch": 1.49, |
| "learning_rate": 2.5287356321839083e-05, |
| "loss": 0.3071, |
| "step": 962 |
| }, |
| { |
| "epoch": 1.5, |
| "learning_rate": 2.526123301985371e-05, |
| "loss": 0.296, |
| "step": 963 |
| }, |
| { |
| "epoch": 1.5, |
| "learning_rate": 2.5235109717868337e-05, |
| "loss": 0.3138, |
| "step": 964 |
| }, |
| { |
| "epoch": 1.5, |
| "learning_rate": 2.520898641588297e-05, |
| "loss": 0.2981, |
| "step": 965 |
| }, |
| { |
| "epoch": 1.5, |
| "learning_rate": 2.5182863113897598e-05, |
| "loss": 0.2914, |
| "step": 966 |
| }, |
| { |
| "epoch": 1.5, |
| "learning_rate": 2.5156739811912226e-05, |
| "loss": 0.3011, |
| "step": 967 |
| }, |
| { |
| "epoch": 1.5, |
| "learning_rate": 2.5130616509926858e-05, |
| "loss": 0.2956, |
| "step": 968 |
| }, |
| { |
| "epoch": 1.51, |
| "learning_rate": 2.5104493207941487e-05, |
| "loss": 0.3211, |
| "step": 969 |
| }, |
| { |
| "epoch": 1.51, |
| "learning_rate": 2.5078369905956112e-05, |
| "loss": 0.2978, |
| "step": 970 |
| }, |
| { |
| "epoch": 1.51, |
| "learning_rate": 2.505224660397074e-05, |
| "loss": 0.3101, |
| "step": 971 |
| }, |
| { |
| "epoch": 1.51, |
| "learning_rate": 2.5026123301985373e-05, |
| "loss": 0.2988, |
| "step": 972 |
| }, |
| { |
| "epoch": 1.51, |
| "learning_rate": 2.5e-05, |
| "loss": 0.2943, |
| "step": 973 |
| }, |
| { |
| "epoch": 1.51, |
| "learning_rate": 2.497387669801463e-05, |
| "loss": 0.3059, |
| "step": 974 |
| }, |
| { |
| "epoch": 1.51, |
| "learning_rate": 2.494775339602926e-05, |
| "loss": 0.3008, |
| "step": 975 |
| }, |
| { |
| "epoch": 1.52, |
| "learning_rate": 2.4921630094043887e-05, |
| "loss": 0.2943, |
| "step": 976 |
| }, |
| { |
| "epoch": 1.52, |
| "learning_rate": 2.489550679205852e-05, |
| "loss": 0.3149, |
| "step": 977 |
| }, |
| { |
| "epoch": 1.52, |
| "learning_rate": 2.4869383490073148e-05, |
| "loss": 0.309, |
| "step": 978 |
| }, |
| { |
| "epoch": 1.52, |
| "learning_rate": 2.4843260188087776e-05, |
| "loss": 0.313, |
| "step": 979 |
| }, |
| { |
| "epoch": 1.52, |
| "learning_rate": 2.4817136886102405e-05, |
| "loss": 0.3033, |
| "step": 980 |
| }, |
| { |
| "epoch": 1.52, |
| "learning_rate": 2.4791013584117033e-05, |
| "loss": 0.3029, |
| "step": 981 |
| }, |
| { |
| "epoch": 1.53, |
| "learning_rate": 2.4764890282131662e-05, |
| "loss": 0.2873, |
| "step": 982 |
| }, |
| { |
| "epoch": 1.53, |
| "learning_rate": 2.473876698014629e-05, |
| "loss": 0.3007, |
| "step": 983 |
| }, |
| { |
| "epoch": 1.53, |
| "learning_rate": 2.4712643678160922e-05, |
| "loss": 0.295, |
| "step": 984 |
| }, |
| { |
| "epoch": 1.53, |
| "learning_rate": 2.468652037617555e-05, |
| "loss": 0.3161, |
| "step": 985 |
| }, |
| { |
| "epoch": 1.53, |
| "learning_rate": 2.466039707419018e-05, |
| "loss": 0.3091, |
| "step": 986 |
| }, |
| { |
| "epoch": 1.53, |
| "learning_rate": 2.4634273772204808e-05, |
| "loss": 0.292, |
| "step": 987 |
| }, |
| { |
| "epoch": 1.53, |
| "learning_rate": 2.4608150470219437e-05, |
| "loss": 0.2951, |
| "step": 988 |
| }, |
| { |
| "epoch": 1.54, |
| "learning_rate": 2.4582027168234065e-05, |
| "loss": 0.3001, |
| "step": 989 |
| }, |
| { |
| "epoch": 1.54, |
| "learning_rate": 2.4555903866248694e-05, |
| "loss": 0.3036, |
| "step": 990 |
| }, |
| { |
| "epoch": 1.54, |
| "learning_rate": 2.4529780564263326e-05, |
| "loss": 0.2927, |
| "step": 991 |
| }, |
| { |
| "epoch": 1.54, |
| "learning_rate": 2.4503657262277955e-05, |
| "loss": 0.3015, |
| "step": 992 |
| }, |
| { |
| "epoch": 1.54, |
| "learning_rate": 2.447753396029258e-05, |
| "loss": 0.3002, |
| "step": 993 |
| }, |
| { |
| "epoch": 1.54, |
| "learning_rate": 2.4451410658307212e-05, |
| "loss": 0.3109, |
| "step": 994 |
| }, |
| { |
| "epoch": 1.55, |
| "learning_rate": 2.442528735632184e-05, |
| "loss": 0.2884, |
| "step": 995 |
| }, |
| { |
| "epoch": 1.55, |
| "learning_rate": 2.439916405433647e-05, |
| "loss": 0.304, |
| "step": 996 |
| }, |
| { |
| "epoch": 1.55, |
| "learning_rate": 2.4373040752351098e-05, |
| "loss": 0.2967, |
| "step": 997 |
| }, |
| { |
| "epoch": 1.55, |
| "learning_rate": 2.4346917450365726e-05, |
| "loss": 0.3039, |
| "step": 998 |
| }, |
| { |
| "epoch": 1.55, |
| "learning_rate": 2.4320794148380358e-05, |
| "loss": 0.3175, |
| "step": 999 |
| }, |
| { |
| "epoch": 1.55, |
| "learning_rate": 2.4294670846394983e-05, |
| "loss": 0.3153, |
| "step": 1000 |
| }, |
| { |
| "epoch": 1.56, |
| "learning_rate": 2.4268547544409615e-05, |
| "loss": 0.2949, |
| "step": 1001 |
| }, |
| { |
| "epoch": 1.56, |
| "learning_rate": 2.4242424242424244e-05, |
| "loss": 0.3183, |
| "step": 1002 |
| }, |
| { |
| "epoch": 1.56, |
| "learning_rate": 2.4216300940438872e-05, |
| "loss": 0.32, |
| "step": 1003 |
| }, |
| { |
| "epoch": 1.56, |
| "learning_rate": 2.41901776384535e-05, |
| "loss": 0.2968, |
| "step": 1004 |
| }, |
| { |
| "epoch": 1.56, |
| "learning_rate": 2.416405433646813e-05, |
| "loss": 0.2879, |
| "step": 1005 |
| }, |
| { |
| "epoch": 1.56, |
| "learning_rate": 2.413793103448276e-05, |
| "loss": 0.2946, |
| "step": 1006 |
| }, |
| { |
| "epoch": 1.56, |
| "learning_rate": 2.4111807732497387e-05, |
| "loss": 0.3054, |
| "step": 1007 |
| }, |
| { |
| "epoch": 1.57, |
| "learning_rate": 2.408568443051202e-05, |
| "loss": 0.3286, |
| "step": 1008 |
| }, |
| { |
| "epoch": 1.57, |
| "learning_rate": 2.4059561128526647e-05, |
| "loss": 0.31, |
| "step": 1009 |
| }, |
| { |
| "epoch": 1.57, |
| "learning_rate": 2.4033437826541276e-05, |
| "loss": 0.3074, |
| "step": 1010 |
| }, |
| { |
| "epoch": 1.57, |
| "learning_rate": 2.4007314524555905e-05, |
| "loss": 0.2954, |
| "step": 1011 |
| }, |
| { |
| "epoch": 1.57, |
| "learning_rate": 2.3981191222570533e-05, |
| "loss": 0.3082, |
| "step": 1012 |
| }, |
| { |
| "epoch": 1.57, |
| "learning_rate": 2.3955067920585165e-05, |
| "loss": 0.283, |
| "step": 1013 |
| }, |
| { |
| "epoch": 1.58, |
| "learning_rate": 2.392894461859979e-05, |
| "loss": 0.3066, |
| "step": 1014 |
| }, |
| { |
| "epoch": 1.58, |
| "learning_rate": 2.3902821316614422e-05, |
| "loss": 0.324, |
| "step": 1015 |
| }, |
| { |
| "epoch": 1.58, |
| "learning_rate": 2.387669801462905e-05, |
| "loss": 0.3016, |
| "step": 1016 |
| }, |
| { |
| "epoch": 1.58, |
| "learning_rate": 2.385057471264368e-05, |
| "loss": 0.2761, |
| "step": 1017 |
| }, |
| { |
| "epoch": 1.58, |
| "learning_rate": 2.3824451410658308e-05, |
| "loss": 0.3021, |
| "step": 1018 |
| }, |
| { |
| "epoch": 1.58, |
| "learning_rate": 2.3798328108672937e-05, |
| "loss": 0.2868, |
| "step": 1019 |
| }, |
| { |
| "epoch": 1.58, |
| "learning_rate": 2.377220480668757e-05, |
| "loss": 0.3012, |
| "step": 1020 |
| }, |
| { |
| "epoch": 1.59, |
| "learning_rate": 2.3746081504702194e-05, |
| "loss": 0.3137, |
| "step": 1021 |
| }, |
| { |
| "epoch": 1.59, |
| "learning_rate": 2.3719958202716826e-05, |
| "loss": 0.3012, |
| "step": 1022 |
| }, |
| { |
| "epoch": 1.59, |
| "learning_rate": 2.3693834900731454e-05, |
| "loss": 0.3095, |
| "step": 1023 |
| }, |
| { |
| "epoch": 1.59, |
| "learning_rate": 2.3667711598746083e-05, |
| "loss": 0.3009, |
| "step": 1024 |
| }, |
| { |
| "epoch": 1.59, |
| "learning_rate": 2.364158829676071e-05, |
| "loss": 0.3116, |
| "step": 1025 |
| }, |
| { |
| "epoch": 1.59, |
| "learning_rate": 2.361546499477534e-05, |
| "loss": 0.2969, |
| "step": 1026 |
| }, |
| { |
| "epoch": 1.6, |
| "learning_rate": 2.3589341692789972e-05, |
| "loss": 0.3208, |
| "step": 1027 |
| }, |
| { |
| "epoch": 1.6, |
| "learning_rate": 2.3563218390804597e-05, |
| "loss": 0.288, |
| "step": 1028 |
| }, |
| { |
| "epoch": 1.6, |
| "learning_rate": 2.3537095088819226e-05, |
| "loss": 0.3061, |
| "step": 1029 |
| }, |
| { |
| "epoch": 1.6, |
| "learning_rate": 2.3510971786833858e-05, |
| "loss": 0.3105, |
| "step": 1030 |
| }, |
| { |
| "epoch": 1.6, |
| "learning_rate": 2.3484848484848487e-05, |
| "loss": 0.3001, |
| "step": 1031 |
| }, |
| { |
| "epoch": 1.6, |
| "learning_rate": 2.3458725182863115e-05, |
| "loss": 0.2875, |
| "step": 1032 |
| }, |
| { |
| "epoch": 1.6, |
| "learning_rate": 2.3432601880877744e-05, |
| "loss": 0.3078, |
| "step": 1033 |
| }, |
| { |
| "epoch": 1.61, |
| "learning_rate": 2.3406478578892372e-05, |
| "loss": 0.2993, |
| "step": 1034 |
| }, |
| { |
| "epoch": 1.61, |
| "learning_rate": 2.3380355276907e-05, |
| "loss": 0.3158, |
| "step": 1035 |
| }, |
| { |
| "epoch": 1.61, |
| "learning_rate": 2.335423197492163e-05, |
| "loss": 0.31, |
| "step": 1036 |
| }, |
| { |
| "epoch": 1.61, |
| "learning_rate": 2.332810867293626e-05, |
| "loss": 0.282, |
| "step": 1037 |
| }, |
| { |
| "epoch": 1.61, |
| "learning_rate": 2.330198537095089e-05, |
| "loss": 0.2948, |
| "step": 1038 |
| }, |
| { |
| "epoch": 1.61, |
| "learning_rate": 2.327586206896552e-05, |
| "loss": 0.3005, |
| "step": 1039 |
| }, |
| { |
| "epoch": 1.62, |
| "learning_rate": 2.3249738766980147e-05, |
| "loss": 0.2877, |
| "step": 1040 |
| }, |
| { |
| "epoch": 1.62, |
| "learning_rate": 2.3223615464994776e-05, |
| "loss": 0.29, |
| "step": 1041 |
| }, |
| { |
| "epoch": 1.62, |
| "learning_rate": 2.3197492163009404e-05, |
| "loss": 0.2896, |
| "step": 1042 |
| }, |
| { |
| "epoch": 1.62, |
| "learning_rate": 2.3171368861024033e-05, |
| "loss": 0.3089, |
| "step": 1043 |
| }, |
| { |
| "epoch": 1.62, |
| "learning_rate": 2.3145245559038665e-05, |
| "loss": 0.2932, |
| "step": 1044 |
| }, |
| { |
| "epoch": 1.62, |
| "learning_rate": 2.3119122257053294e-05, |
| "loss": 0.2843, |
| "step": 1045 |
| }, |
| { |
| "epoch": 1.62, |
| "learning_rate": 2.3092998955067922e-05, |
| "loss": 0.2993, |
| "step": 1046 |
| }, |
| { |
| "epoch": 1.63, |
| "learning_rate": 2.306687565308255e-05, |
| "loss": 0.309, |
| "step": 1047 |
| }, |
| { |
| "epoch": 1.63, |
| "learning_rate": 2.304075235109718e-05, |
| "loss": 0.2999, |
| "step": 1048 |
| }, |
| { |
| "epoch": 1.63, |
| "learning_rate": 2.3014629049111808e-05, |
| "loss": 0.3064, |
| "step": 1049 |
| }, |
| { |
| "epoch": 1.63, |
| "learning_rate": 2.2988505747126437e-05, |
| "loss": 0.3027, |
| "step": 1050 |
| }, |
| { |
| "epoch": 1.63, |
| "learning_rate": 2.296238244514107e-05, |
| "loss": 0.2888, |
| "step": 1051 |
| }, |
| { |
| "epoch": 1.63, |
| "learning_rate": 2.2936259143155697e-05, |
| "loss": 0.3187, |
| "step": 1052 |
| }, |
| { |
| "epoch": 1.64, |
| "learning_rate": 2.2910135841170326e-05, |
| "loss": 0.2936, |
| "step": 1053 |
| }, |
| { |
| "epoch": 1.64, |
| "learning_rate": 2.2884012539184954e-05, |
| "loss": 0.2989, |
| "step": 1054 |
| }, |
| { |
| "epoch": 1.64, |
| "learning_rate": 2.2857889237199583e-05, |
| "loss": 0.2819, |
| "step": 1055 |
| }, |
| { |
| "epoch": 1.64, |
| "learning_rate": 2.283176593521421e-05, |
| "loss": 0.3097, |
| "step": 1056 |
| }, |
| { |
| "epoch": 1.64, |
| "learning_rate": 2.280564263322884e-05, |
| "loss": 0.307, |
| "step": 1057 |
| }, |
| { |
| "epoch": 1.64, |
| "learning_rate": 2.2779519331243472e-05, |
| "loss": 0.2995, |
| "step": 1058 |
| }, |
| { |
| "epoch": 1.65, |
| "learning_rate": 2.27533960292581e-05, |
| "loss": 0.3152, |
| "step": 1059 |
| }, |
| { |
| "epoch": 1.65, |
| "learning_rate": 2.272727272727273e-05, |
| "loss": 0.3125, |
| "step": 1060 |
| }, |
| { |
| "epoch": 1.65, |
| "learning_rate": 2.2701149425287358e-05, |
| "loss": 0.2977, |
| "step": 1061 |
| }, |
| { |
| "epoch": 1.65, |
| "learning_rate": 2.2675026123301986e-05, |
| "loss": 0.3028, |
| "step": 1062 |
| }, |
| { |
| "epoch": 1.65, |
| "learning_rate": 2.2648902821316615e-05, |
| "loss": 0.3049, |
| "step": 1063 |
| }, |
| { |
| "epoch": 1.65, |
| "learning_rate": 2.2622779519331244e-05, |
| "loss": 0.2889, |
| "step": 1064 |
| }, |
| { |
| "epoch": 1.65, |
| "learning_rate": 2.2596656217345876e-05, |
| "loss": 0.3105, |
| "step": 1065 |
| }, |
| { |
| "epoch": 1.66, |
| "learning_rate": 2.2570532915360504e-05, |
| "loss": 0.3226, |
| "step": 1066 |
| }, |
| { |
| "epoch": 1.66, |
| "learning_rate": 2.254440961337513e-05, |
| "loss": 0.2902, |
| "step": 1067 |
| }, |
| { |
| "epoch": 1.66, |
| "learning_rate": 2.251828631138976e-05, |
| "loss": 0.2993, |
| "step": 1068 |
| }, |
| { |
| "epoch": 1.66, |
| "learning_rate": 2.249216300940439e-05, |
| "loss": 0.2975, |
| "step": 1069 |
| }, |
| { |
| "epoch": 1.66, |
| "learning_rate": 2.246603970741902e-05, |
| "loss": 0.3138, |
| "step": 1070 |
| }, |
| { |
| "epoch": 1.66, |
| "learning_rate": 2.2439916405433647e-05, |
| "loss": 0.3002, |
| "step": 1071 |
| }, |
| { |
| "epoch": 1.67, |
| "learning_rate": 2.2413793103448276e-05, |
| "loss": 0.3088, |
| "step": 1072 |
| }, |
| { |
| "epoch": 1.67, |
| "learning_rate": 2.2387669801462908e-05, |
| "loss": 0.2872, |
| "step": 1073 |
| }, |
| { |
| "epoch": 1.67, |
| "learning_rate": 2.2361546499477533e-05, |
| "loss": 0.3016, |
| "step": 1074 |
| }, |
| { |
| "epoch": 1.67, |
| "learning_rate": 2.2335423197492165e-05, |
| "loss": 0.2942, |
| "step": 1075 |
| }, |
| { |
| "epoch": 1.67, |
| "learning_rate": 2.2309299895506794e-05, |
| "loss": 0.2877, |
| "step": 1076 |
| }, |
| { |
| "epoch": 1.67, |
| "learning_rate": 2.2283176593521422e-05, |
| "loss": 0.298, |
| "step": 1077 |
| }, |
| { |
| "epoch": 1.67, |
| "learning_rate": 2.225705329153605e-05, |
| "loss": 0.2979, |
| "step": 1078 |
| }, |
| { |
| "epoch": 1.68, |
| "learning_rate": 2.223092998955068e-05, |
| "loss": 0.293, |
| "step": 1079 |
| }, |
| { |
| "epoch": 1.68, |
| "learning_rate": 2.220480668756531e-05, |
| "loss": 0.3027, |
| "step": 1080 |
| }, |
| { |
| "epoch": 1.68, |
| "learning_rate": 2.2178683385579936e-05, |
| "loss": 0.2898, |
| "step": 1081 |
| }, |
| { |
| "epoch": 1.68, |
| "learning_rate": 2.215256008359457e-05, |
| "loss": 0.3105, |
| "step": 1082 |
| }, |
| { |
| "epoch": 1.68, |
| "learning_rate": 2.2126436781609197e-05, |
| "loss": 0.3074, |
| "step": 1083 |
