| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "global_step": 367, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1e-05, | |
| "loss": 3.2173, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2e-05, | |
| "loss": 3.2031, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.0000000000000004e-05, | |
| "loss": 3.1896, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4e-05, | |
| "loss": 3.5399, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.999925099660787e-05, | |
| "loss": 3.249, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.999700404253208e-05, | |
| "loss": 4.346, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9993259306070256e-05, | |
| "loss": 4.1291, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.998801706770442e-05, | |
| "loss": 3.7064, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.9981277720080015e-05, | |
| "loss": 3.5643, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.9973041767976466e-05, | |
| "loss": 3.5031, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.996330982826937e-05, | |
| "loss": 3.4195, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.995208262988431e-05, | |
| "loss": 3.3598, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.9939361013742275e-05, | |
| "loss": 3.3177, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.99251459326966e-05, | |
| "loss": 3.2443, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.9909438451461695e-05, | |
| "loss": 3.2215, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.989223974653323e-05, | |
| "loss": 3.149, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.9873551106100035e-05, | |
| "loss": 3.1167, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.985337392994763e-05, | |
| "loss": 3.0538, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.983170972935333e-05, | |
| "loss": 3.029, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.9808560126973126e-05, | |
| "loss": 3.0055, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.9783926856720085e-05, | |
| "loss": 2.9598, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.975781176363451e-05, | |
| "loss": 2.9572, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.973021680374571e-05, | |
| "loss": 2.9004, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.9701144043925576e-05, | |
| "loss": 2.8984, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.9670595661733654e-05, | |
| "loss": 2.874, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.963857394525413e-05, | |
| "loss": 2.8689, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.960508129292446e-05, | |
| "loss": 2.8885, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.9570120213355636e-05, | |
| "loss": 2.8331, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.953369332514438e-05, | |
| "loss": 2.8093, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.949580335667699e-05, | |
| "loss": 2.7919, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.945645314592495e-05, | |
| "loss": 2.7876, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 3.9415645640232386e-05, | |
| "loss": 2.791, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 3.937338389609533e-05, | |
| "loss": 2.7751, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 3.932967107893274e-05, | |
| "loss": 2.7673, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.928451046284946e-05, | |
| "loss": 2.74, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.923790543039095e-05, | |
| "loss": 2.7354, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.9189859472289956e-05, | |
| "loss": 2.699, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.9140376187205025e-05, | |
| "loss": 2.7079, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.9089459281451e-05, | |
| "loss": 2.6673, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.903711256872139e-05, | |
| "loss": 2.6733, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.898333996980275e-05, | |
| "loss": 2.659, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 3.8928145512280973e-05, | |
| "loss": 2.6657, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.8871533330239646e-05, | |
| "loss": 2.6348, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.8813507663950404e-05, | |
| "loss": 2.6497, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.8754072859555346e-05, | |
| "loss": 2.5951, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 3.869323336874146e-05, | |
| "loss": 2.6479, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 3.8630993748407274e-05, | |
| "loss": 2.6018, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 3.856735866032145e-05, | |
| "loss": 2.5817, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 3.8502332870773675e-05, | |
| "loss": 2.5903, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 3.843592125021764e-05, | |
| "loss": 2.5757, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 3.8368128772906254e-05, | |
| "loss": 2.5705, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 3.829896051651907e-05, | |
| "loss": 2.5914, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 3.822842166178194e-05, | |
| "loss": 2.5694, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 3.815651749207902e-05, | |
| "loss": 2.5826, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 3.8083253393057006e-05, | |
| "loss": 2.5435, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 3.8008634852221777e-05, | |
| "loss": 2.5286, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 3.793266745852735e-05, | |
| "loss": 2.5152, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 3.785535690195728e-05, | |
| "loss": 2.4879, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 3.7776708973098476e-05, | |
| "loss": 2.5058, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 3.769672956270749e-05, | |
| "loss": 2.5437, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 3.761542466126929e-05, | |
| "loss": 2.4668, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 3.753280035854857e-05, | |
| "loss": 2.501, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 3.7448862843133644e-05, | |
| "loss": 2.4697, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 3.736361840197288e-05, | |
| "loss": 2.4651, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 3.727707341990383e-05, | |
| "loss": 2.4394, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 3.718923437917503e-05, | |
| "loss": 2.4239, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 3.7100107858960404e-05, | |
| "loss": 2.4541, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 3.7009700534866557e-05, | |
| "loss": 2.445, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 3.691801917843273e-05, | |
| "loss": 2.422, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 3.6825070656623626e-05, | |
| "loss": 2.4274, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 3.6730861931315054e-05, | |
| "loss": 2.4062, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.663540005877249e-05, | |
| "loss": 2.4, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.653869218912258e-05, | |
| "loss": 2.4229, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.6440745565817556e-05, | |
| "loss": 2.3909, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.6341567525092727e-05, | |
| "loss": 2.3607, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.6241165495417006e-05, | |
| "loss": 2.399, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.613954699693645e-05, | |
| "loss": 2.3902, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.603671964091107e-05, | |
| "loss": 2.3692, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.593269112914472e-05, | |
| "loss": 2.3436, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.582746925340822e-05, | |
| "loss": 2.3629, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.5721061894855756e-05, | |
| "loss": 2.3287, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.561347702343456e-05, | |
| "loss": 2.3376, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.5504722697288025e-05, | |
| "loss": 2.3418, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.539480706215204e-05, | |
| "loss": 2.3182, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.5283738350744986e-05, | |
| "loss": 2.3287, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.517152488215101e-05, | |
| "loss": 2.3222, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.505817506119698e-05, | |
| "loss": 2.3393, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.494369737782293e-05, | |
| "loss": 2.322, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.4828100406446184e-05, | |
| "loss": 2.2907, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.47113928053191e-05, | |
| "loss": 2.3132, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.45935833158806e-05, | |
| "loss": 2.2972, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.44746807621014e-05, | |
| "loss": 2.2643, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.4354694049823124e-05, | |
| "loss": 2.2527, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.4233632166091205e-05, | |
| "loss": 2.2746, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.4111504178481813e-05, | |
| "loss": 2.2479, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.3988319234422636e-05, | |
| "loss": 2.2954, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.3864086560507785e-05, | |
| "loss": 2.2455, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.373881546180666e-05, | |
| "loss": 2.2442, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.361251532116707e-05, | |
| "loss": 2.2035, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.3485195598512365e-05, | |
| "loss": 2.2234, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.3356865830132976e-05, | |
| "loss": 2.2226, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.322753562797209e-05, | |
| "loss": 2.2167, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.309721467890571e-05, | |
| "loss": 2.2481, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.296591274401712e-05, | |
| "loss": 2.1965, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.28336396578658e-05, | |
| "loss": 2.1935, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.270040532775077e-05, | |
| "loss": 2.196, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.256621973296854e-05, | |
| "loss": 2.1733, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.243109292406568e-05, | |
| "loss": 2.1735, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.229503502208602e-05, | |
| "loss": 2.1915, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.215805621781256e-05, | |
| "loss": 2.1795, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.202016677100422e-05, | |
| "loss": 2.2021, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.188137700962733e-05, | |
| "loss": 2.201, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.174169732908209e-05, | |
| "loss": 2.1857, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.1601138191423966e-05, | |
| "loss": 2.1818, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.145971012458005e-05, | |
| "loss": 2.1438, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.13174237215605e-05, | |
| "loss": 2.1359, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.11742896396652e-05, | |
| "loss": 2.1668, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.103031859968542e-05, | |
| "loss": 2.1769, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.0885521385100885e-05, | |
| "loss": 2.1445, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.0739908841272095e-05, | |
| "loss": 2.1193, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.059349187462798e-05, | |
| "loss": 2.1285, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.044628145184899e-05, | |
| "loss": 2.1411, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 3.0298288599045747e-05, | |
| "loss": 2.1321, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 3.0149524400933114e-05, | |