| }, |
| { |
| "epoch": 1.68, |
| "learning_rate": 2.2100313479623826e-05, |
| "loss": 0.318, |
| "step": 1084 |
| }, |
| { |
| "epoch": 1.69, |
| "learning_rate": 2.2074190177638454e-05, |
| "loss": 0.2947, |
| "step": 1085 |
| }, |
| { |
| "epoch": 1.69, |
| "learning_rate": 2.2048066875653083e-05, |
| "loss": 0.2844, |
| "step": 1086 |
| }, |
| { |
| "epoch": 1.69, |
| "learning_rate": 2.2021943573667715e-05, |
| "loss": 0.2827, |
| "step": 1087 |
| }, |
| { |
| "epoch": 1.69, |
| "learning_rate": 2.199582027168234e-05, |
| "loss": 0.3019, |
| "step": 1088 |
| }, |
| { |
| "epoch": 1.69, |
| "learning_rate": 2.1969696969696972e-05, |
| "loss": 0.2963, |
| "step": 1089 |
| }, |
| { |
| "epoch": 1.69, |
| "learning_rate": 2.19435736677116e-05, |
| "loss": 0.2956, |
| "step": 1090 |
| }, |
| { |
| "epoch": 1.69, |
| "learning_rate": 2.191745036572623e-05, |
| "loss": 0.3107, |
| "step": 1091 |
| }, |
| { |
| "epoch": 1.7, |
| "learning_rate": 2.1891327063740858e-05, |
| "loss": 0.3084, |
| "step": 1092 |
| }, |
| { |
| "epoch": 1.7, |
| "learning_rate": 2.1865203761755486e-05, |
| "loss": 0.2946, |
| "step": 1093 |
| }, |
| { |
| "epoch": 1.7, |
| "learning_rate": 2.183908045977012e-05, |
| "loss": 0.2818, |
| "step": 1094 |
| }, |
| { |
| "epoch": 1.7, |
| "learning_rate": 2.1812957157784744e-05, |
| "loss": 0.2888, |
| "step": 1095 |
| }, |
| { |
| "epoch": 1.7, |
| "learning_rate": 2.1786833855799376e-05, |
| "loss": 0.2906, |
| "step": 1096 |
| }, |
| { |
| "epoch": 1.7, |
| "learning_rate": 2.1760710553814004e-05, |
| "loss": 0.2744, |
| "step": 1097 |
| }, |
| { |
| "epoch": 1.71, |
| "learning_rate": 2.173458725182863e-05, |
| "loss": 0.3016, |
| "step": 1098 |
| }, |
| { |
| "epoch": 1.71, |
| "learning_rate": 2.170846394984326e-05, |
| "loss": 0.3071, |
| "step": 1099 |
| }, |
| { |
| "epoch": 1.71, |
| "learning_rate": 2.168234064785789e-05, |
| "loss": 0.29, |
| "step": 1100 |
| }, |
| { |
| "epoch": 1.71, |
| "learning_rate": 2.1656217345872522e-05, |
| "loss": 0.3144, |
| "step": 1101 |
| }, |
| { |
| "epoch": 1.71, |
| "learning_rate": 2.1630094043887147e-05, |
| "loss": 0.2963, |
| "step": 1102 |
| }, |
| { |
| "epoch": 1.71, |
| "learning_rate": 2.1603970741901776e-05, |
| "loss": 0.3253, |
| "step": 1103 |
| }, |
| { |
| "epoch": 1.72, |
| "learning_rate": 2.1577847439916408e-05, |
| "loss": 0.3035, |
| "step": 1104 |
| }, |
| { |
| "epoch": 1.72, |
| "learning_rate": 2.1551724137931033e-05, |
| "loss": 0.3043, |
| "step": 1105 |
| }, |
| { |
| "epoch": 1.72, |
| "learning_rate": 2.1525600835945665e-05, |
| "loss": 0.2835, |
| "step": 1106 |
| }, |
| { |
| "epoch": 1.72, |
| "learning_rate": 2.1499477533960293e-05, |
| "loss": 0.3371, |
| "step": 1107 |
| }, |
| { |
| "epoch": 1.72, |
| "learning_rate": 2.1473354231974925e-05, |
| "loss": 0.3075, |
| "step": 1108 |
| }, |
| { |
| "epoch": 1.72, |
| "learning_rate": 2.144723092998955e-05, |
| "loss": 0.302, |
| "step": 1109 |
| }, |
| { |
| "epoch": 1.72, |
| "learning_rate": 2.142110762800418e-05, |
| "loss": 0.3133, |
| "step": 1110 |
| }, |
| { |
| "epoch": 1.73, |
| "learning_rate": 2.139498432601881e-05, |
| "loss": 0.3097, |
| "step": 1111 |
| }, |
| { |
| "epoch": 1.73, |
| "learning_rate": 2.1368861024033436e-05, |
| "loss": 0.2821, |
| "step": 1112 |
| }, |
| { |
| "epoch": 1.73, |
| "learning_rate": 2.134273772204807e-05, |
| "loss": 0.2946, |
| "step": 1113 |
| }, |
| { |
| "epoch": 1.73, |
| "learning_rate": 2.1316614420062697e-05, |
| "loss": 0.3052, |
| "step": 1114 |
| }, |
| { |
| "epoch": 1.73, |
| "learning_rate": 2.1290491118077326e-05, |
| "loss": 0.2806, |
| "step": 1115 |
| }, |
| { |
| "epoch": 1.73, |
| "learning_rate": 2.1264367816091954e-05, |
| "loss": 0.2997, |
| "step": 1116 |
| }, |
| { |
| "epoch": 1.74, |
| "learning_rate": 2.1238244514106583e-05, |
| "loss": 0.2987, |
| "step": 1117 |
| }, |
| { |
| "epoch": 1.74, |
| "learning_rate": 2.1212121212121215e-05, |
| "loss": 0.3012, |
| "step": 1118 |
| }, |
| { |
| "epoch": 1.74, |
| "learning_rate": 2.1185997910135843e-05, |
| "loss": 0.3052, |
| "step": 1119 |
| }, |
| { |
| "epoch": 1.74, |
| "learning_rate": 2.1159874608150472e-05, |
| "loss": 0.2898, |
| "step": 1120 |
| }, |
| { |
| "epoch": 1.74, |
| "learning_rate": 2.11337513061651e-05, |
| "loss": 0.2958, |
| "step": 1121 |
| }, |
| { |
| "epoch": 1.74, |
| "learning_rate": 2.110762800417973e-05, |
| "loss": 0.2831, |
| "step": 1122 |
| }, |
| { |
| "epoch": 1.74, |
| "learning_rate": 2.1081504702194358e-05, |
| "loss": 0.3038, |
| "step": 1123 |
| }, |
| { |
| "epoch": 1.75, |
| "learning_rate": 2.1055381400208986e-05, |
| "loss": 0.3024, |
| "step": 1124 |
| }, |
| { |
| "epoch": 1.75, |
| "learning_rate": 2.1029258098223618e-05, |
| "loss": 0.3034, |
| "step": 1125 |
| }, |
| { |
| "epoch": 1.75, |
| "learning_rate": 2.1003134796238247e-05, |
| "loss": 0.2966, |
| "step": 1126 |
| }, |
| { |
| "epoch": 1.75, |
| "learning_rate": 2.0977011494252875e-05, |
| "loss": 0.2984, |
| "step": 1127 |
| }, |
| { |
| "epoch": 1.75, |
| "learning_rate": 2.0950888192267504e-05, |
| "loss": 0.3023, |
| "step": 1128 |
| }, |
| { |
| "epoch": 1.75, |
| "learning_rate": 2.0924764890282133e-05, |
| "loss": 0.3009, |
| "step": 1129 |
| }, |
| { |
| "epoch": 1.76, |
| "learning_rate": 2.089864158829676e-05, |
| "loss": 0.2901, |
| "step": 1130 |
| }, |
| { |
| "epoch": 1.76, |
| "learning_rate": 2.087251828631139e-05, |
| "loss": 0.2996, |
| "step": 1131 |
| }, |
| { |
| "epoch": 1.76, |
| "learning_rate": 2.0846394984326022e-05, |
| "loss": 0.3031, |
| "step": 1132 |
| }, |
| { |
| "epoch": 1.76, |
| "learning_rate": 2.082027168234065e-05, |
| "loss": 0.2943, |
| "step": 1133 |
| }, |
| { |
| "epoch": 1.76, |
| "learning_rate": 2.079414838035528e-05, |
| "loss": 0.2994, |
| "step": 1134 |
| }, |
| { |
| "epoch": 1.76, |
| "learning_rate": 2.0768025078369908e-05, |
| "loss": 0.2969, |
| "step": 1135 |
| }, |
| { |
| "epoch": 1.76, |
| "learning_rate": 2.0741901776384536e-05, |
| "loss": 0.2912, |
| "step": 1136 |
| }, |
| { |
| "epoch": 1.77, |
| "learning_rate": 2.0715778474399165e-05, |
| "loss": 0.2878, |
| "step": 1137 |
| }, |
| { |
| "epoch": 1.77, |
| "learning_rate": 2.0689655172413793e-05, |
| "loss": 0.2879, |
| "step": 1138 |
| }, |
| { |
| "epoch": 1.77, |
| "learning_rate": 2.0663531870428425e-05, |
| "loss": 0.299, |
| "step": 1139 |
| }, |
| { |
| "epoch": 1.77, |
| "learning_rate": 2.0637408568443054e-05, |
| "loss": 0.3194, |
| "step": 1140 |
| }, |
| { |
| "epoch": 1.77, |
| "learning_rate": 2.061128526645768e-05, |
| "loss": 0.2826, |
| "step": 1141 |
| }, |
| { |
| "epoch": 1.77, |
| "learning_rate": 2.058516196447231e-05, |
| "loss": 0.2906, |
| "step": 1142 |
| }, |
| { |
| "epoch": 1.78, |
| "learning_rate": 2.055903866248694e-05, |
| "loss": 0.3135, |
| "step": 1143 |
| }, |
| { |
| "epoch": 1.78, |
| "learning_rate": 2.0532915360501568e-05, |
| "loss": 0.2977, |
| "step": 1144 |
| }, |
| { |
| "epoch": 1.78, |
| "learning_rate": 2.0506792058516197e-05, |
| "loss": 0.2987, |
| "step": 1145 |
| }, |
| { |
| "epoch": 1.78, |
| "learning_rate": 2.0480668756530825e-05, |
| "loss": 0.2798, |
| "step": 1146 |
| }, |
| { |
| "epoch": 1.78, |
| "learning_rate": 2.0454545454545457e-05, |
| "loss": 0.3049, |
| "step": 1147 |
| }, |
| { |
| "epoch": 1.78, |
| "learning_rate": 2.0428422152560083e-05, |
| "loss": 0.2589, |
| "step": 1148 |
| }, |
| { |
| "epoch": 1.78, |
| "learning_rate": 2.0402298850574715e-05, |
| "loss": 0.2938, |
| "step": 1149 |
| }, |
| { |
| "epoch": 1.79, |
| "learning_rate": 2.0376175548589343e-05, |
| "loss": 0.2822, |
| "step": 1150 |
| }, |
| { |
| "epoch": 1.79, |
| "learning_rate": 2.0350052246603972e-05, |
| "loss": 0.3124, |
| "step": 1151 |
| }, |
| { |
| "epoch": 1.79, |
| "learning_rate": 2.03239289446186e-05, |
| "loss": 0.2776, |
| "step": 1152 |
| }, |
| { |
| "epoch": 1.79, |
| "learning_rate": 2.029780564263323e-05, |
| "loss": 0.2975, |
| "step": 1153 |
| }, |
| { |
| "epoch": 1.79, |
| "learning_rate": 2.027168234064786e-05, |
| "loss": 0.2749, |
| "step": 1154 |
| }, |
| { |
| "epoch": 1.79, |
| "learning_rate": 2.0245559038662486e-05, |
| "loss": 0.2965, |
| "step": 1155 |
| }, |
| { |
| "epoch": 1.8, |
| "learning_rate": 2.0219435736677118e-05, |
| "loss": 0.3006, |
| "step": 1156 |
| }, |
| { |
| "epoch": 1.8, |
| "learning_rate": 2.0193312434691747e-05, |
| "loss": 0.2799, |
| "step": 1157 |
| }, |
| { |
| "epoch": 1.8, |
| "learning_rate": 2.0167189132706375e-05, |
| "loss": 0.2949, |
| "step": 1158 |
| }, |
| { |
| "epoch": 1.8, |
| "learning_rate": 2.0141065830721004e-05, |
| "loss": 0.3203, |
| "step": 1159 |
| }, |
| { |
| "epoch": 1.8, |
| "learning_rate": 2.0114942528735632e-05, |
| "loss": 0.3118, |
| "step": 1160 |
| }, |
| { |
| "epoch": 1.8, |
| "learning_rate": 2.0088819226750264e-05, |
| "loss": 0.2961, |
| "step": 1161 |
| }, |
| { |
| "epoch": 1.81, |
| "learning_rate": 2.006269592476489e-05, |
| "loss": 0.2903, |
| "step": 1162 |
| }, |
| { |
| "epoch": 1.81, |
| "learning_rate": 2.003657262277952e-05, |
| "loss": 0.3005, |
| "step": 1163 |
| }, |
| { |
| "epoch": 1.81, |
| "learning_rate": 2.001044932079415e-05, |
| "loss": 0.2877, |
| "step": 1164 |
| }, |
| { |
| "epoch": 1.81, |
| "learning_rate": 1.998432601880878e-05, |
| "loss": 0.3087, |
| "step": 1165 |
| }, |
| { |
| "epoch": 1.81, |
| "learning_rate": 1.9958202716823407e-05, |
| "loss": 0.2988, |
| "step": 1166 |
| }, |
| { |
| "epoch": 1.81, |
| "learning_rate": 1.9932079414838036e-05, |
| "loss": 0.3196, |
| "step": 1167 |
| }, |
| { |
| "epoch": 1.81, |
| "learning_rate": 1.9905956112852668e-05, |
| "loss": 0.295, |
| "step": 1168 |
| }, |
| { |
| "epoch": 1.82, |
| "learning_rate": 1.9879832810867293e-05, |
| "loss": 0.3017, |
| "step": 1169 |
| }, |
| { |
| "epoch": 1.82, |
| "learning_rate": 1.9853709508881925e-05, |
| "loss": 0.289, |
| "step": 1170 |
| }, |
| { |
| "epoch": 1.82, |
| "learning_rate": 1.9827586206896554e-05, |
| "loss": 0.2861, |
| "step": 1171 |
| }, |
| { |
| "epoch": 1.82, |
| "learning_rate": 1.980146290491118e-05, |
| "loss": 0.2992, |
| "step": 1172 |
| }, |
| { |
| "epoch": 1.82, |
| "learning_rate": 1.977533960292581e-05, |
| "loss": 0.2994, |
| "step": 1173 |
| }, |
| { |
| "epoch": 1.82, |
| "learning_rate": 1.974921630094044e-05, |
| "loss": 0.3021, |
| "step": 1174 |
| }, |
| { |
| "epoch": 1.83, |
| "learning_rate": 1.972309299895507e-05, |
| "loss": 0.2931, |
| "step": 1175 |
| }, |
| { |
| "epoch": 1.83, |
| "learning_rate": 1.9696969696969697e-05, |
| "loss": 0.2923, |
| "step": 1176 |
| }, |
| { |
| "epoch": 1.83, |
| "learning_rate": 1.9670846394984325e-05, |
| "loss": 0.3041, |
| "step": 1177 |
| }, |
| { |
| "epoch": 1.83, |
| "learning_rate": 1.9644723092998957e-05, |
| "loss": 0.296, |
| "step": 1178 |
| }, |
| { |
| "epoch": 1.83, |
| "learning_rate": 1.9618599791013582e-05, |
| "loss": 0.2931, |
| "step": 1179 |
| }, |
| { |
| "epoch": 1.83, |
| "learning_rate": 1.9592476489028214e-05, |
| "loss": 0.3002, |
| "step": 1180 |
| }, |
| { |
| "epoch": 1.83, |
| "learning_rate": 1.9566353187042843e-05, |
| "loss": 0.3022, |
| "step": 1181 |
| }, |
| { |
| "epoch": 1.84, |
| "learning_rate": 1.9540229885057475e-05, |
| "loss": 0.2882, |
| "step": 1182 |
| }, |
| { |
| "epoch": 1.84, |
| "learning_rate": 1.95141065830721e-05, |
| "loss": 0.2978, |
| "step": 1183 |
| }, |
| { |
| "epoch": 1.84, |
| "learning_rate": 1.948798328108673e-05, |
| "loss": 0.2838, |
| "step": 1184 |
| }, |
| { |
| "epoch": 1.84, |
| "learning_rate": 1.946185997910136e-05, |
| "loss": 0.2974, |
| "step": 1185 |
| }, |
| { |
| "epoch": 1.84, |
| "learning_rate": 1.9435736677115986e-05, |
| "loss": 0.2831, |
| "step": 1186 |
| }, |
| { |
| "epoch": 1.84, |
| "learning_rate": 1.9409613375130618e-05, |
| "loss": 0.2894, |
| "step": 1187 |
| }, |
| { |
| "epoch": 1.85, |
| "learning_rate": 1.9383490073145247e-05, |
| "loss": 0.2948, |
| "step": 1188 |
| }, |
| { |
| "epoch": 1.85, |
| "learning_rate": 1.9357366771159875e-05, |
| "loss": 0.3034, |
| "step": 1189 |
| }, |
| { |
| "epoch": 1.85, |
| "learning_rate": 1.9331243469174504e-05, |
| "loss": 0.292, |
| "step": 1190 |
| }, |
| { |
| "epoch": 1.85, |
| "learning_rate": 1.9305120167189132e-05, |
| "loss": 0.2897, |
| "step": 1191 |
| }, |
| { |
| "epoch": 1.85, |
| "learning_rate": 1.9278996865203764e-05, |
| "loss": 0.3007, |
| "step": 1192 |
| }, |
| { |
| "epoch": 1.85, |
| "learning_rate": 1.925287356321839e-05, |
| "loss": 0.2984, |
| "step": 1193 |
| }, |
| { |
| "epoch": 1.85, |
| "learning_rate": 1.922675026123302e-05, |
| "loss": 0.2991, |
| "step": 1194 |
| }, |
| { |
| "epoch": 1.86, |
| "learning_rate": 1.920062695924765e-05, |
| "loss": 0.2872, |
| "step": 1195 |
| }, |
| { |
| "epoch": 1.86, |
| "learning_rate": 1.917450365726228e-05, |
| "loss": 0.2815, |
| "step": 1196 |
| }, |
| { |
| "epoch": 1.86, |
| "learning_rate": 1.9148380355276907e-05, |
| "loss": 0.3028, |
| "step": 1197 |
| }, |
| { |
| "epoch": 1.86, |
| "learning_rate": 1.9122257053291536e-05, |
| "loss": 0.3101, |
| "step": 1198 |
| }, |
| { |
| "epoch": 1.86, |
| "learning_rate": 1.9096133751306168e-05, |
| "loss": 0.3093, |
| "step": 1199 |
| }, |
| { |
| "epoch": 1.86, |
| "learning_rate": 1.9070010449320793e-05, |
| "loss": 0.2908, |
| "step": 1200 |
| }, |
| { |
| "epoch": 1.87, |
| "learning_rate": 1.9043887147335425e-05, |
| "loss": 0.3026, |
| "step": 1201 |
| }, |
| { |
| "epoch": 1.87, |
| "learning_rate": 1.9017763845350054e-05, |
| "loss": 0.3029, |
| "step": 1202 |
| }, |
| { |
| "epoch": 1.87, |
| "learning_rate": 1.8991640543364682e-05, |
| "loss": 0.2867, |
| "step": 1203 |
| }, |
| { |
| "epoch": 1.87, |
| "learning_rate": 1.896551724137931e-05, |
| "loss": 0.307, |
| "step": 1204 |
| }, |
| { |
| "epoch": 1.87, |
| "learning_rate": 1.893939393939394e-05, |
| "loss": 0.2957, |
| "step": 1205 |
| }, |
| { |
| "epoch": 1.87, |
| "learning_rate": 1.891327063740857e-05, |
| "loss": 0.306, |
| "step": 1206 |
| }, |
| { |
| "epoch": 1.88, |
| "learning_rate": 1.8887147335423197e-05, |
| "loss": 0.2803, |
| "step": 1207 |
| }, |
| { |
| "epoch": 1.88, |
| "learning_rate": 1.886102403343783e-05, |
| "loss": 0.2757, |
| "step": 1208 |
| }, |
| { |
| "epoch": 1.88, |
| "learning_rate": 1.8834900731452457e-05, |
| "loss": 0.2924, |
| "step": 1209 |
| }, |
| { |
| "epoch": 1.88, |
| "learning_rate": 1.8808777429467086e-05, |
| "loss": 0.3036, |
| "step": 1210 |
| }, |
| { |
| "epoch": 1.88, |
| "learning_rate": 1.8782654127481714e-05, |
| "loss": 0.3163, |
| "step": 1211 |
| }, |
| { |
| "epoch": 1.88, |
| "learning_rate": 1.8756530825496343e-05, |
| "loss": 0.2884, |
| "step": 1212 |
| }, |
| { |
| "epoch": 1.89, |
| "learning_rate": 1.8730407523510975e-05, |
| "loss": 0.3114, |
| "step": 1213 |
| }, |
| { |
| "epoch": 1.89, |
| "learning_rate": 1.8704284221525603e-05, |
| "loss": 0.2888, |
| "step": 1214 |
| }, |
| { |
| "epoch": 1.89, |
| "learning_rate": 1.867816091954023e-05, |
| "loss": 0.2866, |
| "step": 1215 |
| }, |
| { |
| "epoch": 1.89, |
| "learning_rate": 1.865203761755486e-05, |
| "loss": 0.2796, |
| "step": 1216 |
| }, |
| { |
| "epoch": 1.89, |
| "learning_rate": 1.862591431556949e-05, |
| "loss": 0.2988, |
| "step": 1217 |
| }, |
| { |
| "epoch": 1.89, |
| "learning_rate": 1.8599791013584118e-05, |
| "loss": 0.3015, |
| "step": 1218 |
| }, |
| { |
| "epoch": 1.89, |
| "learning_rate": 1.8573667711598746e-05, |
| "loss": 0.2878, |
| "step": 1219 |
| }, |
| { |
| "epoch": 1.9, |
| "learning_rate": 1.8547544409613375e-05, |
| "loss": 0.29, |
| "step": 1220 |
| }, |
| { |
| "epoch": 1.9, |
| "learning_rate": 1.8521421107628007e-05, |
| "loss": 0.2796, |
| "step": 1221 |
| }, |
| { |
| "epoch": 1.9, |
| "learning_rate": 1.8495297805642632e-05, |
| "loss": 0.2895, |
| "step": 1222 |
| }, |
| { |
| "epoch": 1.9, |
| "learning_rate": 1.8469174503657264e-05, |
| "loss": 0.2915, |
| "step": 1223 |
| }, |
| { |
| "epoch": 1.9, |
| "learning_rate": 1.8443051201671893e-05, |
| "loss": 0.2853, |
| "step": 1224 |
| }, |
| { |
| "epoch": 1.9, |
| "learning_rate": 1.841692789968652e-05, |
| "loss": 0.2969, |
| "step": 1225 |
| }, |
| { |
| "epoch": 1.91, |
| "learning_rate": 1.839080459770115e-05, |
| "loss": 0.2843, |
| "step": 1226 |
| }, |
| { |
| "epoch": 1.91, |
| "learning_rate": 1.836468129571578e-05, |
| "loss": 0.2873, |
| "step": 1227 |
| }, |
| { |
| "epoch": 1.91, |
| "learning_rate": 1.833855799373041e-05, |
| "loss": 0.3044, |
| "step": 1228 |
| }, |
| { |
| "epoch": 1.91, |
| "learning_rate": 1.8312434691745036e-05, |
| "loss": 0.298, |
| "step": 1229 |
| }, |
| { |
| "epoch": 1.91, |
| "learning_rate": 1.8286311389759668e-05, |
| "loss": 0.3025, |
| "step": 1230 |
| }, |
| { |
| "epoch": 1.91, |
| "learning_rate": 1.8260188087774296e-05, |
| "loss": 0.2911, |
| "step": 1231 |
| }, |
| { |
| "epoch": 1.92, |
| "learning_rate": 1.8234064785788925e-05, |
| "loss": 0.2969, |
| "step": 1232 |
| }, |
| { |
| "epoch": 1.92, |
| "learning_rate": 1.8207941483803553e-05, |
| "loss": 0.2996, |
| "step": 1233 |
| }, |
| { |
| "epoch": 1.92, |
| "learning_rate": 1.8181818181818182e-05, |
| "loss": 0.2924, |
| "step": 1234 |
| }, |
| { |
| "epoch": 1.92, |
| "learning_rate": 1.8155694879832814e-05, |
| "loss": 0.2832, |
| "step": 1235 |
| }, |
| { |
| "epoch": 1.92, |
| "learning_rate": 1.812957157784744e-05, |
| "loss": 0.2859, |
| "step": 1236 |
| }, |
| { |
| "epoch": 1.92, |
| "learning_rate": 1.810344827586207e-05, |
| "loss": 0.2782, |
| "step": 1237 |
| }, |
| { |
| "epoch": 1.92, |
| "learning_rate": 1.80773249738767e-05, |
| "loss": 0.3126, |
| "step": 1238 |
| }, |
| { |
| "epoch": 1.93, |
| "learning_rate": 1.805120167189133e-05, |
| "loss": 0.2829, |
| "step": 1239 |
| }, |
| { |
| "epoch": 1.93, |
| "learning_rate": 1.8025078369905957e-05, |
| "loss": 0.2937, |
| "step": 1240 |
| }, |
| { |
| "epoch": 1.93, |
| "learning_rate": 1.7998955067920586e-05, |
| "loss": 0.285, |
| "step": 1241 |
| }, |
| { |
| "epoch": 1.93, |
| "learning_rate": 1.7972831765935218e-05, |
| "loss": 0.2895, |
| "step": 1242 |
| }, |
| { |
| "epoch": 1.93, |
| "learning_rate": 1.7946708463949843e-05, |
| "loss": 0.3082, |
| "step": 1243 |
| }, |
| { |
| "epoch": 1.93, |
| "learning_rate": 1.7920585161964475e-05, |
| "loss": 0.28, |
| "step": 1244 |
| }, |
| { |
| "epoch": 1.94, |
| "learning_rate": 1.7894461859979103e-05, |
| "loss": 0.3041, |
| "step": 1245 |
| }, |
| { |
| "epoch": 1.94, |
| "learning_rate": 1.786833855799373e-05, |
| "loss": 0.294, |
| "step": 1246 |
| }, |
| { |
| "epoch": 1.94, |
| "learning_rate": 1.784221525600836e-05, |
| "loss": 0.2923, |
| "step": 1247 |
| }, |
| { |
| "epoch": 1.94, |
| "learning_rate": 1.781609195402299e-05, |
| "loss": 0.2755, |
| "step": 1248 |
| }, |
| { |
| "epoch": 1.94, |
| "learning_rate": 1.778996865203762e-05, |
| "loss": 0.2864, |
| "step": 1249 |
| }, |
| { |
| "epoch": 1.94, |
| "learning_rate": 1.7763845350052246e-05, |
| "loss": 0.2926, |
| "step": 1250 |
| }, |
| { |
| "epoch": 1.94, |
| "learning_rate": 1.7737722048066875e-05, |
| "loss": 0.2872, |
| "step": 1251 |
| }, |
| { |
| "epoch": 1.95, |
| "learning_rate": 1.7711598746081507e-05, |
| "loss": 0.2876, |
| "step": 1252 |
| }, |
| { |
| "epoch": 1.95, |
| "learning_rate": 1.7685475444096132e-05, |
| "loss": 0.2935, |
| "step": 1253 |
| }, |
| { |
| "epoch": 1.95, |
| "learning_rate": 1.7659352142110764e-05, |
| "loss": 0.3101, |
| "step": 1254 |
| }, |
| { |
| "epoch": 1.95, |
| "learning_rate": 1.7633228840125393e-05, |
| "loss": 0.292, |
| "step": 1255 |
| }, |
| { |
| "epoch": 1.95, |
| "learning_rate": 1.7607105538140025e-05, |
| "loss": 0.2881, |
| "step": 1256 |
| }, |
| { |
| "epoch": 1.95, |
| "learning_rate": 1.758098223615465e-05, |
| "loss": 0.2891, |
| "step": 1257 |
| }, |
| { |
| "epoch": 1.96, |
| "learning_rate": 1.755485893416928e-05, |
| "loss": 0.2882, |
| "step": 1258 |
| }, |
| { |
| "epoch": 1.96, |
| "learning_rate": 1.752873563218391e-05, |
| "loss": 0.3096, |
| "step": 1259 |
| }, |
| { |
| "epoch": 1.96, |
| "learning_rate": 1.7502612330198536e-05, |
| "loss": 0.2975, |
| "step": 1260 |
| }, |
| { |
| "epoch": 1.96, |
| "learning_rate": 1.7476489028213168e-05, |
| "loss": 0.2812, |
| "step": 1261 |
| }, |
| { |
| "epoch": 1.96, |
| "learning_rate": 1.7450365726227796e-05, |
| "loss": 0.288, |
| "step": 1262 |
| }, |
| { |
| "epoch": 1.96, |
| "learning_rate": 1.7424242424242425e-05, |
| "loss": 0.2985, |
| "step": 1263 |
| }, |
| { |
| "epoch": 1.96, |
| "learning_rate": 1.7398119122257053e-05, |
| "loss": 0.2826, |
| "step": 1264 |
| }, |
| { |
| "epoch": 1.97, |
| "learning_rate": 1.7371995820271682e-05, |
| "loss": 0.3025, |
| "step": 1265 |
| }, |
| { |
| "epoch": 1.97, |
| "learning_rate": 1.7345872518286314e-05, |
| "loss": 0.2911, |
| "step": 1266 |
| }, |
| { |
| "epoch": 1.97, |
| "learning_rate": 1.731974921630094e-05, |
| "loss": 0.2997, |
| "step": 1267 |
| }, |
| { |
| "epoch": 1.97, |
| "learning_rate": 1.729362591431557e-05, |
| "loss": 0.2753, |
| "step": 1268 |
| }, |
| { |
| "epoch": 1.97, |
| "learning_rate": 1.72675026123302e-05, |
| "loss": 0.3036, |
| "step": 1269 |
| }, |
| { |
| "epoch": 1.97, |
| "learning_rate": 1.7241379310344828e-05, |
| "loss": 0.3155, |
| "step": 1270 |
| }, |
| { |
| "epoch": 1.98, |
| "learning_rate": 1.7215256008359457e-05, |
| "loss": 0.2817, |
| "step": 1271 |
| }, |
| { |
| "epoch": 1.98, |
| "learning_rate": 1.7189132706374085e-05, |
| "loss": 0.2791, |
| "step": 1272 |
| }, |
| { |
| "epoch": 1.98, |
| "learning_rate": 1.7163009404388717e-05, |
| "loss": 0.293, |
| "step": 1273 |
| }, |
| { |
| "epoch": 1.98, |
| "learning_rate": 1.7136886102403343e-05, |
| "loss": 0.2934, |
| "step": 1274 |
| }, |
| { |
| "epoch": 1.98, |
| "learning_rate": 1.7110762800417975e-05, |
| "loss": 0.2923, |
| "step": 1275 |
| }, |
| { |
| "epoch": 1.98, |
| "learning_rate": 1.7084639498432603e-05, |
| "loss": 0.2998, |
| "step": 1276 |
| }, |
| { |
| "epoch": 1.98, |
| "learning_rate": 1.7058516196447232e-05, |
| "loss": 0.2978, |
| "step": 1277 |
| }, |
| { |
| "epoch": 1.99, |
| "learning_rate": 1.703239289446186e-05, |
| "loss": 0.2954, |
| "step": 1278 |
| }, |
| { |
| "epoch": 1.99, |
| "learning_rate": 1.700626959247649e-05, |
| "loss": 0.3174, |
| "step": 1279 |
| }, |
| { |
| "epoch": 1.99, |
| "learning_rate": 1.698014629049112e-05, |
| "loss": 0.2942, |
| "step": 1280 |
| }, |
| { |
| "epoch": 1.99, |
| "learning_rate": 1.6954022988505746e-05, |
| "loss": 0.2834, |
| "step": 1281 |
| }, |
| { |
| "epoch": 1.99, |
| "learning_rate": 1.6927899686520378e-05, |
| "loss": 0.3028, |
| "step": 1282 |
| }, |
| { |
| "epoch": 1.99, |
| "learning_rate": 1.6901776384535007e-05, |
| "loss": 0.2714, |
| "step": 1283 |
| }, |
| { |
| "epoch": 2.0, |
| "learning_rate": 1.6875653082549635e-05, |
| "loss": 0.2945, |
| "step": 1284 |
| }, |
| { |
| "epoch": 2.0, |
| "learning_rate": 1.6849529780564264e-05, |
| "loss": 0.2955, |
| "step": 1285 |
| }, |
| { |
| "epoch": 2.0, |
| "learning_rate": 1.6823406478578893e-05, |
| "loss": 0.2974, |
| "step": 1286 |
| }, |
| { |
| "epoch": 2.0, |
| "learning_rate": 1.6797283176593525e-05, |
| "loss": 0.2738, |
| "step": 1287 |
| }, |
| { |
| "epoch": 2.0, |
| "learning_rate": 1.677115987460815e-05, |
| "loss": 0.1929, |
| "step": 1288 |
| }, |
| { |
| "epoch": 2.0, |
| "learning_rate": 1.6745036572622778e-05, |
| "loss": 0.1997, |
| "step": 1289 |
| }, |
| { |
| "epoch": 2.01, |
| "learning_rate": 1.671891327063741e-05, |
| "loss": 0.1878, |
| "step": 1290 |
| }, |
| { |
| "epoch": 2.01, |
| "learning_rate": 1.669278996865204e-05, |
| "loss": 0.203, |
| "step": 1291 |
| }, |
| { |
| "epoch": 2.01, |
| "learning_rate": 1.6666666666666667e-05, |
| "loss": 0.1847, |
| "step": 1292 |
| }, |
| { |
| "epoch": 2.01, |
| "learning_rate": 1.6640543364681296e-05, |
| "loss": 0.2022, |
| "step": 1293 |
| }, |
| { |
| "epoch": 2.01, |
| "learning_rate": 1.6614420062695925e-05, |
| "loss": 0.1817, |
| "step": 1294 |
| }, |
| { |
| "epoch": 2.01, |
| "learning_rate": 1.6588296760710553e-05, |
| "loss": 0.1843, |
| "step": 1295 |
| }, |
| { |
| "epoch": 2.01, |
| "learning_rate": 1.6562173458725182e-05, |
| "loss": 0.184, |
| "step": 1296 |
| }, |
| { |
| "epoch": 2.02, |
| "learning_rate": 1.6536050156739814e-05, |
| "loss": 0.1867, |
| "step": 1297 |
| }, |
| { |
| "epoch": 2.02, |
| "learning_rate": 1.6509926854754442e-05, |
| "loss": 0.1792, |
| "step": 1298 |
| }, |
| { |
| "epoch": 2.02, |
| "learning_rate": 1.648380355276907e-05, |
| "loss": 0.1901, |
| "step": 1299 |
| }, |
| { |
| "epoch": 2.02, |
| "learning_rate": 1.64576802507837e-05, |
| "loss": 0.1894, |
| "step": 1300 |
| }, |
| { |
| "epoch": 2.02, |
| "learning_rate": 1.6431556948798328e-05, |
| "loss": 0.1824, |
| "step": 1301 |
| }, |
| { |
| "epoch": 2.02, |
| "learning_rate": 1.6405433646812957e-05, |
| "loss": 0.1924, |
| "step": 1302 |
| }, |
| { |
| "epoch": 2.03, |
| "learning_rate": 1.6379310344827585e-05, |
| "loss": 0.1843, |
| "step": 1303 |
| }, |
| { |
| "epoch": 2.03, |
| "learning_rate": 1.6353187042842217e-05, |
| "loss": 0.1873, |
| "step": 1304 |
| }, |
| { |
| "epoch": 2.03, |
| "learning_rate": 1.6327063740856846e-05, |
| "loss": 0.198, |
| "step": 1305 |
| }, |
| { |
| "epoch": 2.03, |
| "learning_rate": 1.6300940438871475e-05, |
| "loss": 0.1831, |
| "step": 1306 |
| }, |
| { |
| "epoch": 2.03, |
| "learning_rate": 1.6274817136886103e-05, |
| "loss": 0.1859, |
| "step": 1307 |
| }, |
| { |
| "epoch": 2.03, |
| "learning_rate": 1.6248693834900732e-05, |
| "loss": 0.1879, |
| "step": 1308 |
| }, |
| { |
| "epoch": 2.03, |
| "learning_rate": 1.6222570532915364e-05, |
| "loss": 0.1871, |
| "step": 1309 |
| }, |
| { |
| "epoch": 2.04, |
| "learning_rate": 1.619644723092999e-05, |
| "loss": 0.2004, |
| "step": 1310 |
| }, |
| { |
| "epoch": 2.04, |
| "learning_rate": 1.617032392894462e-05, |
| "loss": 0.1689, |
| "step": 1311 |
| }, |
| { |
| "epoch": 2.04, |
| "learning_rate": 1.614420062695925e-05, |
| "loss": 0.1851, |
| "step": 1312 |
| }, |
| { |
| "epoch": 2.04, |
| "learning_rate": 1.6118077324973878e-05, |
| "loss": 0.1889, |
| "step": 1313 |
| }, |
| { |
| "epoch": 2.04, |
| "learning_rate": 1.6091954022988507e-05, |
| "loss": 0.1799, |
| "step": 1314 |
| }, |
| { |
| "epoch": 2.04, |
| "learning_rate": 1.6065830721003135e-05, |
| "loss": 0.1902, |
| "step": 1315 |
| }, |
| { |
| "epoch": 2.05, |
| "learning_rate": 1.6039707419017767e-05, |
| "loss": 0.1848, |
| "step": 1316 |
| }, |
| { |
| "epoch": 2.05, |
| "learning_rate": 1.6013584117032392e-05, |
| "loss": 0.1793, |
| "step": 1317 |
| }, |
| { |
| "epoch": 2.05, |
| "learning_rate": 1.5987460815047024e-05, |
| "loss": 0.1948, |
| "step": 1318 |
| }, |
| { |
| "epoch": 2.05, |
| "learning_rate": 1.5961337513061653e-05, |
| "loss": 0.2018, |
| "step": 1319 |
| }, |
| { |
| "epoch": 2.05, |
| "learning_rate": 1.5935214211076278e-05, |
| "loss": 0.1832, |
| "step": 1320 |
| }, |
| { |
| "epoch": 2.05, |
| "learning_rate": 1.590909090909091e-05, |
| "loss": 0.1973, |
| "step": 1321 |
| }, |
| { |
| "epoch": 2.05, |
| "learning_rate": 1.588296760710554e-05, |
| "loss": 0.1937, |
| "step": 1322 |
| }, |
| { |
| "epoch": 2.06, |
| "learning_rate": 1.585684430512017e-05, |
| "loss": 0.1915, |
| "step": 1323 |
| }, |
| { |
| "epoch": 2.06, |
| "learning_rate": 1.5830721003134796e-05, |
| "loss": 0.1798, |
| "step": 1324 |
| }, |
| { |
| "epoch": 2.06, |
| "learning_rate": 1.5804597701149425e-05, |
| "loss": 0.1824, |
| "step": 1325 |
| }, |
| { |
| "epoch": 2.06, |