| "loss": 2.1008, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 3.0000000000000004e-05, | |
| "loss": 2.1437, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.9849726595674756e-05, | |
| "loss": 2.1224, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.9698715443486338e-05, | |
| "loss": 2.083, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.9546977854221266e-05, | |
| "loss": 2.1156, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.9394525193076454e-05, | |
| "loss": 2.1127, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.9241368878807925e-05, | |
| "loss": 2.0949, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 2.908752038287558e-05, | |
| "loss": 2.0821, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 2.8932991228583954e-05, | |
| "loss": 2.0735, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 2.877779299021912e-05, | |
| "loss": 2.0996, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 2.8621937292181768e-05, | |
| "loss": 2.0967, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 2.846543580811656e-05, | |
| "loss": 2.0313, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 2.8308300260037734e-05, | |
| "loss": 2.0891, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 2.8150542417451144e-05, | |
| "loss": 2.0816, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 2.7992174096472714e-05, | |
| "loss": 2.0765, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 2.783320715894341e-05, | |
| "loss": 2.026, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 2.767365351154077e-05, | |
| "loss": 2.0424, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 2.751352510488711e-05, | |
| "loss": 2.074, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 2.7352833932654402e-05, | |
| "loss": 2.0189, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 2.719159203066597e-05, | |
| "loss": 2.0283, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 2.702981147599495e-05, | |
| "loss": 2.0373, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.6867504386059776e-05, | |
| "loss": 2.0141, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.6704682917716528e-05, | |
| "loss": 2.0197, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.6541359266348437e-05, | |
| "loss": 2.0168, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.637754566495238e-05, | |
| "loss": 2.0032, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.6213254383222665e-05, | |
| "loss": 2.0038, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.6048497726632023e-05, | |
| "loss": 1.9901, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.588328803550993e-05, | |
| "loss": 1.9917, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.571763768411829e-05, | |
| "loss": 1.9718, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.555155907972461e-05, | |
| "loss": 2.0155, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.5385064661672692e-05, | |
| "loss": 2.0072, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.5218166900450937e-05, | |
| "loss": 1.9935, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.5050878296758255e-05, | |
| "loss": 1.9879, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.488321138056783e-05, | |
| "loss": 2.0028, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.471517871018855e-05, | |
| "loss": 1.9669, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.4546792871324424e-05, | |
| "loss": 1.9854, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.43780664761319e-05, | |
| "loss": 1.9613, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.4209012162275217e-05, | |
| "loss": 1.9474, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.4039642591979825e-05, | |
| "loss": 1.9424, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.3869970451083996e-05, | |
| "loss": 1.9656, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.370000844808863e-05, | |
| "loss": 1.9686, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.3529769313205423e-05, | |
| "loss": 1.9655, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.3359265797403297e-05, | |
| "loss": 1.97, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.318851067145345e-05, | |
| "loss": 1.9435, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.3017516724972716e-05, | |
| "loss": 1.972, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.2846296765465708e-05, | |
| "loss": 1.9789, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.267486361736546e-05, | |
| "loss": 1.9466, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.250323012107292e-05, | |
| "loss": 1.9231, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.2331409131995186e-05, | |
| "loss": 1.9672, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.2159413519582623e-05, | |
| "loss": 1.9265, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.1987256166364937e-05, | |
| "loss": 1.911, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.1814949966986288e-05, | |
| "loss": 1.9236, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.1642507827239455e-05, | |
| "loss": 1.9543, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.1469942663099208e-05, | |
| "loss": 1.9296, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.129726739975486e-05, | |
| "loss": 1.9292, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.112449497064223e-05, | |
| "loss": 1.9132, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.095163831647485e-05, | |
| "loss": 1.9087, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.0778710384274757e-05, | |
| "loss": 1.9137, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.0605724126402734e-05, | |
| "loss": 1.8797, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.0432692499588174e-05, | |
| "loss": 1.8937, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.025962846395862e-05, | |
| "loss": 1.9124, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.0086544982069046e-05, | |