| "learning_rate": 1.5778474399164057e-05, |
| "loss": 0.1906, |
| "step": 1326 |
| }, |
| { |
| "epoch": 2.06, |
| "learning_rate": 1.5752351097178682e-05, |
| "loss": 0.184, |
| "step": 1327 |
| }, |
| { |
| "epoch": 2.06, |
| "learning_rate": 1.5726227795193314e-05, |
| "loss": 0.1655, |
| "step": 1328 |
| }, |
| { |
| "epoch": 2.07, |
| "learning_rate": 1.5700104493207942e-05, |
| "loss": 0.184, |
| "step": 1329 |
| }, |
| { |
| "epoch": 2.07, |
| "learning_rate": 1.5673981191222574e-05, |
| "loss": 0.1745, |
| "step": 1330 |
| }, |
| { |
| "epoch": 2.07, |
| "learning_rate": 1.56478578892372e-05, |
| "loss": 0.18, |
| "step": 1331 |
| }, |
| { |
| "epoch": 2.07, |
| "learning_rate": 1.5621734587251828e-05, |
| "loss": 0.1891, |
| "step": 1332 |
| }, |
| { |
| "epoch": 2.07, |
| "learning_rate": 1.559561128526646e-05, |
| "loss": 0.1815, |
| "step": 1333 |
| }, |
| { |
| "epoch": 2.07, |
| "learning_rate": 1.5569487983281085e-05, |
| "loss": 0.1888, |
| "step": 1334 |
| }, |
| { |
| "epoch": 2.08, |
| "learning_rate": 1.5543364681295717e-05, |
| "loss": 0.185, |
| "step": 1335 |
| }, |
| { |
| "epoch": 2.08, |
| "learning_rate": 1.5517241379310346e-05, |
| "loss": 0.1838, |
| "step": 1336 |
| }, |
| { |
| "epoch": 2.08, |
| "learning_rate": 1.5491118077324974e-05, |
| "loss": 0.1861, |
| "step": 1337 |
| }, |
| { |
| "epoch": 2.08, |
| "learning_rate": 1.5464994775339603e-05, |
| "loss": 0.1726, |
| "step": 1338 |
| }, |
| { |
| "epoch": 2.08, |
| "learning_rate": 1.543887147335423e-05, |
| "loss": 0.1765, |
| "step": 1339 |
| }, |
| { |
| "epoch": 2.08, |
| "learning_rate": 1.5412748171368864e-05, |
| "loss": 0.1894, |
| "step": 1340 |
| }, |
| { |
| "epoch": 2.08, |
| "learning_rate": 1.538662486938349e-05, |
| "loss": 0.1886, |
| "step": 1341 |
| }, |
| { |
| "epoch": 2.09, |
| "learning_rate": 1.536050156739812e-05, |
| "loss": 0.179, |
| "step": 1342 |
| }, |
| { |
| "epoch": 2.09, |
| "learning_rate": 1.533437826541275e-05, |
| "loss": 0.1923, |
| "step": 1343 |
| }, |
| { |
| "epoch": 2.09, |
| "learning_rate": 1.5308254963427378e-05, |
| "loss": 0.1881, |
| "step": 1344 |
| }, |
| { |
| "epoch": 2.09, |
| "learning_rate": 1.5282131661442007e-05, |
| "loss": 0.1821, |
| "step": 1345 |
| }, |
| { |
| "epoch": 2.09, |
| "learning_rate": 1.5256008359456637e-05, |
| "loss": 0.1805, |
| "step": 1346 |
| }, |
| { |
| "epoch": 2.09, |
| "learning_rate": 1.5229885057471265e-05, |
| "loss": 0.1803, |
| "step": 1347 |
| }, |
| { |
| "epoch": 2.1, |
| "learning_rate": 1.5203761755485892e-05, |
| "loss": 0.1939, |
| "step": 1348 |
| }, |
| { |
| "epoch": 2.1, |
| "learning_rate": 1.5177638453500523e-05, |
| "loss": 0.1724, |
| "step": 1349 |
| }, |
| { |
| "epoch": 2.1, |
| "learning_rate": 1.5151515151515153e-05, |
| "loss": 0.1926, |
| "step": 1350 |
| }, |
| { |
| "epoch": 2.1, |
| "learning_rate": 1.5125391849529783e-05, |
| "loss": 0.1892, |
| "step": 1351 |
| }, |
| { |
| "epoch": 2.1, |
| "learning_rate": 1.509926854754441e-05, |
| "loss": 0.1969, |
| "step": 1352 |
| }, |
| { |
| "epoch": 2.1, |
| "learning_rate": 1.5073145245559039e-05, |
| "loss": 0.1933, |
| "step": 1353 |
| }, |
| { |
| "epoch": 2.1, |
| "learning_rate": 1.5047021943573669e-05, |
| "loss": 0.1838, |
| "step": 1354 |
| }, |
| { |
| "epoch": 2.11, |
| "learning_rate": 1.5020898641588296e-05, |
| "loss": 0.1739, |
| "step": 1355 |
| }, |
| { |
| "epoch": 2.11, |
| "learning_rate": 1.4994775339602926e-05, |
| "loss": 0.1952, |
| "step": 1356 |
| }, |
| { |
| "epoch": 2.11, |
| "learning_rate": 1.4968652037617556e-05, |
| "loss": 0.1833, |
| "step": 1357 |
| }, |
| { |
| "epoch": 2.11, |
| "learning_rate": 1.4942528735632185e-05, |
| "loss": 0.1915, |
| "step": 1358 |
| }, |
| { |
| "epoch": 2.11, |
| "learning_rate": 1.4916405433646814e-05, |
| "loss": 0.1895, |
| "step": 1359 |
| }, |
| { |
| "epoch": 2.11, |
| "learning_rate": 1.4890282131661442e-05, |
| "loss": 0.1822, |
| "step": 1360 |
| }, |
| { |
| "epoch": 2.12, |
| "learning_rate": 1.4864158829676072e-05, |
| "loss": 0.1873, |
| "step": 1361 |
| }, |
| { |
| "epoch": 2.12, |
| "learning_rate": 1.48380355276907e-05, |
| "loss": 0.1906, |
| "step": 1362 |
| }, |
| { |
| "epoch": 2.12, |
| "learning_rate": 1.481191222570533e-05, |
| "loss": 0.1904, |
| "step": 1363 |
| }, |
| { |
| "epoch": 2.12, |
| "learning_rate": 1.478578892371996e-05, |
| "loss": 0.1807, |
| "step": 1364 |
| }, |
| { |
| "epoch": 2.12, |
| "learning_rate": 1.4759665621734589e-05, |
| "loss": 0.19, |
| "step": 1365 |
| }, |
| { |
| "epoch": 2.12, |
| "learning_rate": 1.4733542319749215e-05, |
| "loss": 0.1744, |
| "step": 1366 |
| }, |
| { |
| "epoch": 2.12, |
| "learning_rate": 1.4707419017763846e-05, |
| "loss": 0.1858, |
| "step": 1367 |
| }, |
| { |
| "epoch": 2.13, |
| "learning_rate": 1.4681295715778476e-05, |
| "loss": 0.1931, |
| "step": 1368 |
| }, |
| { |
| "epoch": 2.13, |
| "learning_rate": 1.4655172413793103e-05, |
| "loss": 0.1843, |
| "step": 1369 |
| }, |
| { |
| "epoch": 2.13, |
| "learning_rate": 1.4629049111807733e-05, |
| "loss": 0.1824, |
| "step": 1370 |
| }, |
| { |
| "epoch": 2.13, |
| "learning_rate": 1.4602925809822362e-05, |
| "loss": 0.1899, |
| "step": 1371 |
| }, |
| { |
| "epoch": 2.13, |
| "learning_rate": 1.4576802507836992e-05, |
| "loss": 0.1782, |
| "step": 1372 |
| }, |
| { |
| "epoch": 2.13, |
| "learning_rate": 1.4550679205851619e-05, |
| "loss": 0.1919, |
| "step": 1373 |
| }, |
| { |
| "epoch": 2.14, |
| "learning_rate": 1.452455590386625e-05, |
| "loss": 0.1833, |
| "step": 1374 |
| }, |
| { |
| "epoch": 2.14, |
| "learning_rate": 1.449843260188088e-05, |
| "loss": 0.1854, |
| "step": 1375 |
| }, |
| { |
| "epoch": 2.14, |
| "learning_rate": 1.4472309299895506e-05, |
| "loss": 0.1775, |
| "step": 1376 |
| }, |
| { |
| "epoch": 2.14, |
| "learning_rate": 1.4446185997910137e-05, |
| "loss": 0.188, |
| "step": 1377 |
| }, |
| { |
| "epoch": 2.14, |
| "learning_rate": 1.4420062695924765e-05, |
| "loss": 0.1777, |
| "step": 1378 |
| }, |
| { |
| "epoch": 2.14, |
| "learning_rate": 1.4393939393939396e-05, |
| "loss": 0.1759, |
| "step": 1379 |
| }, |
| { |
| "epoch": 2.14, |
| "learning_rate": 1.4367816091954022e-05, |
| "loss": 0.194, |
| "step": 1380 |
| }, |
| { |
| "epoch": 2.15, |
| "learning_rate": 1.4341692789968653e-05, |
| "loss": 0.1762, |
| "step": 1381 |
| }, |
| { |
| "epoch": 2.15, |
| "learning_rate": 1.4315569487983283e-05, |
| "loss": 0.1748, |
| "step": 1382 |
| }, |
| { |
| "epoch": 2.15, |
| "learning_rate": 1.428944618599791e-05, |
| "loss": 0.1714, |
| "step": 1383 |
| }, |
| { |
| "epoch": 2.15, |
| "learning_rate": 1.4263322884012539e-05, |
| "loss": 0.1917, |
| "step": 1384 |
| }, |
| { |
| "epoch": 2.15, |
| "learning_rate": 1.4237199582027169e-05, |
| "loss": 0.1872, |
| "step": 1385 |
| }, |
| { |
| "epoch": 2.15, |
| "learning_rate": 1.4211076280041799e-05, |
| "loss": 0.1867, |
| "step": 1386 |
| }, |
| { |
| "epoch": 2.16, |
| "learning_rate": 1.4184952978056426e-05, |
| "loss": 0.1824, |
| "step": 1387 |
| }, |
| { |
| "epoch": 2.16, |
| "learning_rate": 1.4158829676071056e-05, |
| "loss": 0.1816, |
| "step": 1388 |
| }, |
| { |
| "epoch": 2.16, |
| "learning_rate": 1.4132706374085687e-05, |
| "loss": 0.1877, |
| "step": 1389 |
| }, |
| { |
| "epoch": 2.16, |
| "learning_rate": 1.4106583072100313e-05, |
| "loss": 0.1956, |
| "step": 1390 |
| }, |
| { |
| "epoch": 2.16, |
| "learning_rate": 1.4080459770114942e-05, |
| "loss": 0.1775, |
| "step": 1391 |
| }, |
| { |
| "epoch": 2.16, |
| "learning_rate": 1.4054336468129572e-05, |
| "loss": 0.1911, |
| "step": 1392 |
| }, |
| { |
| "epoch": 2.17, |
| "learning_rate": 1.4028213166144203e-05, |
| "loss": 0.1775, |
| "step": 1393 |
| }, |
| { |
| "epoch": 2.17, |
| "learning_rate": 1.400208986415883e-05, |
| "loss": 0.182, |
| "step": 1394 |
| }, |
| { |
| "epoch": 2.17, |
| "learning_rate": 1.397596656217346e-05, |
| "loss": 0.1916, |
| "step": 1395 |
| }, |
| { |
| "epoch": 2.17, |
| "learning_rate": 1.3949843260188088e-05, |
| "loss": 0.1818, |
| "step": 1396 |
| }, |
| { |
| "epoch": 2.17, |
| "learning_rate": 1.3923719958202717e-05, |
| "loss": 0.1944, |
| "step": 1397 |
| }, |
| { |
| "epoch": 2.17, |
| "learning_rate": 1.3897596656217346e-05, |
| "loss": 0.1811, |
| "step": 1398 |
| }, |
| { |
| "epoch": 2.17, |
| "learning_rate": 1.3871473354231976e-05, |
| "loss": 0.1986, |
| "step": 1399 |
| }, |
| { |
| "epoch": 2.18, |
| "learning_rate": 1.3845350052246606e-05, |
| "loss": 0.1901, |
| "step": 1400 |
| }, |
| { |
| "epoch": 2.18, |
| "learning_rate": 1.3819226750261233e-05, |
| "loss": 0.1824, |
| "step": 1401 |
| }, |
| { |
| "epoch": 2.18, |
| "learning_rate": 1.3793103448275863e-05, |
| "loss": 0.1796, |
| "step": 1402 |
| }, |
| { |
| "epoch": 2.18, |
| "learning_rate": 1.3766980146290492e-05, |
| "loss": 0.1917, |
| "step": 1403 |
| }, |
| { |
| "epoch": 2.18, |
| "learning_rate": 1.3740856844305122e-05, |
| "loss": 0.1919, |
| "step": 1404 |
| }, |
| { |
| "epoch": 2.18, |
| "learning_rate": 1.3714733542319749e-05, |
| "loss": 0.1767, |
| "step": 1405 |
| }, |
| { |
| "epoch": 2.19, |
| "learning_rate": 1.368861024033438e-05, |
| "loss": 0.1849, |
| "step": 1406 |
| }, |
| { |
| "epoch": 2.19, |
| "learning_rate": 1.366248693834901e-05, |
| "loss": 0.1932, |
| "step": 1407 |
| }, |
| { |
| "epoch": 2.19, |
| "learning_rate": 1.3636363636363637e-05, |
| "loss": 0.1933, |
| "step": 1408 |
| }, |
| { |
| "epoch": 2.19, |
| "learning_rate": 1.3610240334378265e-05, |
| "loss": 0.175, |
| "step": 1409 |
| }, |
| { |
| "epoch": 2.19, |
| "learning_rate": 1.3584117032392895e-05, |
| "loss": 0.1899, |
| "step": 1410 |
| }, |
| { |
| "epoch": 2.19, |
| "learning_rate": 1.3557993730407526e-05, |
| "loss": 0.1751, |
| "step": 1411 |
| }, |
| { |
| "epoch": 2.19, |
| "learning_rate": 1.3531870428422153e-05, |
| "loss": 0.1769, |
| "step": 1412 |
| }, |
| { |
| "epoch": 2.2, |
| "learning_rate": 1.3505747126436783e-05, |
| "loss": 0.1821, |
| "step": 1413 |
| }, |
| { |
| "epoch": 2.2, |
| "learning_rate": 1.3479623824451411e-05, |
| "loss": 0.1838, |
| "step": 1414 |
| }, |
| { |
| "epoch": 2.2, |
| "learning_rate": 1.345350052246604e-05, |
| "loss": 0.1763, |
| "step": 1415 |
| }, |
| { |
| "epoch": 2.2, |
| "learning_rate": 1.3427377220480669e-05, |
| "loss": 0.1803, |
| "step": 1416 |
| }, |
| { |
| "epoch": 2.2, |
| "learning_rate": 1.3401253918495299e-05, |
| "loss": 0.1758, |
| "step": 1417 |
| }, |
| { |
| "epoch": 2.2, |
| "learning_rate": 1.337513061650993e-05, |
| "loss": 0.1895, |
| "step": 1418 |
| }, |
| { |
| "epoch": 2.21, |
| "learning_rate": 1.3349007314524556e-05, |
| "loss": 0.1762, |
| "step": 1419 |
| }, |
| { |
| "epoch": 2.21, |
| "learning_rate": 1.3322884012539186e-05, |
| "loss": 0.1806, |
| "step": 1420 |
| }, |
| { |
| "epoch": 2.21, |
| "learning_rate": 1.3296760710553815e-05, |
| "loss": 0.1777, |
| "step": 1421 |
| }, |
| { |
| "epoch": 2.21, |
| "learning_rate": 1.3270637408568442e-05, |
| "loss": 0.1844, |
| "step": 1422 |
| }, |
| { |
| "epoch": 2.21, |
| "learning_rate": 1.3244514106583072e-05, |
| "loss": 0.1983, |
| "step": 1423 |
| }, |
| { |
| "epoch": 2.21, |
| "learning_rate": 1.3218390804597702e-05, |
| "loss": 0.1787, |
| "step": 1424 |
| }, |
| { |
| "epoch": 2.21, |
| "learning_rate": 1.3192267502612333e-05, |
| "loss": 0.1899, |
| "step": 1425 |
| }, |
| { |
| "epoch": 2.22, |
| "learning_rate": 1.316614420062696e-05, |
| "loss": 0.1912, |
| "step": 1426 |
| }, |
| { |
| "epoch": 2.22, |
| "learning_rate": 1.3140020898641588e-05, |
| "loss": 0.198, |
| "step": 1427 |
| }, |
| { |
| "epoch": 2.22, |
| "learning_rate": 1.3113897596656219e-05, |
| "loss": 0.1797, |
| "step": 1428 |
| }, |
| { |
| "epoch": 2.22, |
| "learning_rate": 1.3087774294670845e-05, |
| "loss": 0.1932, |
| "step": 1429 |
| }, |
| { |
| "epoch": 2.22, |
| "learning_rate": 1.3061650992685476e-05, |
| "loss": 0.185, |
| "step": 1430 |
| }, |
| { |
| "epoch": 2.22, |
| "learning_rate": 1.3035527690700106e-05, |
| "loss": 0.1848, |
| "step": 1431 |
| }, |
| { |
| "epoch": 2.23, |
| "learning_rate": 1.3009404388714735e-05, |
| "loss": 0.1747, |
| "step": 1432 |
| }, |
| { |
| "epoch": 2.23, |
| "learning_rate": 1.2983281086729363e-05, |
| "loss": 0.1819, |
| "step": 1433 |
| }, |
| { |
| "epoch": 2.23, |
| "learning_rate": 1.2957157784743992e-05, |
| "loss": 0.1801, |
| "step": 1434 |
| }, |
| { |
| "epoch": 2.23, |
| "learning_rate": 1.2931034482758622e-05, |
| "loss": 0.1765, |
| "step": 1435 |
| }, |
| { |
| "epoch": 2.23, |
| "learning_rate": 1.2904911180773249e-05, |
| "loss": 0.1748, |
| "step": 1436 |
| }, |
| { |
| "epoch": 2.23, |
| "learning_rate": 1.287878787878788e-05, |
| "loss": 0.1812, |
| "step": 1437 |
| }, |
| { |
| "epoch": 2.24, |
| "learning_rate": 1.285266457680251e-05, |
| "loss": 0.186, |
| "step": 1438 |
| }, |
| { |
| "epoch": 2.24, |
| "learning_rate": 1.2826541274817138e-05, |
| "loss": 0.1933, |
| "step": 1439 |
| }, |
| { |
| "epoch": 2.24, |
| "learning_rate": 1.2800417972831765e-05, |
| "loss": 0.1733, |
| "step": 1440 |
| }, |
| { |
| "epoch": 2.24, |
| "learning_rate": 1.2774294670846395e-05, |
| "loss": 0.1845, |
| "step": 1441 |
| }, |
| { |
| "epoch": 2.24, |
| "learning_rate": 1.2748171368861026e-05, |
| "loss": 0.1839, |
| "step": 1442 |
| }, |
| { |
| "epoch": 2.24, |
| "learning_rate": 1.2722048066875652e-05, |
| "loss": 0.1783, |
| "step": 1443 |
| }, |
| { |
| "epoch": 2.24, |
| "learning_rate": 1.2695924764890283e-05, |
| "loss": 0.1806, |
| "step": 1444 |
| }, |
| { |
| "epoch": 2.25, |
| "learning_rate": 1.2669801462904911e-05, |
| "loss": 0.1837, |
| "step": 1445 |
| }, |
| { |
| "epoch": 2.25, |
| "learning_rate": 1.2643678160919542e-05, |
| "loss": 0.1738, |
| "step": 1446 |
| }, |
| { |
| "epoch": 2.25, |
| "learning_rate": 1.2617554858934169e-05, |
| "loss": 0.182, |
| "step": 1447 |
| }, |
| { |
| "epoch": 2.25, |
| "learning_rate": 1.2591431556948799e-05, |
| "loss": 0.1919, |
| "step": 1448 |
| }, |
| { |
| "epoch": 2.25, |
| "learning_rate": 1.2565308254963429e-05, |
| "loss": 0.1791, |
| "step": 1449 |
| }, |
| { |
| "epoch": 2.25, |
| "learning_rate": 1.2539184952978056e-05, |
| "loss": 0.1856, |
| "step": 1450 |
| }, |
| { |
| "epoch": 2.26, |
| "learning_rate": 1.2513061650992686e-05, |
| "loss": 0.1798, |
| "step": 1451 |
| }, |
| { |
| "epoch": 2.26, |
| "learning_rate": 1.2486938349007315e-05, |
| "loss": 0.1708, |
| "step": 1452 |
| }, |
| { |
| "epoch": 2.26, |
| "learning_rate": 1.2460815047021943e-05, |
| "loss": 0.2019, |
| "step": 1453 |
| }, |
| { |
| "epoch": 2.26, |
| "learning_rate": 1.2434691745036574e-05, |
| "loss": 0.1978, |
| "step": 1454 |
| }, |
| { |
| "epoch": 2.26, |
| "learning_rate": 1.2408568443051202e-05, |
| "loss": 0.1831, |
| "step": 1455 |
| }, |
| { |
| "epoch": 2.26, |
| "learning_rate": 1.2382445141065831e-05, |
| "loss": 0.1795, |
| "step": 1456 |
| }, |
| { |
| "epoch": 2.26, |
| "learning_rate": 1.2356321839080461e-05, |
| "loss": 0.1745, |
| "step": 1457 |
| }, |
| { |
| "epoch": 2.27, |
| "learning_rate": 1.233019853709509e-05, |
| "loss": 0.1715, |
| "step": 1458 |
| }, |
| { |
| "epoch": 2.27, |
| "learning_rate": 1.2304075235109718e-05, |
| "loss": 0.1896, |
| "step": 1459 |
| }, |
| { |
| "epoch": 2.27, |
| "learning_rate": 1.2277951933124347e-05, |
| "loss": 0.1813, |
| "step": 1460 |
| }, |
| { |
| "epoch": 2.27, |
| "learning_rate": 1.2251828631138977e-05, |
| "loss": 0.1824, |
| "step": 1461 |
| }, |
| { |
| "epoch": 2.27, |
| "learning_rate": 1.2225705329153606e-05, |
| "loss": 0.1799, |
| "step": 1462 |
| }, |
| { |
| "epoch": 2.27, |
| "learning_rate": 1.2199582027168234e-05, |
| "loss": 0.1766, |
| "step": 1463 |
| }, |
| { |
| "epoch": 2.28, |
| "learning_rate": 1.2173458725182863e-05, |
| "loss": 0.1866, |
| "step": 1464 |
| }, |
| { |
| "epoch": 2.28, |
| "learning_rate": 1.2147335423197492e-05, |
| "loss": 0.18, |
| "step": 1465 |
| }, |
| { |
| "epoch": 2.28, |
| "learning_rate": 1.2121212121212122e-05, |
| "loss": 0.1812, |
| "step": 1466 |
| }, |
| { |
| "epoch": 2.28, |
| "learning_rate": 1.209508881922675e-05, |
| "loss": 0.1853, |
| "step": 1467 |
| }, |
| { |
| "epoch": 2.28, |
| "learning_rate": 1.206896551724138e-05, |
| "loss": 0.202, |
| "step": 1468 |
| }, |
| { |
| "epoch": 2.28, |
| "learning_rate": 1.204284221525601e-05, |
| "loss": 0.1884, |
| "step": 1469 |
| }, |
| { |
| "epoch": 2.28, |
| "learning_rate": 1.2016718913270638e-05, |
| "loss": 0.17, |
| "step": 1470 |
| }, |
| { |
| "epoch": 2.29, |
| "learning_rate": 1.1990595611285267e-05, |
| "loss": 0.1822, |
| "step": 1471 |
| }, |
| { |
| "epoch": 2.29, |
| "learning_rate": 1.1964472309299895e-05, |
| "loss": 0.1811, |
| "step": 1472 |
| }, |
| { |
| "epoch": 2.29, |
| "learning_rate": 1.1938349007314525e-05, |
| "loss": 0.1985, |
| "step": 1473 |
| }, |
| { |
| "epoch": 2.29, |
| "learning_rate": 1.1912225705329154e-05, |
| "loss": 0.198, |
| "step": 1474 |
| }, |
| { |
| "epoch": 2.29, |
| "learning_rate": 1.1886102403343784e-05, |
| "loss": 0.1818, |
| "step": 1475 |
| }, |
| { |
| "epoch": 2.29, |
| "learning_rate": 1.1859979101358413e-05, |
| "loss": 0.1884, |
| "step": 1476 |
| }, |
| { |
| "epoch": 2.3, |
| "learning_rate": 1.1833855799373042e-05, |
| "loss": 0.1832, |
| "step": 1477 |
| }, |
| { |
| "epoch": 2.3, |
| "learning_rate": 1.180773249738767e-05, |
| "loss": 0.188, |
| "step": 1478 |
| }, |
| { |
| "epoch": 2.3, |
| "learning_rate": 1.1781609195402299e-05, |
| "loss": 0.186, |
| "step": 1479 |
| }, |
| { |
| "epoch": 2.3, |
| "learning_rate": 1.1755485893416929e-05, |
| "loss": 0.1865, |
| "step": 1480 |
| }, |
| { |
| "epoch": 2.3, |
| "learning_rate": 1.1729362591431558e-05, |
| "loss": 0.175, |
| "step": 1481 |
| }, |
| { |
| "epoch": 2.3, |
| "learning_rate": 1.1703239289446186e-05, |
| "loss": 0.1658, |
| "step": 1482 |
| }, |
| { |
| "epoch": 2.3, |
| "learning_rate": 1.1677115987460815e-05, |
| "loss": 0.1795, |
| "step": 1483 |
| }, |
| { |
| "epoch": 2.31, |
| "learning_rate": 1.1650992685475445e-05, |
| "loss": 0.1758, |
| "step": 1484 |
| }, |
| { |
| "epoch": 2.31, |
| "learning_rate": 1.1624869383490074e-05, |
| "loss": 0.1843, |
| "step": 1485 |
| }, |
| { |
| "epoch": 2.31, |
| "learning_rate": 1.1598746081504702e-05, |
| "loss": 0.1803, |
| "step": 1486 |
| }, |
| { |
| "epoch": 2.31, |
| "learning_rate": 1.1572622779519333e-05, |
| "loss": 0.1839, |
| "step": 1487 |
| }, |
| { |
| "epoch": 2.31, |
| "learning_rate": 1.1546499477533961e-05, |
| "loss": 0.1745, |
| "step": 1488 |
| }, |
| { |
| "epoch": 2.31, |
| "learning_rate": 1.152037617554859e-05, |
| "loss": 0.1796, |
| "step": 1489 |
| }, |
| { |
| "epoch": 2.32, |
| "learning_rate": 1.1494252873563218e-05, |
| "loss": 0.1956, |
| "step": 1490 |
| }, |
| { |
| "epoch": 2.32, |
| "learning_rate": 1.1468129571577849e-05, |
| "loss": 0.1874, |
| "step": 1491 |
| }, |
| { |
| "epoch": 2.32, |
| "learning_rate": 1.1442006269592477e-05, |
| "loss": 0.1805, |
| "step": 1492 |
| }, |
| { |
| "epoch": 2.32, |
| "learning_rate": 1.1415882967607106e-05, |
| "loss": 0.1909, |
| "step": 1493 |
| }, |
| { |
| "epoch": 2.32, |
| "learning_rate": 1.1389759665621736e-05, |
| "loss": 0.1785, |
| "step": 1494 |
| }, |
| { |
| "epoch": 2.32, |
| "learning_rate": 1.1363636363636365e-05, |
| "loss": 0.1861, |
| "step": 1495 |
| }, |
| { |
| "epoch": 2.33, |
| "learning_rate": 1.1337513061650993e-05, |
| "loss": 0.1686, |
| "step": 1496 |
| }, |
| { |
| "epoch": 2.33, |
| "learning_rate": 1.1311389759665622e-05, |
| "loss": 0.1748, |
| "step": 1497 |
| }, |
| { |
| "epoch": 2.33, |
| "learning_rate": 1.1285266457680252e-05, |
| "loss": 0.1778, |
| "step": 1498 |
| }, |
| { |
| "epoch": 2.33, |
| "learning_rate": 1.125914315569488e-05, |
| "loss": 0.1983, |
| "step": 1499 |
| }, |
| { |
| "epoch": 2.33, |
| "learning_rate": 1.123301985370951e-05, |
| "loss": 0.1796, |
| "step": 1500 |
| }, |
| { |
| "epoch": 2.33, |
| "learning_rate": 1.1206896551724138e-05, |
| "loss": 0.1886, |
| "step": 1501 |
| }, |
| { |
| "epoch": 2.33, |
| "learning_rate": 1.1180773249738766e-05, |
| "loss": 0.182, |
| "step": 1502 |
| }, |
| { |
| "epoch": 2.34, |
| "learning_rate": 1.1154649947753397e-05, |
| "loss": 0.1893, |
| "step": 1503 |
| }, |
| { |
| "epoch": 2.34, |
| "learning_rate": 1.1128526645768025e-05, |
| "loss": 0.166, |
| "step": 1504 |
| }, |
| { |
| "epoch": 2.34, |
| "learning_rate": 1.1102403343782656e-05, |
| "loss": 0.1892, |
| "step": 1505 |
| }, |
| { |
| "epoch": 2.34, |
| "learning_rate": 1.1076280041797284e-05, |
| "loss": 0.1881, |
| "step": 1506 |
| }, |
| { |
| "epoch": 2.34, |
| "learning_rate": 1.1050156739811913e-05, |
| "loss": 0.165, |
| "step": 1507 |
| }, |
| { |
| "epoch": 2.34, |
| "learning_rate": 1.1024033437826541e-05, |
| "loss": 0.1856, |
| "step": 1508 |
| }, |
| { |
| "epoch": 2.35, |
| "learning_rate": 1.099791013584117e-05, |
| "loss": 0.1791, |
| "step": 1509 |
| }, |
| { |
| "epoch": 2.35, |
| "learning_rate": 1.09717868338558e-05, |
| "loss": 0.1834, |
| "step": 1510 |
| }, |
| { |
| "epoch": 2.35, |
| "learning_rate": 1.0945663531870429e-05, |
| "loss": 0.1885, |
| "step": 1511 |
| }, |
| { |
| "epoch": 2.35, |
| "learning_rate": 1.091954022988506e-05, |
| "loss": 0.18, |
| "step": 1512 |
| }, |
| { |
| "epoch": 2.35, |
| "learning_rate": 1.0893416927899688e-05, |
| "loss": 0.1678, |
| "step": 1513 |
| }, |
| { |
| "epoch": 2.35, |
| "learning_rate": 1.0867293625914315e-05, |
| "loss": 0.1829, |
| "step": 1514 |
| }, |
| { |
| "epoch": 2.35, |
| "learning_rate": 1.0841170323928945e-05, |
| "loss": 0.2008, |
| "step": 1515 |
| }, |
| { |
| "epoch": 2.36, |
| "learning_rate": 1.0815047021943574e-05, |
| "loss": 0.1795, |
| "step": 1516 |
| }, |
| { |
| "epoch": 2.36, |
| "learning_rate": 1.0788923719958204e-05, |
| "loss": 0.185, |
| "step": 1517 |
| }, |
| { |
| "epoch": 2.36, |
| "learning_rate": 1.0762800417972832e-05, |
| "loss": 0.1788, |
| "step": 1518 |
| }, |
| { |
| "epoch": 2.36, |
| "learning_rate": 1.0736677115987463e-05, |
| "loss": 0.1854, |
| "step": 1519 |
| }, |
| { |
| "epoch": 2.36, |
| "learning_rate": 1.071055381400209e-05, |
| "loss": 0.1843, |
| "step": 1520 |
| }, |
| { |
| "epoch": 2.36, |
| "learning_rate": 1.0684430512016718e-05, |
| "loss": 0.1887, |
| "step": 1521 |
| }, |
| { |
| "epoch": 2.37, |
| "learning_rate": 1.0658307210031348e-05, |
| "loss": 0.1844, |
| "step": 1522 |
| }, |
| { |
| "epoch": 2.37, |
| "learning_rate": 1.0632183908045977e-05, |
| "loss": 0.1871, |
| "step": 1523 |
| }, |
| { |
| "epoch": 2.37, |
| "learning_rate": 1.0606060606060607e-05, |
| "loss": 0.1649, |
| "step": 1524 |
| }, |
| { |
| "epoch": 2.37, |
| "learning_rate": 1.0579937304075236e-05, |
| "loss": 0.2, |
| "step": 1525 |
| }, |
| { |
| "epoch": 2.37, |
| "learning_rate": 1.0553814002089865e-05, |
| "loss": 0.1663, |
| "step": 1526 |
| }, |
| { |
| "epoch": 2.37, |
| "learning_rate": 1.0527690700104493e-05, |
| "loss": 0.175, |
| "step": 1527 |
| }, |
| { |
| "epoch": 2.37, |
| "learning_rate": 1.0501567398119123e-05, |
| "loss": 0.1706, |
| "step": 1528 |
| }, |
| { |
| "epoch": 2.38, |
| "learning_rate": 1.0475444096133752e-05, |
| "loss": 0.1855, |
| "step": 1529 |
| }, |
| { |
| "epoch": 2.38, |
| "learning_rate": 1.044932079414838e-05, |
| "loss": 0.1848, |
| "step": 1530 |
| }, |
| { |
| "epoch": 2.38, |
| "learning_rate": 1.0423197492163011e-05, |
| "loss": 0.1724, |
| "step": 1531 |
| }, |
| { |
| "epoch": 2.38, |
| "learning_rate": 1.039707419017764e-05, |
| "loss": 0.1759, |
| "step": 1532 |
| }, |
| { |
| "epoch": 2.38, |
| "learning_rate": 1.0370950888192268e-05, |
| "loss": 0.1889, |
| "step": 1533 |
| }, |
| { |
| "epoch": 2.38, |
| "learning_rate": 1.0344827586206897e-05, |
| "loss": 0.1907, |
| "step": 1534 |
| }, |
| { |
| "epoch": 2.39, |
| "learning_rate": 1.0318704284221527e-05, |
| "loss": 0.1825, |
| "step": 1535 |
| }, |
| { |
| "epoch": 2.39, |
| "learning_rate": 1.0292580982236156e-05, |
| "loss": 0.1813, |
| "step": 1536 |
| }, |
| { |
| "epoch": 2.39, |
| "learning_rate": 1.0266457680250784e-05, |
| "loss": 0.182, |
| "step": 1537 |
| }, |
| { |
| "epoch": 2.39, |
| "learning_rate": 1.0240334378265413e-05, |
| "loss": 0.1878, |
| "step": 1538 |
| }, |
| { |
| "epoch": 2.39, |
| "learning_rate": 1.0214211076280041e-05, |
| "loss": 0.1731, |
| "step": 1539 |
| }, |
| { |
| "epoch": 2.39, |
| "learning_rate": 1.0188087774294672e-05, |
| "loss": 0.1876, |
| "step": 1540 |
| }, |
| { |
| "epoch": 2.4, |
| "learning_rate": 1.01619644723093e-05, |
| "loss": 0.183, |
| "step": 1541 |
| }, |
| { |
| "epoch": 2.4, |
| "learning_rate": 1.013584117032393e-05, |
| "loss": 0.1823, |
| "step": 1542 |
| }, |
| { |
| "epoch": 2.4, |
| "learning_rate": 1.0109717868338559e-05, |
| "loss": 0.1867, |
| "step": 1543 |
| }, |
| { |
| "epoch": 2.4, |
| "learning_rate": 1.0083594566353188e-05, |
| "loss": 0.1706, |
| "step": 1544 |
| }, |
| { |
| "epoch": 2.4, |
| "learning_rate": 1.0057471264367816e-05, |
| "loss": 0.1776, |
| "step": 1545 |
| }, |
| { |
| "epoch": 2.4, |
| "learning_rate": 1.0031347962382445e-05, |
| "loss": 0.1751, |
| "step": 1546 |
| }, |
| { |
| "epoch": 2.4, |
| "learning_rate": 1.0005224660397075e-05, |
| "loss": 0.19, |
| "step": 1547 |
| }, |
| { |
| "epoch": 2.41, |
| "learning_rate": 9.979101358411704e-06, |
| "loss": 0.1789, |
| "step": 1548 |
| }, |
| { |
| "epoch": 2.41, |
| "learning_rate": 9.952978056426334e-06, |
| "loss": 0.1819, |
| "step": 1549 |
| }, |
| { |
| "epoch": 2.41, |
| "learning_rate": 9.926854754440963e-06, |
| "loss": 0.1839, |
| "step": 1550 |
| }, |
| { |
| "epoch": 2.41, |
| "learning_rate": 9.90073145245559e-06, |
| "loss": 0.1739, |
| "step": 1551 |
| }, |
| { |
| "epoch": 2.41, |
| "learning_rate": 9.87460815047022e-06, |
| "loss": 0.1697, |
| "step": 1552 |
| }, |
| { |
| "epoch": 2.41, |
| "learning_rate": 9.848484848484848e-06, |
| "loss": 0.1729, |
| "step": 1553 |
| }, |
| { |
| "epoch": 2.42, |
| "learning_rate": 9.822361546499479e-06, |
| "loss": 0.1725, |
| "step": 1554 |
| }, |
| { |
| "epoch": 2.42, |
| "learning_rate": 9.796238244514107e-06, |
| "loss": 0.1868, |
| "step": 1555 |
| }, |
| { |
| "epoch": 2.42, |
| "learning_rate": 9.770114942528738e-06, |
| "loss": 0.18, |
| "step": 1556 |
| }, |
| { |
| "epoch": 2.42, |
| "learning_rate": 9.743991640543364e-06, |
| "loss": 0.1709, |
| "step": 1557 |
| }, |
| { |
| "epoch": 2.42, |
| "learning_rate": 9.717868338557993e-06, |
| "loss": 0.1816, |
| "step": 1558 |
| }, |
| { |
| "epoch": 2.42, |
| "learning_rate": 9.691745036572623e-06, |
| "loss": 0.1876, |
| "step": 1559 |
| }, |
| { |
| "epoch": 2.42, |
| "learning_rate": 9.665621734587252e-06, |
| "loss": 0.199, |
| "step": 1560 |
| }, |
| { |
| "epoch": 2.43, |
| "learning_rate": 9.639498432601882e-06, |
| "loss": 0.1807, |
| "step": 1561 |
| }, |
| { |
| "epoch": 2.43, |
| "learning_rate": 9.61337513061651e-06, |
| "loss": 0.1759, |
| "step": 1562 |
| }, |
| { |
| "epoch": 2.43, |
| "learning_rate": 9.58725182863114e-06, |
| "loss": 0.179, |
| "step": 1563 |
| }, |
| { |
| "epoch": 2.43, |
| "learning_rate": 9.561128526645768e-06, |
| "loss": 0.1895, |
| "step": 1564 |
| }, |
| { |
| "epoch": 2.43, |
| "learning_rate": 9.535005224660397e-06, |
| "loss": 0.1848, |
| "step": 1565 |
| }, |
| { |
| "epoch": 2.43, |
| "learning_rate": 9.508881922675027e-06, |
| "loss": 0.1745, |
| "step": 1566 |
| }, |
| { |
| "epoch": 2.44, |
| "learning_rate": 9.482758620689655e-06, |
| "loss": 0.1855, |
| "step": 1567 |
| }, |
| { |
| "epoch": 2.44, |
| "learning_rate": 9.456635318704286e-06, |
| "loss": 0.1669, |
| "step": 1568 |
| }, |
| { |
| "epoch": 2.44, |
| "learning_rate": 9.430512016718914e-06, |
| "loss": 0.1799, |
| "step": 1569 |
| }, |
| { |
| "epoch": 2.44, |
| "learning_rate": 9.404388714733543e-06, |
| "loss": 0.1809, |
| "step": 1570 |
| }, |
| { |
| "epoch": 2.44, |
| "learning_rate": 9.378265412748171e-06, |
| "loss": 0.1801, |
| "step": 1571 |
| }, |
| { |
| "epoch": 2.44, |
| "learning_rate": 9.352142110762802e-06, |
| "loss": 0.1788, |
| "step": 1572 |
| }, |
| { |
| "epoch": 2.44, |
| "learning_rate": 9.32601880877743e-06, |
| "loss": 0.179, |
| "step": 1573 |
| }, |
| { |
| "epoch": 2.45, |
| "learning_rate": 9.299895506792059e-06, |
| "loss": 0.1783, |
| "step": 1574 |
| }, |
| { |
| "epoch": 2.45, |
| "learning_rate": 9.273772204806688e-06, |
| "loss": 0.1893, |
| "step": 1575 |
| }, |
| { |
| "epoch": 2.45, |
| "learning_rate": 9.247648902821316e-06, |
| "loss": 0.1872, |
| "step": 1576 |
| }, |
| { |
| "epoch": 2.45, |
| "learning_rate": 9.221525600835946e-06, |
| "loss": 0.1817, |
| "step": 1577 |
| }, |
| { |
| "epoch": 2.45, |
| "learning_rate": 9.195402298850575e-06, |
| "loss": 0.1761, |
| "step": 1578 |
| }, |
| { |
| "epoch": 2.45, |
| "learning_rate": 9.169278996865205e-06, |
| "loss": 0.174, |
| "step": 1579 |
| }, |
| { |
| "epoch": 2.46, |
| "learning_rate": 9.143155694879834e-06, |
| "loss": 0.1835, |
| "step": 1580 |
| }, |
| { |
| "epoch": 2.46, |
| "learning_rate": 9.117032392894462e-06, |
| "loss": 0.1922, |
| "step": 1581 |
| }, |
| { |
| "epoch": 2.46, |
| "learning_rate": 9.090909090909091e-06, |
| "loss": 0.1708, |
| "step": 1582 |
| }, |
| { |
| "epoch": 2.46, |
| "learning_rate": 9.06478578892372e-06, |
| "loss": 0.1758, |
| "step": 1583 |
| }, |
| { |
| "epoch": 2.46, |
| "learning_rate": 9.03866248693835e-06, |
| "loss": 0.2029, |
| "step": 1584 |
| }, |
| { |
| "epoch": 2.46, |
| "learning_rate": 9.012539184952979e-06, |
| "loss": 0.18, |
| "step": 1585 |
| }, |
| { |
| "epoch": 2.46, |
| "learning_rate": 8.986415882967609e-06, |
| "loss": 0.1885, |
| "step": 1586 |
| }, |
| { |
| "epoch": 2.47, |
| "learning_rate": 8.960292580982237e-06, |
| "loss": 0.183, |
| "step": 1587 |
| }, |
| { |
| "epoch": 2.47, |
| "learning_rate": 8.934169278996864e-06, |
| "loss": 0.1635, |
| "step": 1588 |
| }, |
| { |
| "epoch": 2.47, |
| "learning_rate": 8.908045977011495e-06, |
| "loss": 0.1761, |
| "step": 1589 |
| }, |
| { |
| "epoch": 2.47, |
| "learning_rate": 8.881922675026123e-06, |
| "loss": 0.1764, |
| "step": 1590 |
| }, |
| { |
| "epoch": 2.47, |
| "learning_rate": 8.855799373040753e-06, |
| "loss": 0.1773, |
| "step": 1591 |
| }, |
| { |
| "epoch": 2.47, |
| "learning_rate": 8.829676071055382e-06, |
| "loss": 0.1771, |
| "step": 1592 |
| }, |
| { |
| "epoch": 2.48, |
| "learning_rate": 8.803552769070012e-06, |
| "loss": 0.1683, |
| "step": 1593 |
| }, |
| { |
| "epoch": 2.48, |
| "learning_rate": 8.77742946708464e-06, |
| "loss": 0.1805, |
| "step": 1594 |
| }, |
| { |
| "epoch": 2.48, |
| "learning_rate": 8.751306165099268e-06, |
| "loss": 0.1888, |
| "step": 1595 |
| }, |
| { |
| "epoch": 2.48, |
| "learning_rate": 8.725182863113898e-06, |
| "loss": 0.1886, |
| "step": 1596 |
| }, |
| { |
| "epoch": 2.48, |
| "learning_rate": 8.699059561128527e-06, |
| "loss": 0.1785, |
| "step": 1597 |
| }, |
| { |
| "epoch": 2.48, |
| "learning_rate": 8.672936259143157e-06, |
| "loss": 0.1895, |
| "step": 1598 |
| }, |
| { |
| "epoch": 2.49, |
| "learning_rate": 8.646812957157786e-06, |
| "loss": 0.1668, |
| "step": 1599 |
| }, |
| { |
| "epoch": 2.49, |
| "learning_rate": 8.620689655172414e-06, |
| "loss": 0.1779, |
| "step": 1600 |
| }, |
| { |
| "epoch": 2.49, |
| "learning_rate": 8.594566353187043e-06, |
| "loss": 0.1897, |
| "step": 1601 |
| }, |
| { |
| "epoch": 2.49, |
| "learning_rate": 8.568443051201671e-06, |
| "loss": 0.1892, |
| "step": 1602 |
| }, |
| { |
| "epoch": 2.49, |
| "learning_rate": 8.542319749216302e-06, |
| "loss": 0.1848, |
| "step": 1603 |
| }, |
| { |
| "epoch": 2.49, |
| "learning_rate": 8.51619644723093e-06, |
| "loss": 0.1737, |
| "step": 1604 |
| }, |
| { |
| "epoch": 2.49, |
| "learning_rate": 8.49007314524556e-06, |
| "loss": 0.2052, |
| "step": 1605 |
| }, |
| { |
| "epoch": 2.5, |
| "learning_rate": 8.463949843260189e-06, |
| "loss": 0.1782, |
| "step": 1606 |
| }, |
| { |
| "epoch": 2.5, |
| "learning_rate": 8.437826541274818e-06, |
| "loss": 0.1818, |
| "step": 1607 |
| }, |
| { |
| "epoch": 2.5, |
| "learning_rate": 8.411703239289446e-06, |
| "loss": 0.1618, |
| "step": 1608 |
| }, |
| { |
| "epoch": 2.5, |
| "learning_rate": 8.385579937304075e-06, |
| "loss": 0.1746, |
| "step": 1609 |
| }, |
| { |
| "epoch": 2.5, |
| "learning_rate": 8.359456635318705e-06, |
| "loss": 0.1887, |
| "step": 1610 |
| }, |
| { |
| "epoch": 2.5, |
| "learning_rate": 8.333333333333334e-06, |
| "loss": 0.196, |
| "step": 1611 |
| }, |
| { |
| "epoch": 2.51, |
| "learning_rate": 8.307210031347962e-06, |
| "loss": 0.186, |
| "step": 1612 |
| }, |
| { |
| "epoch": 2.51, |
| "learning_rate": 8.281086729362591e-06, |
| "loss": 0.1745, |
| "step": 1613 |
| }, |
| { |
| "epoch": 2.51, |
| "learning_rate": 8.254963427377221e-06, |
| "loss": 0.1705, |
| "step": 1614 |
| }, |
| { |
| "epoch": 2.51, |
| "learning_rate": 8.22884012539185e-06, |
| "loss": 0.1801, |
| "step": 1615 |
| }, |
| { |
| "epoch": 2.51, |
| "learning_rate": 8.202716823406478e-06, |
| "loss": 0.1873, |
| "step": 1616 |
| }, |
| { |
| "epoch": 2.51, |
| "learning_rate": 8.176593521421109e-06, |
| "loss": 0.1746, |
| "step": 1617 |
| }, |
| { |
| "epoch": 2.51, |
| "learning_rate": 8.150470219435737e-06, |
| "loss": 0.1731, |
| "step": 1618 |
| }, |
| { |
| "epoch": 2.52, |
| "learning_rate": 8.124346917450366e-06, |
| "loss": 0.1769, |
| "step": 1619 |
| }, |
| { |
| "epoch": 2.52, |
| "learning_rate": 8.098223615464994e-06, |
| "loss": 0.172, |
| "step": 1620 |
| }, |
| { |
| "epoch": 2.52, |
| "learning_rate": 8.072100313479625e-06, |
| "loss": 0.1871, |
| "step": 1621 |
| }, |
| { |
| "epoch": 2.52, |
| "learning_rate": 8.045977011494253e-06, |
| "loss": 0.18, |
| "step": 1622 |
| }, |
| { |
| "epoch": 2.52, |
| "learning_rate": 8.019853709508884e-06, |
| "loss": 0.1895, |
| "step": 1623 |
| }, |
| { |
| "epoch": 2.52, |
| "learning_rate": 7.993730407523512e-06, |
| "loss": 0.1872, |
| "step": 1624 |
| }, |
| { |
| "epoch": 2.53, |
| "learning_rate": 7.967607105538139e-06, |
| "loss": 0.1764, |
| "step": 1625 |
| }, |
| { |
| "epoch": 2.53, |
| "learning_rate": 7.94148380355277e-06, |
| "loss": 0.1846, |
| "step": 1626 |
| }, |
| { |
| "epoch": 2.53, |
| "learning_rate": 7.915360501567398e-06, |
| "loss": 0.1767, |
| "step": 1627 |
| }, |
| { |
| "epoch": 2.53, |
| "learning_rate": 7.889237199582028e-06, |
| "loss": 0.1699, |
| "step": 1628 |
| }, |
| { |
| "epoch": 2.53, |
| "learning_rate": 7.863113897596657e-06, |
| "loss": 0.1722, |
| "step": 1629 |
| }, |
| { |
| "epoch": 2.53, |
| "learning_rate": 7.836990595611287e-06, |
| "loss": 0.1774, |
| "step": 1630 |
| }, |
| { |
| "epoch": 2.53, |
| "learning_rate": 7.810867293625914e-06, |
| "loss": 0.1769, |
| "step": 1631 |
| }, |
| { |
| "epoch": 2.54, |
| "learning_rate": 7.784743991640543e-06, |
| "loss": 0.1845, |
| "step": 1632 |
| }, |
| { |
| "epoch": 2.54, |
| "learning_rate": 7.758620689655173e-06, |
| "loss": 0.177, |
| "step": 1633 |
| }, |
| { |
| "epoch": 2.54, |
| "learning_rate": 7.732497387669801e-06, |
| "loss": 0.1948, |
| "step": 1634 |
| }, |
| { |
| "epoch": 2.54, |
| "learning_rate": 7.706374085684432e-06, |
| "loss": 0.1827, |
| "step": 1635 |
| }, |
| { |
| "epoch": 2.54, |
| "learning_rate": 7.68025078369906e-06, |
| "loss": 0.1769, |
| "step": 1636 |
| }, |
| { |
| "epoch": 2.54, |
| "learning_rate": 7.654127481713689e-06, |
| "loss": 0.1786, |
| "step": 1637 |
| }, |
| { |
| "epoch": 2.55, |
| "learning_rate": 7.628004179728318e-06, |
| "loss": 0.1657, |
| "step": 1638 |
| }, |
| { |
| "epoch": 2.55, |
| "learning_rate": 7.601880877742946e-06, |
| "loss": 0.1807, |
| "step": 1639 |
| }, |
| { |
| "epoch": 2.55, |
| "learning_rate": 7.5757575757575764e-06, |
| "loss": 0.1768, |
| "step": 1640 |
| }, |
| { |
| "epoch": 2.55, |
| "learning_rate": 7.549634273772205e-06, |
| "loss": 0.1797, |
| "step": 1641 |
| }, |
| { |
| "epoch": 2.55, |
| "learning_rate": 7.5235109717868345e-06, |
| "loss": 0.1914, |
| "step": 1642 |
| }, |
| { |
| "epoch": 2.55, |
| "learning_rate": 7.497387669801463e-06, |
| "loss": 0.1838, |
| "step": 1643 |
| }, |
| { |
| "epoch": 2.56, |
| "learning_rate": 7.4712643678160925e-06, |
| "loss": 0.1814, |
| "step": 1644 |
| }, |
| { |
| "epoch": 2.56, |
| "learning_rate": 7.445141065830721e-06, |
| "loss": 0.1658, |
| "step": 1645 |
| }, |
| { |
| "epoch": 2.56, |
| "learning_rate": 7.41901776384535e-06, |
| "loss": 0.1882, |
| "step": 1646 |
| }, |
| { |
| "epoch": 2.56, |
| "learning_rate": 7.39289446185998e-06, |
| "loss": 0.1908, |
| "step": 1647 |
| }, |
| { |
| "epoch": 2.56, |
| "learning_rate": 7.366771159874608e-06, |
| "loss": 0.1755, |
| "step": 1648 |
| }, |
| { |
| "epoch": 2.56, |
| "learning_rate": 7.340647857889238e-06, |
| "loss": 0.1809, |
| "step": 1649 |
| }, |
| { |
| "epoch": 2.56, |
| "learning_rate": 7.3145245559038666e-06, |
| "loss": 0.1779, |
| "step": 1650 |
| }, |
| { |
| "epoch": 2.57, |
| "learning_rate": 7.288401253918496e-06, |
| "loss": 0.1714, |
| "step": 1651 |
| }, |
| { |
| "epoch": 2.57, |
| "learning_rate": 7.262277951933125e-06, |
| "loss": 0.1776, |
| "step": 1652 |
| }, |
| { |
| "epoch": 2.57, |
| "learning_rate": 7.236154649947753e-06, |
| "loss": 0.1909, |
| "step": 1653 |
| }, |
| { |
| "epoch": 2.57, |
| "learning_rate": 7.210031347962383e-06, |
| "loss": 0.1848, |
| "step": 1654 |
| }, |
| { |
| "epoch": 2.57, |
| "learning_rate": 7.183908045977011e-06, |
| "loss": 0.1717, |
| "step": 1655 |
| }, |
| { |
| "epoch": 2.57, |
| "learning_rate": 7.1577847439916415e-06, |
| "loss": 0.1769, |
| "step": 1656 |
| }, |
| { |
| "epoch": 2.58, |
| "learning_rate": 7.131661442006269e-06, |
| "loss": 0.175, |
| "step": 1657 |
| }, |
| { |
| "epoch": 2.58, |
| "learning_rate": 7.1055381400208995e-06, |
| "loss": 0.1827, |
| "step": 1658 |
| }, |
| { |
| "epoch": 2.58, |
| "learning_rate": 7.079414838035528e-06, |
| "loss": 0.176, |
| "step": 1659 |
| }, |
| { |
| "epoch": 2.58, |
| "learning_rate": 7.053291536050157e-06, |
| "loss": 0.191, |
| "step": 1660 |
| }, |
| { |
| "epoch": 2.58, |
| "learning_rate": 7.027168234064786e-06, |
| "loss": 0.1781, |
| "step": 1661 |
| }, |
| { |
| "epoch": 2.58, |
| "learning_rate": 7.001044932079415e-06, |
| "loss": 0.1681, |
| "step": 1662 |
| }, |
| { |
| "epoch": 2.58, |
| "learning_rate": 6.974921630094044e-06, |
| "loss": 0.1831, |
| "step": 1663 |
| }, |
| { |
| "epoch": 2.59, |
| "learning_rate": 6.948798328108673e-06, |
| "loss": 0.1679, |
| "step": 1664 |
| }, |
| { |
| "epoch": 2.59, |
| "learning_rate": 6.922675026123303e-06, |
| "loss": 0.1747, |
| "step": 1665 |
| }, |
| { |
| "epoch": 2.59, |
| "learning_rate": 6.896551724137932e-06, |
| "loss": 0.175, |
| "step": 1666 |
| }, |
| { |
| "epoch": 2.59, |
| "learning_rate": 6.870428422152561e-06, |
| "loss": 0.1788, |
| "step": 1667 |
| }, |
| { |
| "epoch": 2.59, |
| "learning_rate": 6.84430512016719e-06, |
| "loss": 0.1795, |
| "step": 1668 |
| }, |
| { |
| "epoch": 2.59, |
| "learning_rate": 6.818181818181818e-06, |
| "loss": 0.1785, |
| "step": 1669 |
| }, |
| { |
| "epoch": 2.6, |
| "learning_rate": 6.792058516196448e-06, |
| "loss": 0.1795, |
| "step": 1670 |
| }, |
| { |
| "epoch": 2.6, |
| "learning_rate": 6.765935214211076e-06, |
| "loss": 0.1725, |
| "step": 1671 |
| }, |
| { |
| "epoch": 2.6, |
| "learning_rate": 6.739811912225706e-06, |
| "loss": 0.1687, |
| "step": 1672 |
| }, |
| { |
| "epoch": 2.6, |
| "learning_rate": 6.713688610240334e-06, |
| "loss": 0.1663, |
| "step": 1673 |
| }, |
| { |
| "epoch": 2.6, |
| "learning_rate": 6.687565308254965e-06, |
| "loss": 0.1868, |
| "step": 1674 |
| }, |
| { |
| "epoch": 2.6, |
| "learning_rate": 6.661442006269593e-06, |
| "loss": 0.1827, |
| "step": 1675 |
| }, |
| { |
| "epoch": 2.6, |
| "learning_rate": 6.635318704284221e-06, |
| "loss": 0.1784, |
| "step": 1676 |
| }, |
| { |
| "epoch": 2.61, |
| "learning_rate": 6.609195402298851e-06, |
| "loss": 0.1744, |
| "step": 1677 |
| }, |
| { |
| "epoch": 2.61, |
| "learning_rate": 6.58307210031348e-06, |
| "loss": 0.1727, |
| "step": 1678 |
| }, |
| { |
| "epoch": 2.61, |
| "learning_rate": 6.556948798328109e-06, |
| "loss": 0.167, |
| "step": 1679 |
| }, |
| { |
| "epoch": 2.61, |
| "learning_rate": 6.530825496342738e-06, |
| "loss": 0.1768, |
| "step": 1680 |
| }, |
| { |
| "epoch": 2.61, |
| "learning_rate": 6.504702194357367e-06, |
| "loss": 0.1862, |
| "step": 1681 |
| }, |
| { |
| "epoch": 2.61, |
| "learning_rate": 6.478578892371996e-06, |
| "loss": 0.1688, |
| "step": 1682 |
| }, |
| { |
| "epoch": 2.62, |
| "learning_rate": 6.4524555903866245e-06, |
| "loss": 0.1789, |
| "step": 1683 |
| }, |
| { |
| "epoch": 2.62, |
| "learning_rate": 6.426332288401255e-06, |
| "loss": 0.1649, |
| "step": 1684 |
| }, |
| { |
| "epoch": 2.62, |
| "learning_rate": 6.4002089864158825e-06, |
| "loss": 0.1915, |
| "step": 1685 |
| }, |
| { |
| "epoch": 2.62, |
| "learning_rate": 6.374085684430513e-06, |
| "loss": 0.1725, |
| "step": 1686 |
| }, |
| { |
| "epoch": 2.62, |
| "learning_rate": 6.347962382445141e-06, |
| "loss": 0.1606, |
| "step": 1687 |
| }, |
| { |
| "epoch": 2.62, |
| "learning_rate": 6.321839080459771e-06, |
| "loss": 0.1862, |
| "step": 1688 |
| }, |
| { |
| "epoch": 2.62, |
| "learning_rate": 6.295715778474399e-06, |
| "loss": 0.1714, |
| "step": 1689 |
| }, |
| { |
| "epoch": 2.63, |
| "learning_rate": 6.269592476489028e-06, |
| "loss": 0.1834, |
| "step": 1690 |
| }, |
| { |
| "epoch": 2.63, |
| "learning_rate": 6.2434691745036574e-06, |
| "loss": 0.1735, |
| "step": 1691 |
| }, |
| { |
| "epoch": 2.63, |
| "learning_rate": 6.217345872518287e-06, |
| "loss": 0.1731, |
| "step": 1692 |
| }, |
| { |
| "epoch": 2.63, |
| "learning_rate": 6.1912225705329155e-06, |
| "loss": 0.1764, |
| "step": 1693 |
| }, |
| { |
| "epoch": 2.63, |
| "learning_rate": 6.165099268547545e-06, |
| "loss": 0.1877, |
| "step": 1694 |
| }, |
| { |
| "epoch": 2.63, |
| "learning_rate": 6.1389759665621735e-06, |
| "loss": 0.1742, |
| "step": 1695 |
| }, |
| { |
| "epoch": 2.64, |
| "learning_rate": 6.112852664576803e-06, |
| "loss": 0.1659, |
| "step": 1696 |
| }, |
| { |
| "epoch": 2.64, |
| "learning_rate": 6.0867293625914315e-06, |
| "loss": 0.179, |
| "step": 1697 |
| }, |
| { |
| "epoch": 2.64, |
| "learning_rate": 6.060606060606061e-06, |
| "loss": 0.1707, |
| "step": 1698 |
| }, |
| { |
| "epoch": 2.64, |
| "learning_rate": 6.03448275862069e-06, |
| "loss": 0.1693, |
| "step": 1699 |
| }, |
| { |
| "epoch": 2.64, |
| "learning_rate": 6.008359456635319e-06, |
| "loss": 0.1669, |
| "step": 1700 |
| }, |
| { |
| "epoch": 2.64, |
| "learning_rate": 5.982236154649948e-06, |
| "loss": 0.1763, |
| "step": 1701 |
| }, |
| { |
| "epoch": 2.65, |
| "learning_rate": 5.956112852664577e-06, |
| "loss": 0.1712, |
| "step": 1702 |
| }, |
| { |
| "epoch": 2.65, |
| "learning_rate": 5.9299895506792065e-06, |
| "loss": 0.176, |
| "step": 1703 |
| }, |
| { |
| "epoch": 2.65, |
| "learning_rate": 5.903866248693835e-06, |
| "loss": 0.1698, |
| "step": 1704 |
| }, |
| { |
| "epoch": 2.65, |
| "learning_rate": 5.8777429467084645e-06, |
| "loss": 0.173, |
| "step": 1705 |
| }, |
| { |
| "epoch": 2.65, |
| "learning_rate": 5.851619644723093e-06, |
| "loss": 0.1862, |
| "step": 1706 |
| }, |
| { |
| "epoch": 2.65, |
| "learning_rate": 5.8254963427377225e-06, |
| "loss": 0.1721, |
| "step": 1707 |
| }, |
| { |
| "epoch": 2.65, |
| "learning_rate": 5.799373040752351e-06, |
| "loss": 0.1661, |
| "step": 1708 |
| }, |
| { |
| "epoch": 2.66, |
| "learning_rate": 5.7732497387669806e-06, |
| "loss": 0.1759, |
| "step": 1709 |
| }, |
| { |
| "epoch": 2.66, |
| "learning_rate": 5.747126436781609e-06, |
| "loss": 0.1762, |
| "step": 1710 |
| }, |
| { |
| "epoch": 2.66, |
| "learning_rate": 5.721003134796239e-06, |
| "loss": 0.1757, |
| "step": 1711 |
| }, |
| { |
| "epoch": 2.66, |
| "learning_rate": 5.694879832810868e-06, |
| "loss": 0.1843, |
| "step": 1712 |
| }, |
| { |
| "epoch": 2.66, |
| "learning_rate": 5.668756530825497e-06, |
| "loss": 0.1731, |
| "step": 1713 |
| }, |
| { |
| "epoch": 2.66, |
| "learning_rate": 5.642633228840126e-06, |
| "loss": 0.1768, |
| "step": 1714 |
| }, |
| { |
| "epoch": 2.67, |
| "learning_rate": 5.616509926854755e-06, |
| "loss": 0.1738, |
| "step": 1715 |
| }, |
| { |
| "epoch": 2.67, |
| "learning_rate": 5.590386624869383e-06, |
| "loss": 0.1778, |
| "step": 1716 |
| }, |
| { |
| "epoch": 2.67, |
| "learning_rate": 5.564263322884013e-06, |
| "loss": 0.1822, |
| "step": 1717 |
| }, |
| { |
| "epoch": 2.67, |
| "learning_rate": 5.538140020898642e-06, |
| "loss": 0.173, |
| "step": 1718 |
| }, |
| { |
| "epoch": 2.67, |
| "learning_rate": 5.512016718913271e-06, |
| "loss": 0.1725, |
| "step": 1719 |
| }, |
| { |
| "epoch": 2.67, |
| "learning_rate": 5.4858934169279e-06, |
| "loss": 0.1789, |
| "step": 1720 |
| }, |
| { |
| "epoch": 2.67, |
| "learning_rate": 5.45977011494253e-06, |
| "loss": 0.1797, |
| "step": 1721 |
| }, |
| { |
| "epoch": 2.68, |
| "learning_rate": 5.433646812957157e-06, |
| "loss": 0.1689, |
| "step": 1722 |
| }, |
| { |
| "epoch": 2.68, |
| "learning_rate": 5.407523510971787e-06, |
| "loss": 0.1956, |
| "step": 1723 |
| }, |
| { |
| "epoch": 2.68, |
| "learning_rate": 5.381400208986416e-06, |
| "loss": 0.1696, |
| "step": 1724 |
| }, |
| { |
| "epoch": 2.68, |
| "learning_rate": 5.355276907001045e-06, |
| "loss": 0.1755, |
| "step": 1725 |
| }, |
| { |
| "epoch": 2.68, |
| "learning_rate": 5.329153605015674e-06, |
| "loss": 0.1656, |
| "step": 1726 |
| }, |
| { |
| "epoch": 2.68, |
| "learning_rate": 5.303030303030304e-06, |
| "loss": 0.1661, |
| "step": 1727 |
| }, |
| { |
| "epoch": 2.69, |
| "learning_rate": 5.276907001044932e-06, |
| "loss": 0.1748, |
| "step": 1728 |
| }, |
| { |
| "epoch": 2.69, |
| "learning_rate": 5.250783699059562e-06, |
| "loss": 0.1636, |
| "step": 1729 |
| }, |
| { |
| "epoch": 2.69, |
| "learning_rate": 5.22466039707419e-06, |
| "loss": 0.1805, |
| "step": 1730 |
| }, |
| { |
| "epoch": 2.69, |
| "learning_rate": 5.19853709508882e-06, |
| "loss": 0.1795, |
| "step": 1731 |
| }, |
| { |
| "epoch": 2.69, |
| "learning_rate": 5.172413793103448e-06, |
| "loss": 0.1816, |
| "step": 1732 |
| }, |
| { |
| "epoch": 2.69, |
| "learning_rate": 5.146290491118078e-06, |
| "loss": 0.1728, |
| "step": 1733 |
| }, |
| { |
| "epoch": 2.69, |
| "learning_rate": 5.120167189132706e-06, |
| "loss": 0.1802, |
| "step": 1734 |
| }, |
| { |
| "epoch": 2.7, |
| "learning_rate": 5.094043887147336e-06, |
| "loss": 0.1778, |
| "step": 1735 |
| }, |
| { |
| "epoch": 2.7, |
| "learning_rate": 5.067920585161965e-06, |
| "loss": 0.1683, |
| "step": 1736 |
| }, |
| { |
| "epoch": 2.7, |
| "learning_rate": 5.041797283176594e-06, |
| "loss": 0.1682, |
| "step": 1737 |
| }, |
| { |
| "epoch": 2.7, |
| "learning_rate": 5.015673981191222e-06, |
| "loss": 0.1729, |
| "step": 1738 |
| }, |
| { |
| "epoch": 2.7, |
| "learning_rate": 4.989550679205852e-06, |
| "loss": 0.1685, |
| "step": 1739 |
| }, |
| { |
| "epoch": 2.7, |
| "learning_rate": 4.963427377220481e-06, |
| "loss": 0.1598, |
| "step": 1740 |
| }, |
| { |
| "epoch": 2.71, |
| "learning_rate": 4.93730407523511e-06, |
| "loss": 0.1774, |
| "step": 1741 |
| }, |
| { |
| "epoch": 2.71, |
| "learning_rate": 4.911180773249739e-06, |
| "loss": 0.1786, |
| "step": 1742 |
| }, |
| { |
| "epoch": 2.71, |
| "learning_rate": 4.885057471264369e-06, |
| "loss": 0.1824, |
| "step": 1743 |
| }, |
| { |
| "epoch": 2.71, |
| "learning_rate": 4.8589341692789965e-06, |
| "loss": 0.1686, |
| "step": 1744 |
| }, |
| { |
| "epoch": 2.71, |
| "learning_rate": 4.832810867293626e-06, |
| "loss": 0.1774, |
| "step": 1745 |
| }, |
| { |
| "epoch": 2.71, |
| "learning_rate": 4.806687565308255e-06, |
| "loss": 0.1766, |
| "step": 1746 |
| }, |
| { |
| "epoch": 2.72, |
| "learning_rate": 4.780564263322884e-06, |
| "loss": 0.1717, |
| "step": 1747 |
| }, |
| { |
| "epoch": 2.72, |
| "learning_rate": 4.754440961337513e-06, |
| "loss": 0.1707, |
| "step": 1748 |
| }, |
| { |
| "epoch": 2.72, |
| "learning_rate": 4.728317659352143e-06, |
| "loss": 0.1764, |
| "step": 1749 |
| }, |
| { |
| "epoch": 2.72, |
| "learning_rate": 4.7021943573667714e-06, |
| "loss": 0.1584, |
| "step": 1750 |
| }, |
| { |
| "epoch": 2.72, |
| "learning_rate": 4.676071055381401e-06, |
| "loss": 0.1698, |
| "step": 1751 |
| }, |
| { |
| "epoch": 2.72, |
| "learning_rate": 4.6499477533960295e-06, |
| "loss": 0.1613, |
| "step": 1752 |
| }, |
| { |
| "epoch": 2.72, |
| "learning_rate": 4.623824451410658e-06, |
| "loss": 0.1946, |
| "step": 1753 |
| }, |
| { |
| "epoch": 2.73, |
| "learning_rate": 4.5977011494252875e-06, |
| "loss": 0.1677, |
| "step": 1754 |
| }, |
| { |
| "epoch": 2.73, |
| "learning_rate": 4.571577847439917e-06, |
| "loss": 0.1914, |
| "step": 1755 |
| }, |
| { |
| "epoch": 2.73, |
| "learning_rate": 4.5454545454545455e-06, |
| "loss": 0.1781, |
| "step": 1756 |
| }, |
| { |
| "epoch": 2.73, |
| "learning_rate": 4.519331243469175e-06, |
| "loss": 0.1686, |
| "step": 1757 |
| }, |
| { |
| "epoch": 2.73, |
| "learning_rate": 4.493207941483804e-06, |
| "loss": 0.1807, |
| "step": 1758 |
| }, |
| { |
| "epoch": 2.73, |
| "learning_rate": 4.467084639498432e-06, |
| "loss": 0.162, |
| "step": 1759 |
| }, |
| { |
| "epoch": 2.74, |
| "learning_rate": 4.440961337513062e-06, |
| "loss": 0.1701, |
| "step": 1760 |
| }, |
| { |
| "epoch": 2.74, |
| "learning_rate": 4.414838035527691e-06, |
| "loss": 0.1658, |
| "step": 1761 |
| }, |
| { |
| "epoch": 2.74, |
| "learning_rate": 4.38871473354232e-06, |
| "loss": 0.1745, |
| "step": 1762 |
| }, |
| { |
| "epoch": 2.74, |
| "learning_rate": 4.362591431556949e-06, |
| "loss": 0.1854, |
| "step": 1763 |
| }, |
| { |
| "epoch": 2.74, |
| "learning_rate": 4.3364681295715785e-06, |
| "loss": 0.1731, |
| "step": 1764 |
| }, |
| { |
| "epoch": 2.74, |
| "learning_rate": 4.310344827586207e-06, |
| "loss": 0.1811, |
| "step": 1765 |
| }, |
| { |
| "epoch": 2.74, |
| "learning_rate": 4.284221525600836e-06, |
| "loss": 0.1678, |
| "step": 1766 |
| }, |
| { |
| "epoch": 2.75, |
| "learning_rate": 4.258098223615465e-06, |
| "loss": 0.1732, |
| "step": 1767 |
| }, |
| { |
| "epoch": 2.75, |
| "learning_rate": 4.2319749216300945e-06, |
| "loss": 0.1818, |
| "step": 1768 |
| }, |
| { |
| "epoch": 2.75, |
| "learning_rate": 4.205851619644723e-06, |
| "loss": 0.17, |
| "step": 1769 |
| }, |
| { |
| "epoch": 2.75, |
| "learning_rate": 4.1797283176593526e-06, |
| "loss": 0.1728, |
| "step": 1770 |
| }, |
| { |
| "epoch": 2.75, |
| "learning_rate": 4.153605015673981e-06, |
| "loss": 0.1698, |
| "step": 1771 |
| }, |
| { |
| "epoch": 2.75, |
| "learning_rate": 4.127481713688611e-06, |
| "loss": 0.1752, |
| "step": 1772 |
| }, |
| { |
| "epoch": 2.76, |
| "learning_rate": 4.101358411703239e-06, |
| "loss": 0.1572, |
| "step": 1773 |
| }, |
| { |
| "epoch": 2.76, |
| "learning_rate": 4.075235109717869e-06, |
| "loss": 0.1682, |
| "step": 1774 |
| }, |
| { |
| "epoch": 2.76, |
| "learning_rate": 4.049111807732497e-06, |
| "loss": 0.1822, |
| "step": 1775 |
| }, |
| { |
| "epoch": 2.76, |
| "learning_rate": 4.022988505747127e-06, |
| "loss": 0.1763, |
| "step": 1776 |
| }, |
| { |
| "epoch": 2.76, |
| "learning_rate": 3.996865203761756e-06, |
| "loss": 0.1783, |
| "step": 1777 |
| }, |
| { |
| "epoch": 2.76, |
| "learning_rate": 3.970741901776385e-06, |
| "loss": 0.1701, |
| "step": 1778 |
| }, |
| { |
| "epoch": 2.76, |
| "learning_rate": 3.944618599791014e-06, |
| "loss": 0.1755, |
| "step": 1779 |
| }, |
| { |
| "epoch": 2.77, |
| "learning_rate": 3.9184952978056436e-06, |
| "loss": 0.1683, |
| "step": 1780 |
| }, |
| { |
| "epoch": 2.77, |
| "learning_rate": 3.892371995820271e-06, |
| "loss": 0.1759, |
| "step": 1781 |
| }, |
| { |
| "epoch": 2.77, |
| "learning_rate": 3.866248693834901e-06, |
| "loss": 0.1741, |
| "step": 1782 |
| }, |
| { |
| "epoch": 2.77, |
| "learning_rate": 3.84012539184953e-06, |
| "loss": 0.169, |
| "step": 1783 |
| }, |
| { |
| "epoch": 2.77, |
| "learning_rate": 3.814002089864159e-06, |
| "loss": 0.1715, |
| "step": 1784 |
| }, |
| { |
| "epoch": 2.77, |
| "learning_rate": 3.7878787878787882e-06, |
| "loss": 0.1858, |
| "step": 1785 |
| }, |
| { |
| "epoch": 2.78, |
| "learning_rate": 3.7617554858934172e-06, |
| "loss": 0.1738, |
| "step": 1786 |
| }, |
| { |
| "epoch": 2.78, |
| "learning_rate": 3.7356321839080462e-06, |
| "loss": 0.175, |
| "step": 1787 |
| }, |
| { |
| "epoch": 2.78, |
| "learning_rate": 3.709508881922675e-06, |
| "loss": 0.1721, |
| "step": 1788 |
| }, |
| { |
| "epoch": 2.78, |
| "learning_rate": 3.683385579937304e-06, |
| "loss": 0.1834, |
| "step": 1789 |
| }, |
| { |
| "epoch": 2.78, |
| "learning_rate": 3.6572622779519333e-06, |
| "loss": 0.1684, |
| "step": 1790 |
| }, |
| { |
| "epoch": 2.78, |
| "learning_rate": 3.6311389759665623e-06, |
| "loss": 0.1839, |
| "step": 1791 |
| }, |
| { |
| "epoch": 2.78, |
| "learning_rate": 3.6050156739811913e-06, |
| "loss": 0.182, |
| "step": 1792 |
| }, |
| { |
| "epoch": 2.79, |
| "learning_rate": 3.5788923719958208e-06, |
| "loss": 0.1823, |
| "step": 1793 |
| }, |
| { |
| "epoch": 2.79, |
| "learning_rate": 3.5527690700104498e-06, |
| "loss": 0.1751, |
| "step": 1794 |
| }, |
| { |
| "epoch": 2.79, |
| "learning_rate": 3.5266457680250784e-06, |
| "loss": 0.154, |
| "step": 1795 |
| }, |
| { |
| "epoch": 2.79, |
| "learning_rate": 3.5005224660397074e-06, |
| "loss": 0.1855, |
| "step": 1796 |
| }, |
| { |
| "epoch": 2.79, |
| "learning_rate": 3.4743991640543364e-06, |
| "loss": 0.1733, |
| "step": 1797 |
| }, |
| { |
| "epoch": 2.79, |
| "learning_rate": 3.448275862068966e-06, |
| "loss": 0.1761, |
| "step": 1798 |
| }, |
| { |
| "epoch": 2.8, |
| "learning_rate": 3.422152560083595e-06, |
| "loss": 0.1757, |
| "step": 1799 |
| }, |
| { |
| "epoch": 2.8, |
| "learning_rate": 3.396029258098224e-06, |
| "loss": 0.1717, |
| "step": 1800 |
| }, |
| { |
| "epoch": 2.8, |
| "learning_rate": 3.369905956112853e-06, |
| "loss": 0.1759, |
| "step": 1801 |
| }, |
| { |
| "epoch": 2.8, |
| "learning_rate": 3.3437826541274823e-06, |
| "loss": 0.1716, |
| "step": 1802 |
| }, |
| { |
| "epoch": 2.8, |
| "learning_rate": 3.3176593521421105e-06, |
| "loss": 0.1792, |
| "step": 1803 |
| }, |
| { |
| "epoch": 2.8, |
| "learning_rate": 3.29153605015674e-06, |
| "loss": 0.1693, |
| "step": 1804 |
| }, |
| { |
| "epoch": 2.81, |
| "learning_rate": 3.265412748171369e-06, |
| "loss": 0.1673, |
| "step": 1805 |
| }, |
| { |
| "epoch": 2.81, |
| "learning_rate": 3.239289446185998e-06, |
| "loss": 0.1738, |
| "step": 1806 |
| }, |
| { |
| "epoch": 2.81, |
| "learning_rate": 3.2131661442006274e-06, |
| "loss": 0.1581, |
| "step": 1807 |
| }, |
| { |
| "epoch": 2.81, |
| "learning_rate": 3.1870428422152564e-06, |
| "loss": 0.1701, |
| "step": 1808 |
| }, |
| { |
| "epoch": 2.81, |
| "learning_rate": 3.1609195402298854e-06, |
| "loss": 0.1764, |
| "step": 1809 |
| }, |
| { |
| "epoch": 2.81, |
| "learning_rate": 3.134796238244514e-06, |
| "loss": 0.1609, |
| "step": 1810 |
| }, |
| { |
| "epoch": 2.81, |
| "learning_rate": 3.1086729362591434e-06, |
| "loss": 0.1723, |
| "step": 1811 |
| }, |
| { |
| "epoch": 2.82, |
| "learning_rate": 3.0825496342737725e-06, |
| "loss": 0.1729, |
| "step": 1812 |
| }, |
| { |
| "epoch": 2.82, |
| "learning_rate": 3.0564263322884015e-06, |
| "loss": 0.1655, |
| "step": 1813 |
| }, |
| { |
| "epoch": 2.82, |
| "learning_rate": 3.0303030303030305e-06, |
| "loss": 0.1641, |
| "step": 1814 |
| }, |
| { |
| "epoch": 2.82, |
| "learning_rate": 3.0041797283176595e-06, |
| "loss": 0.1687, |
| "step": 1815 |
| }, |
| { |
| "epoch": 2.82, |
| "learning_rate": 2.9780564263322885e-06, |
| "loss": 0.1635, |
| "step": 1816 |
| }, |
| { |
| "epoch": 2.82, |
| "learning_rate": 2.9519331243469175e-06, |
| "loss": 0.1773, |
| "step": 1817 |
| }, |
| { |
| "epoch": 2.83, |
| "learning_rate": 2.9258098223615465e-06, |
| "loss": 0.1861, |
| "step": 1818 |
| }, |
| { |
| "epoch": 2.83, |
| "learning_rate": 2.8996865203761756e-06, |
| "loss": 0.17, |
| "step": 1819 |
| }, |
| { |
| "epoch": 2.83, |
| "learning_rate": 2.8735632183908046e-06, |
| "loss": 0.1784, |
| "step": 1820 |
| }, |
| { |
| "epoch": 2.83, |
| "learning_rate": 2.847439916405434e-06, |
| "loss": 0.1679, |
| "step": 1821 |
| }, |
| { |
| "epoch": 2.83, |
| "learning_rate": 2.821316614420063e-06, |
| "loss": 0.165, |
| "step": 1822 |
| }, |
| { |
| "epoch": 2.83, |
| "learning_rate": 2.7951933124346916e-06, |
| "loss": 0.1674, |
| "step": 1823 |
| }, |
| { |
| "epoch": 2.83, |
| "learning_rate": 2.769070010449321e-06, |
| "loss": 0.1695, |
| "step": 1824 |
| }, |
| { |
| "epoch": 2.84, |
| "learning_rate": 2.74294670846395e-06, |
| "loss": 0.1753, |
| "step": 1825 |
| }, |
| { |
| "epoch": 2.84, |
| "learning_rate": 2.7168234064785787e-06, |
| "loss": 0.1663, |
| "step": 1826 |
| }, |
| { |
| "epoch": 2.84, |
| "learning_rate": 2.690700104493208e-06, |
| "loss": 0.182, |
| "step": 1827 |
| }, |
| { |
| "epoch": 2.84, |
| "learning_rate": 2.664576802507837e-06, |
| "loss": 0.1777, |
| "step": 1828 |
| }, |
| { |
| "epoch": 2.84, |
| "learning_rate": 2.638453500522466e-06, |
| "loss": 0.1693, |
| "step": 1829 |
| }, |
| { |
| "epoch": 2.84, |
| "learning_rate": 2.612330198537095e-06, |
| "loss": 0.1817, |
| "step": 1830 |
| }, |
| { |
| "epoch": 2.85, |
| "learning_rate": 2.586206896551724e-06, |
| "loss": 0.1703, |
| "step": 1831 |
| }, |
| { |
| "epoch": 2.85, |
| "learning_rate": 2.560083594566353e-06, |
| "loss": 0.1765, |
| "step": 1832 |
| }, |
| { |
| "epoch": 2.85, |
| "learning_rate": 2.5339602925809826e-06, |
| "loss": 0.161, |
| "step": 1833 |
| }, |
| { |
| "epoch": 2.85, |
| "learning_rate": 2.507836990595611e-06, |
| "loss": 0.1705, |
| "step": 1834 |
| }, |
| { |
| "epoch": 2.85, |
| "learning_rate": 2.4817136886102406e-06, |
| "loss": 0.1721, |
| "step": 1835 |
| }, |
| { |
| "epoch": 2.85, |
| "learning_rate": 2.4555903866248697e-06, |
| "loss": 0.1838, |
| "step": 1836 |
| }, |
| { |
| "epoch": 2.85, |
| "learning_rate": 2.4294670846394982e-06, |
| "loss": 0.1582, |
| "step": 1837 |
| }, |
| { |
| "epoch": 2.86, |
| "learning_rate": 2.4033437826541277e-06, |
| "loss": 0.1574, |
| "step": 1838 |
| }, |
| { |
| "epoch": 2.86, |
| "learning_rate": 2.3772204806687567e-06, |
| "loss": 0.1682, |
| "step": 1839 |
| }, |
| { |
| "epoch": 2.86, |
| "learning_rate": 2.3510971786833857e-06, |
| "loss": 0.1658, |
| "step": 1840 |
| }, |
| { |
| "epoch": 2.86, |
| "learning_rate": 2.3249738766980147e-06, |
| "loss": 0.1721, |
| "step": 1841 |
| }, |
| { |
| "epoch": 2.86, |
| "learning_rate": 2.2988505747126437e-06, |
| "loss": 0.1722, |
| "step": 1842 |
| }, |
| { |
| "epoch": 2.86, |
| "learning_rate": 2.2727272727272728e-06, |
| "loss": 0.1913, |
| "step": 1843 |
| }, |
| { |
| "epoch": 2.87, |
| "learning_rate": 2.246603970741902e-06, |
| "loss": 0.1805, |
| "step": 1844 |
| }, |
| { |
| "epoch": 2.87, |
| "learning_rate": 2.220480668756531e-06, |
| "loss": 0.1807, |
| "step": 1845 |
| }, |
| { |
| "epoch": 2.87, |
| "learning_rate": 2.19435736677116e-06, |
| "loss": 0.1742, |
| "step": 1846 |
| }, |
| { |
| "epoch": 2.87, |
| "learning_rate": 2.1682340647857892e-06, |
| "loss": 0.1804, |
| "step": 1847 |
| }, |
| { |
| "epoch": 2.87, |
| "learning_rate": 2.142110762800418e-06, |
| "loss": 0.1704, |
| "step": 1848 |
| }, |
| { |
| "epoch": 2.87, |
| "learning_rate": 2.1159874608150473e-06, |
| "loss": 0.1586, |
| "step": 1849 |
| }, |
| { |
| "epoch": 2.88, |
| "learning_rate": 2.0898641588296763e-06, |
| "loss": 0.1737, |
| "step": 1850 |
| }, |
| { |
| "epoch": 2.88, |
| "learning_rate": 2.0637408568443053e-06, |
| "loss": 0.1618, |
| "step": 1851 |
| }, |
| { |
| "epoch": 2.88, |
| "learning_rate": 2.0376175548589343e-06, |
| "loss": 0.1814, |
| "step": 1852 |
| }, |
| { |
| "epoch": 2.88, |
| "learning_rate": 2.0114942528735633e-06, |
| "loss": 0.1834, |
| "step": 1853 |
| }, |
| { |
| "epoch": 2.88, |
| "learning_rate": 1.9853709508881923e-06, |
| "loss": 0.1736, |
| "step": 1854 |
| }, |
| { |
| "epoch": 2.88, |
| "learning_rate": 1.9592476489028218e-06, |
| "loss": 0.1718, |
| "step": 1855 |
| }, |
| { |
| "epoch": 2.88, |
| "learning_rate": 1.9331243469174504e-06, |
| "loss": 0.1824, |
| "step": 1856 |
| }, |
| { |
| "epoch": 2.89, |
| "learning_rate": 1.9070010449320796e-06, |
| "loss": 0.1682, |
| "step": 1857 |
| }, |
| { |
| "epoch": 2.89, |
| "learning_rate": 1.8808777429467086e-06, |
| "loss": 0.1792, |
| "step": 1858 |
| }, |
| { |
| "epoch": 2.89, |
| "learning_rate": 1.8547544409613374e-06, |
| "loss": 0.1823, |
| "step": 1859 |
| }, |
| { |
| "epoch": 2.89, |
| "learning_rate": 1.8286311389759666e-06, |
| "loss": 0.1738, |
| "step": 1860 |
| }, |
| { |
| "epoch": 2.89, |
| "learning_rate": 1.8025078369905957e-06, |
| "loss": 0.1579, |
| "step": 1861 |
| }, |
| { |
| "epoch": 2.89, |
| "learning_rate": 1.7763845350052249e-06, |
| "loss": 0.1709, |
| "step": 1862 |
| }, |
| { |
| "epoch": 2.9, |
| "learning_rate": 1.7502612330198537e-06, |
| "loss": 0.178, |
| "step": 1863 |
| }, |
| { |
| "epoch": 2.9, |
| "learning_rate": 1.724137931034483e-06, |
| "loss": 0.1756, |
| "step": 1864 |
| }, |
| { |
| "epoch": 2.9, |
| "learning_rate": 1.698014629049112e-06, |
| "loss": 0.1657, |
| "step": 1865 |
| }, |
| { |
| "epoch": 2.9, |
| "learning_rate": 1.6718913270637412e-06, |
| "loss": 0.1662, |
| "step": 1866 |
| }, |
| { |
| "epoch": 2.9, |
| "learning_rate": 1.64576802507837e-06, |
| "loss": 0.163, |
| "step": 1867 |
| }, |
| { |
| "epoch": 2.9, |
| "learning_rate": 1.619644723092999e-06, |
| "loss": 0.1885, |
| "step": 1868 |
| }, |
| { |
| "epoch": 2.9, |
| "learning_rate": 1.5935214211076282e-06, |
| "loss": 0.1605, |
| "step": 1869 |
| }, |
| { |
| "epoch": 2.91, |
| "learning_rate": 1.567398119122257e-06, |
| "loss": 0.1837, |
| "step": 1870 |
| }, |
| { |
| "epoch": 2.91, |
| "learning_rate": 1.5412748171368862e-06, |
| "loss": 0.1748, |
| "step": 1871 |
| }, |
| { |
| "epoch": 2.91, |
| "learning_rate": 1.5151515151515152e-06, |
| "loss": 0.1754, |
| "step": 1872 |
| }, |
| { |
| "epoch": 2.91, |
| "learning_rate": 1.4890282131661443e-06, |
| "loss": 0.1788, |
| "step": 1873 |
| }, |
| { |
| "epoch": 2.91, |
| "learning_rate": 1.4629049111807733e-06, |
| "loss": 0.1818, |
| "step": 1874 |
| }, |
| { |
| "epoch": 2.91, |
| "learning_rate": 1.4367816091954023e-06, |
| "loss": 0.1595, |
| "step": 1875 |
| }, |
| { |
| "epoch": 2.92, |
| "learning_rate": 1.4106583072100315e-06, |
| "loss": 0.1694, |
| "step": 1876 |
| }, |
| { |
| "epoch": 2.92, |
| "learning_rate": 1.3845350052246605e-06, |
| "loss": 0.1774, |
| "step": 1877 |
| }, |
| { |
| "epoch": 2.92, |
| "learning_rate": 1.3584117032392893e-06, |
| "loss": 0.1699, |
| "step": 1878 |
| }, |
| { |
| "epoch": 2.92, |
| "learning_rate": 1.3322884012539186e-06, |
| "loss": 0.1835, |
| "step": 1879 |
| }, |
| { |
| "epoch": 2.92, |
| "learning_rate": 1.3061650992685476e-06, |
| "loss": 0.1689, |
| "step": 1880 |
| }, |
| { |
| "epoch": 2.92, |
| "learning_rate": 1.2800417972831766e-06, |
| "loss": 0.1765, |
| "step": 1881 |
| }, |
| { |
| "epoch": 2.92, |
| "learning_rate": 1.2539184952978056e-06, |
| "loss": 0.1813, |
| "step": 1882 |
| }, |
| { |
| "epoch": 2.93, |
| "learning_rate": 1.2277951933124348e-06, |
| "loss": 0.1575, |
| "step": 1883 |
| }, |
| { |
| "epoch": 2.93, |
| "learning_rate": 1.2016718913270638e-06, |
| "loss": 0.1775, |
| "step": 1884 |
| }, |
| { |
| "epoch": 2.93, |
| "learning_rate": 1.1755485893416929e-06, |
| "loss": 0.1774, |
| "step": 1885 |
| }, |
| { |
| "epoch": 2.93, |
| "learning_rate": 1.1494252873563219e-06, |
| "loss": 0.1581, |
| "step": 1886 |
| }, |
| { |
| "epoch": 2.93, |
| "learning_rate": 1.123301985370951e-06, |
| "loss": 0.1616, |
| "step": 1887 |
| }, |
| { |
| "epoch": 2.93, |
| "learning_rate": 1.09717868338558e-06, |
| "loss": 0.1751, |
| "step": 1888 |
| }, |
| { |
| "epoch": 2.94, |
| "learning_rate": 1.071055381400209e-06, |
| "loss": 0.1709, |
| "step": 1889 |
| }, |
| { |
| "epoch": 2.94, |
| "learning_rate": 1.0449320794148381e-06, |
| "loss": 0.1793, |
| "step": 1890 |
| }, |
| { |
| "epoch": 2.94, |
| "learning_rate": 1.0188087774294672e-06, |
| "loss": 0.1658, |
| "step": 1891 |
| }, |
| { |
| "epoch": 2.94, |
| "learning_rate": 9.926854754440962e-07, |
| "loss": 0.1766, |
| "step": 1892 |
| }, |
| { |
| "epoch": 2.94, |
| "learning_rate": 9.665621734587252e-07, |
| "loss": 0.1937, |
| "step": 1893 |
| }, |
| { |
| "epoch": 2.94, |
| "learning_rate": 9.404388714733543e-07, |
| "loss": 0.1786, |
| "step": 1894 |
| }, |
| { |
| "epoch": 2.94, |
| "learning_rate": 9.143155694879833e-07, |
| "loss": 0.17, |
| "step": 1895 |
| }, |
| { |
| "epoch": 2.95, |
| "learning_rate": 8.881922675026124e-07, |
| "loss": 0.1781, |
| "step": 1896 |
| }, |
| { |
| "epoch": 2.95, |
| "learning_rate": 8.620689655172415e-07, |
| "loss": 0.1704, |
| "step": 1897 |
| }, |
| { |
| "epoch": 2.95, |
| "learning_rate": 8.359456635318706e-07, |
| "loss": 0.1748, |
| "step": 1898 |
| }, |
| { |
| "epoch": 2.95, |
| "learning_rate": 8.098223615464995e-07, |
| "loss": 0.1845, |
| "step": 1899 |
| }, |
| { |
| "epoch": 2.95, |
| "learning_rate": 7.836990595611285e-07, |
| "loss": 0.1792, |
| "step": 1900 |
| }, |
| { |
| "epoch": 2.95, |
| "learning_rate": 7.575757575757576e-07, |
| "loss": 0.1801, |
| "step": 1901 |
| }, |
| { |
| "epoch": 2.96, |
| "learning_rate": 7.314524555903866e-07, |
| "loss": 0.1624, |
| "step": 1902 |
| }, |
| { |
| "epoch": 2.96, |
| "learning_rate": 7.053291536050158e-07, |
| "loss": 0.1701, |
| "step": 1903 |
| }, |
| { |
| "epoch": 2.96, |
| "learning_rate": 6.792058516196447e-07, |
| "loss": 0.1685, |
| "step": 1904 |
| }, |
| { |
| "epoch": 2.96, |
| "learning_rate": 6.530825496342738e-07, |
| "loss": 0.1705, |
| "step": 1905 |
| }, |
| { |
| "epoch": 2.96, |
| "learning_rate": 6.269592476489028e-07, |
| "loss": 0.1851, |
| "step": 1906 |
| }, |
| { |
| "epoch": 2.96, |
| "learning_rate": 6.008359456635319e-07, |
| "loss": 0.1811, |
| "step": 1907 |
| }, |
| { |
| "epoch": 2.97, |
| "learning_rate": 5.747126436781609e-07, |
| "loss": 0.1754, |
| "step": 1908 |
| }, |
| { |
| "epoch": 2.97, |
| "learning_rate": 5.4858934169279e-07, |
| "loss": 0.1706, |
| "step": 1909 |
| }, |
| { |
| "epoch": 2.97, |
| "learning_rate": 5.224660397074191e-07, |
| "loss": 0.1756, |
| "step": 1910 |
| }, |
| { |
| "epoch": 2.97, |
| "learning_rate": 4.963427377220481e-07, |
| "loss": 0.1732, |
| "step": 1911 |
| }, |
| { |
| "epoch": 2.97, |
| "learning_rate": 4.7021943573667715e-07, |
| "loss": 0.155, |
| "step": 1912 |
| }, |
| { |
| "epoch": 2.97, |
| "learning_rate": 4.440961337513062e-07, |
| "loss": 0.1703, |
| "step": 1913 |
| }, |
| { |
| "epoch": 2.97, |
| "learning_rate": 4.179728317659353e-07, |
| "loss": 0.1668, |
| "step": 1914 |
| }, |
| { |
| "epoch": 2.98, |
| "learning_rate": 3.9184952978056425e-07, |
| "loss": 0.1869, |
| "step": 1915 |
| }, |
| { |
| "epoch": 2.98, |
| "learning_rate": 3.657262277951933e-07, |
| "loss": 0.1693, |
| "step": 1916 |
| }, |
| { |
| "epoch": 2.98, |
| "learning_rate": 3.3960292580982233e-07, |
| "loss": 0.1721, |
| "step": 1917 |
| }, |
| { |
| "epoch": 2.98, |
| "learning_rate": 3.134796238244514e-07, |
| "loss": 0.1728, |
| "step": 1918 |
| }, |
| { |
| "epoch": 2.98, |
| "learning_rate": 2.8735632183908047e-07, |
| "loss": 0.1676, |
| "step": 1919 |
| }, |
| { |
| "epoch": 2.98, |
| "learning_rate": 2.6123301985370954e-07, |
| "loss": 0.172, |
| "step": 1920 |
| }, |
| { |
| "epoch": 2.99, |
| "learning_rate": 2.3510971786833858e-07, |
| "loss": 0.1614, |
| "step": 1921 |
| }, |
| { |
| "epoch": 2.99, |
| "learning_rate": 2.0898641588296764e-07, |
| "loss": 0.1867, |
| "step": 1922 |
| }, |
| { |
| "epoch": 2.99, |
| "learning_rate": 1.8286311389759666e-07, |
| "loss": 0.1615, |
| "step": 1923 |
| }, |
| { |
| "epoch": 2.99, |
| "learning_rate": 1.567398119122257e-07, |
| "loss": 0.1675, |
| "step": 1924 |
| }, |
| { |
| "epoch": 2.99, |
| "learning_rate": 1.3061650992685477e-07, |
| "loss": 0.1795, |
| "step": 1925 |
| }, |
| { |
| "epoch": 2.99, |
| "learning_rate": 1.0449320794148382e-07, |
| "loss": 0.1749, |
| "step": 1926 |
| }, |
| { |
| "epoch": 2.99, |
| "learning_rate": 7.836990595611285e-08, |
| "loss": 0.1772, |
| "step": 1927 |
| }, |
| { |
| "epoch": 3.0, |
| "learning_rate": 5.224660397074191e-08, |
| "loss": 0.1775, |
| "step": 1928 |
| }, |
| { |
| "epoch": 3.0, |
| "learning_rate": 2.6123301985370956e-08, |
| "loss": 0.1611, |
| "step": 1929 |
| }, |
| { |
| "epoch": 3.0, |
| "step": 1929, |
| "total_flos": 481400140136448.0, |
| "train_loss": 0.07281383225465451, |
| "train_runtime": 49663.0159, |
| "train_samples_per_second": 9.954, |
| "train_steps_per_second": 0.039 |
| } |
| ], |
| "logging_steps": 1.0, |
| "max_steps": 1929, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 300, |
| "total_flos": 481400140136448.0, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|