| "loss": 1.8763, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.9913455017930964e-05, | |
| "loss": 1.8699, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.9740371536041388e-05, | |
| "loss": 1.8841, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.9567307500411833e-05, | |
| "loss": 1.9116, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.939427587359727e-05, | |
| "loss": 1.8997, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.922128961572525e-05, | |
| "loss": 1.8716, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.9048361683525155e-05, | |
| "loss": 1.8651, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.8875505029357775e-05, | |
| "loss": 1.8827, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.8702732600245138e-05, | |
| "loss": 1.871, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.8530057336900805e-05, | |
| "loss": 1.8788, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.835749217276055e-05, | |
| "loss": 1.8641, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.8185050033013715e-05, | |
| "loss": 1.8856, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.8012743833635067e-05, | |
| "loss": 1.8524, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.7840586480417387e-05, | |
| "loss": 1.8142, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.766859086800482e-05, | |
| "loss": 1.8361, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.7496769878927085e-05, | |
| "loss": 1.8197, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.7325136382634547e-05, | |
| "loss": 1.8723, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.7153703234534302e-05, | |
| "loss": 1.8088, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.6982483275027287e-05, | |
| "loss": 1.8264, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.6811489328546557e-05, | |
| "loss": 1.7955, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.6640734202596702e-05, | |
| "loss": 1.8337, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.647023068679459e-05, | |
| "loss": 1.8571, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.6299991551911373e-05, | |
| "loss": 1.8662, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.6130029548916007e-05, | |
| "loss": 1.8593, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.5960357408020178e-05, | |
| "loss": 1.8387, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.579098783772479e-05, | |
| "loss": 1.8395, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.5621933523868106e-05, | |
| "loss": 1.8271, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.5453207128675583e-05, | |
| "loss": 1.8738, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.5284821289811453e-05, | |
| "loss": 1.8598, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.5116788619432177e-05, | |
| "loss": 1.832, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.4949121703241747e-05, | |
| "loss": 1.8176, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.4781833099549072e-05, | |
| "loss": 1.8302, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.461493533832731e-05, | |
| "loss": 1.848, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.4448440920275402e-05, | |
| "loss": 1.7919, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.4282362315881719e-05, | |
| "loss": 1.8084, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.4116711964490076e-05, | |
| "loss": 1.7979, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.395150227336798e-05, | |
| "loss": 1.8167, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.3786745616777348e-05, | |
| "loss": 1.8015, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.3622454335047631e-05, | |
| "loss": 1.8359, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.345864073365157e-05, | |
| "loss": 1.8111, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.329531708228347e-05, | |
| "loss": 1.814, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.3132495613940237e-05, | |
| "loss": 1.7876, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.2970188524005058e-05, | |
| "loss": 1.7877, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.2808407969334037e-05, | |
| "loss": 1.7683, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.2647166067345598e-05, | |
| "loss": 1.8017, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.24864748951129e-05, | |
| "loss": 1.7989, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.2326346488459237e-05, | |
| "loss": 1.752, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.2166792841056596e-05, | |
| "loss": 1.7771, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.2007825903527287e-05, | |
| "loss": 1.7739, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.1849457582548864e-05, | |
| "loss": 1.7821, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.1691699739962275e-05, | |
| "loss": 1.7745, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.153456419188345e-05, | |
| "loss": 1.7759, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.137806270781824e-05, | |
| "loss": 1.795, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.1222207009780888e-05, | |
| "loss": 1.7348, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.1067008771416047e-05, | |
| "loss": 1.7385, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.091247961712442e-05, | |
| "loss": 1.798, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.0758631121192075e-05, | |
| "loss": 1.7615, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.0605474806923556e-05, | |
| "loss": 1.7768, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.0453022145778742e-05, | |
| "loss": 1.735, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.0301284556513669e-05, | |
| "loss": 1.7635, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.0150273404325244e-05, | |
| "loss": 1.7395, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.0000000000000006e-05, | |
| "loss": 1.7238, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 9.85047559906689e-06, | |
| "loss": 1.7633, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.70171140095426e-06, | |
| "loss": 1.7585, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.553718548151011e-06, | |
| "loss": 1.7379, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.406508125372034e-06, | |
| "loss": 1.7427, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.260091158727913e-06, | |
| "loss": 1.7205, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.114478614899123e-06, | |
| "loss": 1.7576, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 8.969681400314589e-06, | |
| "loss": 1.7559, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 8.825710360334812e-06, | |
| "loss": 1.7465, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 8.682576278439504e-06, | |
| "loss": 1.7571, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 8.540289875419962e-06, | |
| "loss": 1.7239, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 8.39886180857604e-06, | |
| "loss": 1.737, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 8.258302670917915e-06, | |
| "loss": 1.7449, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 8.118622990372676e-06, | |
| "loss": 1.7399, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.979833228995782e-06, | |
| "loss": 1.7411, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.841943782187435e-06, | |
| "loss": 1.7333, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.704964977913984e-06, | |
| "loss": 1.7641, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.568907075934322e-06, | |
| "loss": 1.7311, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.433780267031463e-06, | |
| "loss": 1.7022, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.299594672249231e-06, | |
| "loss": 1.7178, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.166360342134202e-06, | |
| "loss": 1.7497, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.034087255982882e-06, | |
| "loss": 1.7196, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 6.902785321094301e-06, | |
| "loss": 1.707, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 6.7724643720279156e-06, | |
| "loss": 1.7525, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 6.643134169867031e-06, | |
| "loss": 1.6995, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 6.514804401487642e-06, | |
| "loss": 1.7229, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 6.38748467883294e-06, | |
| "loss": 1.7021, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 6.261184538193341e-06, | |
| "loss": 1.6753, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 6.135913439492227e-06, | |
| "loss": 1.7111, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 6.01168076557737e-06, | |
| "loss": 1.7046, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 5.888495821518194e-06, | |
| "loss": 1.7362, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 5.7663678339087995e-06, | |
| "loss": 1.7811, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 5.6453059501768806e-06, | |
| "loss": 1.6819, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 5.5253192378985966e-06, | |
| "loss": 1.7154, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 5.4064166841194e-06, | |
| "loss": 1.7069, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 5.288607194680899e-06, | |
| "loss": 1.715, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 5.171899593553824e-06, | |
| "loss": 1.7173, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 5.056302622177074e-06, | |
| "loss": 1.6873, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.941824938803024e-06, | |
| "loss": 1.7291, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.828475117848992e-06, | |
| "loss": 1.6928, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.716261649255021e-06, | |
| "loss": 1.6815, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.605192937847962e-06, | |
| "loss": 1.6701, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.495277302711982e-06, | |
| "loss": 1.7119, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.386522976565439e-06, | |
| "loss": 1.6813, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.278938105144255e-06, | |
| "loss": 1.6945, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.172530746591783e-06, | |
| "loss": 1.7221, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.06730887085528e-06, | |
| "loss": 1.7254, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.963280359088933e-06, | |
| "loss": 1.6873, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.86045300306356e-06, | |
| "loss": 1.6773, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.7588345045830044e-06, | |
| "loss": 1.7121, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.658432474907274e-06, | |
| "loss": 1.6798, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.559254434182451e-06, | |
| "loss": 1.6762, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.461307810877428e-06, | |
| "loss": 1.6934, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.364599941227513e-06, | |
| "loss": 1.6862, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.2691380686849517e-06, | |
| "loss": 1.6708, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.174929343376374e-06, | |
| "loss": 1.6764, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.081980821567272e-06, | |
| "loss": 1.6886, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.990299465133446e-06, | |
| "loss": 1.6679, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.8998921410396e-06, | |
| "loss": 1.6854, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.8107656208249733e-06, | |
| "loss": 1.681, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.72292658009617e-06, | |
| "loss": 1.6987, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.6363815980271248e-06, | |
| "loss": 1.6574, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.551137156866357e-06, | |
| "loss": 1.7037, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.4671996414514276e-06, | |
| "loss": 1.702, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.384575338730717e-06, | |
| "loss": 1.6788, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.3032704372925176e-06, | |
| "loss": 1.6624, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.223291026901533e-06, | |
| "loss": 1.693, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.144643098042727e-06, | |
| "loss": 1.6687, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.0673325414726574e-06, | |
| "loss": 1.6914, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.991365147778228e-06, | |
| "loss": 1.6708, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.9167466069429964e-06, | |
| "loss": 1.6752, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.8434825079209884e-06, | |
| "loss": 1.6893, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.7715783382180672e-06, | |
| "loss": 1.6681, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.7010394834809373e-06, | |
| "loss": 1.6982, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.6318712270937442e-06, | |
| "loss": 1.6421, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.5640787497823585e-06, | |
| "loss": 1.7007, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.4976671292263257e-06, | |
| "loss": 1.6832, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.4326413396785488e-06, | |
| "loss": 1.6584, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.3690062515927239e-06, | |
| "loss": 1.7009, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.306766631258536e-06, | |
| "loss": 1.6755, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.245927140444665e-06, | |
| "loss": 1.686, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.1864923360496028e-06, | |
| "loss": 1.6916, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.128466669760362e-06, | |
| "loss": 1.6923, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.0718544877190306e-06, | |
| "loss": 1.7079, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.0166600301972517e-06, | |
| "loss": 1.6865, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 9.628874312786096e-07, | |
| "loss": 1.6624, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 9.105407185490067e-07, | |
| "loss": 1.6489, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.59623812794983e-07, | |
| "loss": 1.6923, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.101405277100549e-07, | |
| "loss": 1.678, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 7.620945696090532e-07, | |
| "loss": 1.6302, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 7.154895371505421e-07, | |
| "loss": 1.6555, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.703289210672603e-07, | |
| "loss": 1.6805, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.266161039046737e-07, | |
| "loss": 1.6744, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 5.843543597676138e-07, | |
| "loss": 1.6585, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.435468540750544e-07, | |
| "loss": 1.6951, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.041966433230094e-07, | |
| "loss": 1.6666, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.6630667485561885e-07, | |
| "loss": 1.6757, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.2987978664436936e-07, | |
| "loss": 1.6719, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.9491870707554445e-07, | |
| "loss": 1.6575, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.614260547458659e-07, | |
| "loss": 1.6398, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.2940433826635257e-07, | |
| "loss": 1.6523, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.9885595607443086e-07, | |
| "loss": 1.6521, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.697831962542874e-07, | |
| "loss": 1.6598, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.4218823636549703e-07, | |
| "loss": 1.6721, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.1607314327991791e-07, | |
| "loss": 1.6975, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.9143987302687738e-07, | |
| "loss": 1.6652, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.682902706466738e-07, | |
| "loss": 1.6554, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.4662607005237805e-07, | |
| "loss": 1.6603, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.264488938999664e-07, | |
| "loss": 1.6843, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.0776025346677321e-07, | |
| "loss": 1.69, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 9.056154853830823e-08, | |
| "loss": 1.6755, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 7.485406730340483e-08, | |
| "loss": 1.6577, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.06389862577328e-08, | |
| "loss": 1.7059, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.7917370115688756e-08, | |
| "loss": 1.7009, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.66901717306356e-08, | |
| "loss": 1.6863, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.6958232023539532e-08, | |
| "loss": 1.6557, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.8722279919987098e-08, | |
| "loss": 1.6717, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.1982932295582227e-08, | |
| "loss": 1.6401, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.7406939297520734e-09, | |
| "loss": 1.6846, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.995957467923916e-09, | |
| "loss": 1.6775, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 7.490033921331296e-10, | |
| "loss": 1.675, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0, | |
| "loss": 1.6419, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 367, | |
| "total_flos": 3.839475069608788e+18, | |
| "train_loss": 2.069092570598509, | |
| "train_runtime": 4018.2223, | |
| "train_samples_per_second": 174.946, | |
| "train_steps_per_second": 0.091 | |
| } | |
| ], | |
| "max_steps": 367, | |
| "num_train_epochs": 1, | |
| "total_flos": 3.839475069608788e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |