|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 0.9985710993316432, |
|
"global_step": 677, |
|
"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": 1.5272, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 0.0, |
|
"loss": 1.4949, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 0.0, |
|
"loss": 1.5184, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.0, |
|
"loss": 1.5134, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.0, |
|
"loss": 1.5608, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.0, |
|
"loss": 1.5162, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.0, |
|
"loss": 1.508, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.0, |
|
"loss": 1.5244, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.0, |
|
"loss": 1.5005, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 0.0, |
|
"loss": 1.5176, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 0.0, |
|
"loss": 1.5584, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 0.0, |
|
"loss": 1.5084, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 0.0, |
|
"loss": 1.5425, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 0.0, |
|
"loss": 1.4883, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 0.0, |
|
"loss": 1.4934, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 0.0, |
|
"loss": 1.4992, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 0.0, |
|
"loss": 1.5702, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 0.0, |
|
"loss": 1.5081, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 2e-05, |
|
"loss": 1.4992, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.9999892011980525e-05, |
|
"loss": 1.5015, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.9999568050254373e-05, |
|
"loss": 1.4444, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.999902812181835e-05, |
|
"loss": 1.3563, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.9998272238333606e-05, |
|
"loss": 1.2379, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.9997300416125426e-05, |
|
"loss": 1.1627, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.999611267618283e-05, |
|
"loss": 1.1329, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.9994709044158157e-05, |
|
"loss": 1.0667, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.99930895503665e-05, |
|
"loss": 1.0723, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.9991254229785043e-05, |
|
"loss": 1.0248, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.998920312205231e-05, |
|
"loss": 1.0191, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.9986936271467316e-05, |
|
"loss": 1.0021, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.99844537269886e-05, |
|
"loss": 0.9416, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.9981755542233175e-05, |
|
"loss": 0.9749, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.9978841775475368e-05, |
|
"loss": 0.938, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.997571248964556e-05, |
|
"loss": 0.9259, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.9972367752328824e-05, |
|
"loss": 0.8736, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.9968807635763472e-05, |
|
"loss": 0.852, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.9965032216839493e-05, |
|
"loss": 0.8858, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.996104157709689e-05, |
|
"loss": 0.864, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.9956835802723916e-05, |
|
"loss": 0.8736, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.9952414984555225e-05, |
|
"loss": 0.8471, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.994777921806989e-05, |
|
"loss": 0.8156, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.9942928603389366e-05, |
|
"loss": 0.8051, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.9937863245275303e-05, |
|
"loss": 0.7974, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.9932583253127302e-05, |
|
"loss": 0.8034, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.992708874098054e-05, |
|
"loss": 0.7963, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.9921379827503316e-05, |
|
"loss": 0.7613, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.991545663599448e-05, |
|
"loss": 0.7443, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.990931929438078e-05, |
|
"loss": 0.7833, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.990296793521408e-05, |
|
"loss": 0.7663, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.989640269566853e-05, |
|
"loss": 0.745, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.9889623717537564e-05, |
|
"loss": 0.7292, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.9882631147230874e-05, |
|
"loss": 0.7518, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.987542513577122e-05, |
|
"loss": 0.7151, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.9868005838791185e-05, |
|
"loss": 0.7377, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.9860373416529804e-05, |
|
"loss": 0.7527, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.985252803382911e-05, |
|
"loss": 0.7243, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.984446986013057e-05, |
|
"loss": 0.719, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.983619906947144e-05, |
|
"loss": 0.7639, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.9827715840480962e-05, |
|
"loss": 0.7259, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.9819020356376562e-05, |
|
"loss": 0.7116, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.9810112804959867e-05, |
|
"loss": 0.7193, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.980099337861264e-05, |
|
"loss": 0.7253, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.9791662274292638e-05, |
|
"loss": 0.6758, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.9782119693529358e-05, |
|
"loss": 0.6998, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.977236584241968e-05, |
|
"loss": 0.6996, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.9762400931623413e-05, |
|
"loss": 0.6726, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.9752225176358757e-05, |
|
"loss": 0.7184, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.9741838796397638e-05, |
|
"loss": 0.6589, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.9731242016060985e-05, |
|
"loss": 0.6933, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.972043506421386e-05, |
|
"loss": 0.6596, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.9709418174260523e-05, |
|
"loss": 0.6668, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.9698191584139402e-05, |
|
"loss": 0.6776, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.9686755536317945e-05, |
|
"loss": 0.6579, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.967511027778738e-05, |
|
"loss": 0.7176, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.9663256060057395e-05, |
|
"loss": 0.6585, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.965119313915068e-05, |
|
"loss": 0.667, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.9638921775597428e-05, |
|
"loss": 0.6656, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.9626442234429684e-05, |
|
"loss": 0.6649, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.961375478517564e-05, |
|
"loss": 0.6594, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.9600859701853796e-05, |
|
"loss": 0.6573, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.958775726296706e-05, |
|
"loss": 0.6532, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.9574447751496706e-05, |
|
"loss": 0.6542, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.95609314548963e-05, |
|
"loss": 0.6272, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.954720866508546e-05, |
|
"loss": 0.6486, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.953327967844356e-05, |
|
"loss": 0.6347, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.9519144795803342e-05, |
|
"loss": 0.657, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.95048043224444e-05, |
|
"loss": 0.6321, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.94902585680866e-05, |
|
"loss": 0.6416, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.9475507846883377e-05, |
|
"loss": 0.6549, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.9460552477414972e-05, |
|
"loss": 0.6452, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 1.9445392782681523e-05, |
|
"loss": 0.6469, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.9430029090096118e-05, |
|
"loss": 0.6323, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.94144617314777e-05, |
|
"loss": 0.6191, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.939869104304392e-05, |
|
"loss": 0.6201, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.9382717365403854e-05, |
|
"loss": 0.6576, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.9366541043550667e-05, |
|
"loss": 0.6481, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.9350162426854152e-05, |
|
"loss": 0.5918, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 1.933358186905318e-05, |
|
"loss": 0.6981, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.9316799728248074e-05, |
|
"loss": 0.6157, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.9299816366892865e-05, |
|
"loss": 0.6427, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.9282632151787462e-05, |
|
"loss": 0.6758, |
|
"step": 101 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.9265247454069736e-05, |
|
"loss": 0.6384, |
|
"step": 102 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.924766264920751e-05, |
|
"loss": 0.5973, |
|
"step": 103 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.922987811699042e-05, |
|
"loss": 0.6351, |
|
"step": 104 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.9211894241521757e-05, |
|
"loss": 0.5967, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.9193711411210138e-05, |
|
"loss": 0.6282, |
|
"step": 106 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.917533001876113e-05, |
|
"loss": 0.6332, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.9156750461168768e-05, |
|
"loss": 0.6619, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.9137973139706973e-05, |
|
"loss": 0.6336, |
|
"step": 109 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.91189984599209e-05, |
|
"loss": 0.6117, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.9099826831618168e-05, |
|
"loss": 0.6246, |
|
"step": 111 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.908045866886001e-05, |
|
"loss": 0.6245, |
|
"step": 112 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.9060894389952328e-05, |
|
"loss": 0.573, |
|
"step": 113 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.9041134417436674e-05, |
|
"loss": 0.6469, |
|
"step": 114 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.9021179178081107e-05, |
|
"loss": 0.6374, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.9001029102870982e-05, |
|
"loss": 0.5943, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.898068462699964e-05, |
|
"loss": 0.6565, |
|
"step": 117 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.8960146189859014e-05, |
|
"loss": 0.611, |
|
"step": 118 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.8939414235030137e-05, |
|
"loss": 0.5979, |
|
"step": 119 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.891848921027355e-05, |
|
"loss": 0.5958, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.889737156751965e-05, |
|
"loss": 0.6053, |
|
"step": 121 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.887606176285893e-05, |
|
"loss": 0.5823, |
|
"step": 122 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.8854560256532098e-05, |
|
"loss": 0.6342, |
|
"step": 123 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.883286751292018e-05, |
|
"loss": 0.6615, |
|
"step": 124 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 1.8810984000534457e-05, |
|
"loss": 0.6333, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.8788910192006363e-05, |
|
"loss": 0.5888, |
|
"step": 126 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.8766646564077265e-05, |
|
"loss": 0.5922, |
|
"step": 127 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.8744193597588185e-05, |
|
"loss": 0.5927, |
|
"step": 128 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.8721551777469397e-05, |
|
"loss": 0.5726, |
|
"step": 129 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.869872159272997e-05, |
|
"loss": 0.616, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.8675703536447178e-05, |
|
"loss": 0.5978, |
|
"step": 131 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 1.8652498105755898e-05, |
|
"loss": 0.5984, |
|
"step": 132 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.862910580183782e-05, |
|
"loss": 0.6455, |
|
"step": 133 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.8605527129910663e-05, |
|
"loss": 0.5821, |
|
"step": 134 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.858176259921724e-05, |
|
"loss": 0.6221, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.8557812723014476e-05, |
|
"loss": 0.6218, |
|
"step": 136 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.853367801856231e-05, |
|
"loss": 0.6202, |
|
"step": 137 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.8509359007112523e-05, |
|
"loss": 0.6062, |
|
"step": 138 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.8484856213897496e-05, |
|
"loss": 0.6144, |
|
"step": 139 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.8460170168118857e-05, |
|
"loss": 0.6027, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.843530140293603e-05, |
|
"loss": 0.5999, |
|
"step": 141 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.841025045545477e-05, |
|
"loss": 0.574, |
|
"step": 142 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.8385017866715507e-05, |
|
"loss": 0.5678, |
|
"step": 143 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.8359604181681703e-05, |
|
"loss": 0.6106, |
|
"step": 144 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 1.833400994922806e-05, |
|
"loss": 0.5789, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.8308235722128674e-05, |
|
"loss": 0.5799, |
|
"step": 146 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.8282282057045087e-05, |
|
"loss": 0.6149, |
|
"step": 147 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.8256149514514284e-05, |
|
"loss": 0.5768, |
|
"step": 148 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.8229838658936566e-05, |
|
"loss": 0.5947, |
|
"step": 149 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.8203350058563366e-05, |
|
"loss": 0.5843, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.8176684285484985e-05, |
|
"loss": 0.5966, |
|
"step": 151 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.814984191561823e-05, |
|
"loss": 0.5823, |
|
"step": 152 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 1.8122823528693966e-05, |
|
"loss": 0.6425, |
|
"step": 153 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 1.809562970824462e-05, |
|
"loss": 0.5965, |
|
"step": 154 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 1.8068261041591548e-05, |
|
"loss": 0.5852, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 1.8040718119832378e-05, |
|
"loss": 0.5719, |
|
"step": 156 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 1.8013001537828213e-05, |
|
"loss": 0.6235, |
|
"step": 157 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 1.7985111894190827e-05, |
|
"loss": 0.6111, |
|
"step": 158 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 1.7957049791269684e-05, |
|
"loss": 0.5686, |
|
"step": 159 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.7928815835138976e-05, |
|
"loss": 0.5974, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.79004106355845e-05, |
|
"loss": 0.6141, |
|
"step": 161 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.7871834806090502e-05, |
|
"loss": 0.5686, |
|
"step": 162 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.7843088963826437e-05, |
|
"loss": 0.5842, |
|
"step": 163 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.7814173729633607e-05, |
|
"loss": 0.572, |
|
"step": 164 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.7785089728011798e-05, |
|
"loss": 0.5781, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 1.775583758710575e-05, |
|
"loss": 0.5929, |
|
"step": 166 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.772641793869162e-05, |
|
"loss": 0.5732, |
|
"step": 167 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.7696831418163318e-05, |
|
"loss": 0.5995, |
|
"step": 168 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.7667078664518796e-05, |
|
"loss": 0.5693, |
|
"step": 169 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.763716032034624e-05, |
|
"loss": 0.573, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.7607077031810204e-05, |
|
"loss": 0.576, |
|
"step": 171 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.7576829448637628e-05, |
|
"loss": 0.5658, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.7546418224103838e-05, |
|
"loss": 0.5602, |
|
"step": 173 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.7515844015018416e-05, |
|
"loss": 0.5764, |
|
"step": 174 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.7485107481711014e-05, |
|
"loss": 0.5784, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.7454209288017105e-05, |
|
"loss": 0.5901, |
|
"step": 176 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.7423150101263645e-05, |
|
"loss": 0.5796, |
|
"step": 177 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.7391930592254636e-05, |
|
"loss": 0.5961, |
|
"step": 178 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 1.7360551435256673e-05, |
|
"loss": 0.5925, |
|
"step": 179 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.7329013307984363e-05, |
|
"loss": 0.601, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.729731689158568e-05, |
|
"loss": 0.5748, |
|
"step": 181 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.726546287062727e-05, |
|
"loss": 0.5902, |
|
"step": 182 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.7233451933079663e-05, |
|
"loss": 0.6207, |
|
"step": 183 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.720128477030241e-05, |
|
"loss": 0.5962, |
|
"step": 184 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.7168962077029146e-05, |
|
"loss": 0.5547, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.7136484551352608e-05, |
|
"loss": 0.573, |
|
"step": 186 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 1.7103852894709517e-05, |
|
"loss": 0.6041, |
|
"step": 187 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 1.7071067811865477e-05, |
|
"loss": 0.5902, |
|
"step": 188 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 1.7038130010899716e-05, |
|
"loss": 0.5499, |
|
"step": 189 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 1.7005040203189818e-05, |
|
"loss": 0.5892, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 1.6971799103396332e-05, |
|
"loss": 0.5512, |
|
"step": 191 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 1.693840742944738e-05, |
|
"loss": 0.574, |
|
"step": 192 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 1.6904865902523098e-05, |
|
"loss": 0.5995, |
|
"step": 193 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 1.68711752470401e-05, |
|
"loss": 0.5861, |
|
"step": 194 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 1.6837336190635824e-05, |
|
"loss": 0.5991, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 1.6803349464152798e-05, |
|
"loss": 0.5985, |
|
"step": 196 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 1.6769215801622884e-05, |
|
"loss": 0.5471, |
|
"step": 197 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 1.6734935940251404e-05, |
|
"loss": 0.5734, |
|
"step": 198 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 1.6700510620401223e-05, |
|
"loss": 0.5515, |
|
"step": 199 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 1.666594058557677e-05, |
|
"loss": 0.6147, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.6631226582407954e-05, |
|
"loss": 0.592, |
|
"step": 201 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.659636936063407e-05, |
|
"loss": 0.5551, |
|
"step": 202 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.6561369673087588e-05, |
|
"loss": 0.5489, |
|
"step": 203 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.6526228275677892e-05, |
|
"loss": 0.5616, |
|
"step": 204 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.649094592737497e-05, |
|
"loss": 0.5751, |
|
"step": 205 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.6455523390193e-05, |
|
"loss": 0.575, |
|
"step": 206 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.641996142917391e-05, |
|
"loss": 0.5901, |
|
"step": 207 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.638426081237085e-05, |
|
"loss": 0.581, |
|
"step": 208 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.63484223108316e-05, |
|
"loss": 0.585, |
|
"step": 209 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.6312446698581915e-05, |
|
"loss": 0.5652, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.6276334752608823e-05, |
|
"loss": 0.5775, |
|
"step": 211 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.624008725284383e-05, |
|
"loss": 0.5748, |
|
"step": 212 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1.6203704982146073e-05, |
|
"loss": 0.5968, |
|
"step": 213 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.6167188726285433e-05, |
|
"loss": 0.5698, |
|
"step": 214 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.613053927392553e-05, |
|
"loss": 0.5618, |
|
"step": 215 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.6093757416606725e-05, |
|
"loss": 0.538, |
|
"step": 216 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.6056843948729e-05, |
|
"loss": 0.5684, |
|
"step": 217 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.6019799667534813e-05, |
|
"loss": 0.583, |
|
"step": 218 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.5982625373091877e-05, |
|
"loss": 0.5659, |
|
"step": 219 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.594532186827588e-05, |
|
"loss": 0.5608, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.5907889958753134e-05, |
|
"loss": 0.5544, |
|
"step": 221 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.5870330452963206e-05, |
|
"loss": 0.5679, |
|
"step": 222 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.5870330452963206e-05, |
|
"loss": 0.5388, |
|
"step": 223 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.5832644162101417e-05, |
|
"loss": 0.558, |
|
"step": 224 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.5794831900101352e-05, |
|
"loss": 0.5771, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.5756894483617268e-05, |
|
"loss": 0.535, |
|
"step": 226 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 1.571883273200646e-05, |
|
"loss": 0.5397, |
|
"step": 227 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.568064746731156e-05, |
|
"loss": 0.5313, |
|
"step": 228 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.564233951424279e-05, |
|
"loss": 0.5737, |
|
"step": 229 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.560390970016015e-05, |
|
"loss": 0.5334, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.5565358855055535e-05, |
|
"loss": 0.5564, |
|
"step": 231 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.552668781153484e-05, |
|
"loss": 0.5812, |
|
"step": 232 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 1.5487897404799932e-05, |
|
"loss": 0.5385, |
|
"step": 233 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 1.5448988472630654e-05, |
|
"loss": 0.5733, |
|
"step": 234 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 1.5409961855366718e-05, |
|
"loss": 0.594, |
|
"step": 235 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 1.5370818395889536e-05, |
|
"loss": 0.5372, |
|
"step": 236 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 1.533155893960405e-05, |
|
"loss": 0.5599, |
|
"step": 237 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 1.5292184334420434e-05, |
|
"loss": 0.5396, |
|
"step": 238 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 1.5252695430735824e-05, |
|
"loss": 0.552, |
|
"step": 239 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 1.521309308141592e-05, |
|
"loss": 0.5583, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 1.5173378141776569e-05, |
|
"loss": 0.5708, |
|
"step": 241 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 1.5133551469565313e-05, |
|
"loss": 0.5462, |
|
"step": 242 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 1.5093613924942837e-05, |
|
"loss": 0.581, |
|
"step": 243 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 1.5053566370464416e-05, |
|
"loss": 0.5743, |
|
"step": 244 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 1.5013409671061267e-05, |
|
"loss": 0.5609, |
|
"step": 245 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 1.4973144694021874e-05, |
|
"loss": 0.6103, |
|
"step": 246 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 1.4932772308973267e-05, |
|
"loss": 0.561, |
|
"step": 247 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.4892293387862221e-05, |
|
"loss": 0.5909, |
|
"step": 248 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.4851708804936449e-05, |
|
"loss": 0.5469, |
|
"step": 249 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.4811019436725684e-05, |
|
"loss": 0.5435, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.4770226162022799e-05, |
|
"loss": 0.5326, |
|
"step": 251 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.472932986186477e-05, |
|
"loss": 0.5658, |
|
"step": 252 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.4688331419513693e-05, |
|
"loss": 0.5912, |
|
"step": 253 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.4647231720437687e-05, |
|
"loss": 0.5641, |
|
"step": 254 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.4606031652291772e-05, |
|
"loss": 0.5332, |
|
"step": 255 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.4564732104898702e-05, |
|
"loss": 0.5527, |
|
"step": 256 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.4523333970229741e-05, |
|
"loss": 0.5444, |
|
"step": 257 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.4481838142385403e-05, |
|
"loss": 0.5516, |
|
"step": 258 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.4440245517576141e-05, |
|
"loss": 0.5258, |
|
"step": 259 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.4398556994102996e-05, |
|
"loss": 0.5104, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 1.4356773472338174e-05, |
|
"loss": 0.5477, |
|
"step": 261 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.4314895854705641e-05, |
|
"loss": 0.5375, |
|
"step": 262 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.4272925045661583e-05, |
|
"loss": 0.5793, |
|
"step": 263 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.4230861951674914e-05, |
|
"loss": 0.5597, |
|
"step": 264 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.4188707481207677e-05, |
|
"loss": 0.6143, |
|
"step": 265 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.4146462544695428e-05, |
|
"loss": 0.5553, |
|
"step": 266 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 1.410412805452757e-05, |
|
"loss": 0.5296, |
|
"step": 267 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.4061704925027653e-05, |
|
"loss": 0.5451, |
|
"step": 268 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.4019194072433626e-05, |
|
"loss": 0.5559, |
|
"step": 269 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.3976596414878044e-05, |
|
"loss": 0.5331, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.393391287236824e-05, |
|
"loss": 0.5572, |
|
"step": 271 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.3891144366766457e-05, |
|
"loss": 0.5492, |
|
"step": 272 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.3848291821769944e-05, |
|
"loss": 0.502, |
|
"step": 273 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.380535616289099e-05, |
|
"loss": 0.5721, |
|
"step": 274 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.3762338317436948e-05, |
|
"loss": 0.5543, |
|
"step": 275 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.3719239214490203e-05, |
|
"loss": 0.5464, |
|
"step": 276 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.3676059784888112e-05, |
|
"loss": 0.5541, |
|
"step": 277 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.363280096120289e-05, |
|
"loss": 0.5453, |
|
"step": 278 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.358946367772148e-05, |
|
"loss": 0.5163, |
|
"step": 279 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.3546048870425356e-05, |
|
"loss": 0.5263, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 1.3502557476970339e-05, |
|
"loss": 0.5369, |
|
"step": 281 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.3458990436666313e-05, |
|
"loss": 0.5603, |
|
"step": 282 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.3415348690456955e-05, |
|
"loss": 0.5526, |
|
"step": 283 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.3371633180899417e-05, |
|
"loss": 0.5255, |
|
"step": 284 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.3327844852143956e-05, |
|
"loss": 0.5572, |
|
"step": 285 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.3283984649913552e-05, |
|
"loss": 0.5599, |
|
"step": 286 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.3240053521483483e-05, |
|
"loss": 0.517, |
|
"step": 287 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.3196052415660856e-05, |
|
"loss": 0.5277, |
|
"step": 288 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 1.3151982282764126e-05, |
|
"loss": 0.5506, |
|
"step": 289 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 1.3107844074602566e-05, |
|
"loss": 0.532, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 1.3107844074602566e-05, |
|
"loss": 0.5637, |
|
"step": 291 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 1.3063638744455713e-05, |
|
"loss": 0.536, |
|
"step": 292 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 1.3019367247052781e-05, |
|
"loss": 0.5947, |
|
"step": 293 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 1.297503053855203e-05, |
|
"loss": 0.5287, |
|
"step": 294 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.2930629576520133e-05, |
|
"loss": 0.554, |
|
"step": 295 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.2886165319911474e-05, |
|
"loss": 0.562, |
|
"step": 296 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.2841638729047463e-05, |
|
"loss": 0.5398, |
|
"step": 297 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.2797050765595767e-05, |
|
"loss": 0.5341, |
|
"step": 298 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.2752402392549556e-05, |
|
"loss": 0.5519, |
|
"step": 299 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.2707694574206712e-05, |
|
"loss": 0.527, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 1.2662928276148985e-05, |
|
"loss": 0.5426, |
|
"step": 301 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.2618104465221148e-05, |
|
"loss": 0.5561, |
|
"step": 302 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.2573224109510112e-05, |
|
"loss": 0.5305, |
|
"step": 303 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.2528288178324029e-05, |
|
"loss": 0.565, |
|
"step": 304 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.2483297642171332e-05, |
|
"loss": 0.5429, |
|
"step": 305 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.2438253472739805e-05, |
|
"loss": 0.5725, |
|
"step": 306 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.2393156642875579e-05, |
|
"loss": 0.5878, |
|
"step": 307 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.234800812656212e-05, |
|
"loss": 0.5376, |
|
"step": 308 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.23028088988992e-05, |
|
"loss": 0.5502, |
|
"step": 309 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.2257559936081833e-05, |
|
"loss": 0.5396, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.2212262215379199e-05, |
|
"loss": 0.5494, |
|
"step": 311 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.2166916715113521e-05, |
|
"loss": 0.5773, |
|
"step": 312 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.2121524414638958e-05, |
|
"loss": 0.5631, |
|
"step": 313 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.207608629432043e-05, |
|
"loss": 0.5486, |
|
"step": 314 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 1.2030603335512467e-05, |
|
"loss": 0.5547, |
|
"step": 315 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.1985076520537995e-05, |
|
"loss": 0.5226, |
|
"step": 316 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.1939506832667129e-05, |
|
"loss": 0.5549, |
|
"step": 317 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.1893895256095937e-05, |
|
"loss": 0.5555, |
|
"step": 318 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.1848242775925188e-05, |
|
"loss": 0.5226, |
|
"step": 319 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.180255037813906e-05, |
|
"loss": 0.5165, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.1756819049583861e-05, |
|
"loss": 0.529, |
|
"step": 321 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.1711049777946717e-05, |
|
"loss": 0.5699, |
|
"step": 322 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.166524355173422e-05, |
|
"loss": 0.5567, |
|
"step": 323 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.1619401360251104e-05, |
|
"loss": 0.5473, |
|
"step": 324 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.1573524193578863e-05, |
|
"loss": 0.5565, |
|
"step": 325 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.1527613042554368e-05, |
|
"loss": 0.5408, |
|
"step": 326 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.1481668898748474e-05, |
|
"loss": 0.5403, |
|
"step": 327 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 1.1435692754444604e-05, |
|
"loss": 0.5377, |
|
"step": 328 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.1389685602617302e-05, |
|
"loss": 0.5327, |
|
"step": 329 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.134364843691082e-05, |
|
"loss": 0.5259, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.1297582251617618e-05, |
|
"loss": 0.5424, |
|
"step": 331 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.1251488041656929e-05, |
|
"loss": 0.5182, |
|
"step": 332 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.1205366802553231e-05, |
|
"loss": 0.5209, |
|
"step": 333 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.1159219530414786e-05, |
|
"loss": 0.5292, |
|
"step": 334 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 1.1113047221912097e-05, |
|
"loss": 0.5792, |
|
"step": 335 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.1066850874256387e-05, |
|
"loss": 0.5524, |
|
"step": 336 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.1020631485178084e-05, |
|
"loss": 0.5481, |
|
"step": 337 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.0974390052905238e-05, |
|
"loss": 0.5496, |
|
"step": 338 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.0928127576141992e-05, |
|
"loss": 0.5446, |
|
"step": 339 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.0881845054046988e-05, |
|
"loss": 0.5433, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.0835543486211815e-05, |
|
"loss": 0.5208, |
|
"step": 341 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.0789223872639385e-05, |
|
"loss": 0.5707, |
|
"step": 342 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.0742887213722372e-05, |
|
"loss": 0.5387, |
|
"step": 343 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.0696534510221575e-05, |
|
"loss": 0.5474, |
|
"step": 344 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.065016676324433e-05, |
|
"loss": 0.5303, |
|
"step": 345 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.0603784974222862e-05, |
|
"loss": 0.546, |
|
"step": 346 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.0557390144892684e-05, |
|
"loss": 0.5021, |
|
"step": 347 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.0510983277270945e-05, |
|
"loss": 0.5499, |
|
"step": 348 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 1.0464565373634784e-05, |
|
"loss": 0.5445, |
|
"step": 349 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.0418137436499699e-05, |
|
"loss": 0.5578, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.0371700468597886e-05, |
|
"loss": 0.5312, |
|
"step": 351 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.0325255472856586e-05, |
|
"loss": 0.56, |
|
"step": 352 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.0278803452376416e-05, |
|
"loss": 0.501, |
|
"step": 353 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.0232345410409718e-05, |
|
"loss": 0.5123, |
|
"step": 354 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.018588235033888e-05, |
|
"loss": 0.5272, |
|
"step": 355 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 1.0139415275654671e-05, |
|
"loss": 0.5484, |
|
"step": 356 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 1.0092945189934558e-05, |
|
"loss": 0.5149, |
|
"step": 357 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 1.0046473096821047e-05, |
|
"loss": 0.5443, |
|
"step": 358 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 1e-05, |
|
"loss": 0.5473, |
|
"step": 359 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 9.953526903178956e-06, |
|
"loss": 0.5825, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 9.907054810065446e-06, |
|
"loss": 0.5895, |
|
"step": 361 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 9.860584724345334e-06, |
|
"loss": 0.54, |
|
"step": 362 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.81411764966112e-06, |
|
"loss": 0.5421, |
|
"step": 363 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.767654589590284e-06, |
|
"loss": 0.5609, |
|
"step": 364 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.721196547623585e-06, |
|
"loss": 0.5303, |
|
"step": 365 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.674744527143419e-06, |
|
"loss": 0.5431, |
|
"step": 366 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.628299531402118e-06, |
|
"loss": 0.5301, |
|
"step": 367 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.581862563500305e-06, |
|
"loss": 0.5379, |
|
"step": 368 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.535434626365221e-06, |
|
"loss": 0.5389, |
|
"step": 369 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.489016722729059e-06, |
|
"loss": 0.5143, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.442609855107317e-06, |
|
"loss": 0.5171, |
|
"step": 371 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.39621502577714e-06, |
|
"loss": 0.5292, |
|
"step": 372 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.349833236755675e-06, |
|
"loss": 0.548, |
|
"step": 373 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.30346548977843e-06, |
|
"loss": 0.5394, |
|
"step": 374 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.257112786277631e-06, |
|
"loss": 0.4985, |
|
"step": 375 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.21077612736062e-06, |
|
"loss": 0.5576, |
|
"step": 376 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 9.164456513788186e-06, |
|
"loss": 0.5745, |
|
"step": 377 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 9.118154945953015e-06, |
|
"loss": 0.5394, |
|
"step": 378 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 9.07187242385801e-06, |
|
"loss": 0.5688, |
|
"step": 379 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 9.025609947094765e-06, |
|
"loss": 0.5443, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 8.979368514821917e-06, |
|
"loss": 0.4974, |
|
"step": 381 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 8.933149125743615e-06, |
|
"loss": 0.5325, |
|
"step": 382 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 8.88695277808791e-06, |
|
"loss": 0.544, |
|
"step": 383 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.840780469585215e-06, |
|
"loss": 0.5299, |
|
"step": 384 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.79463319744677e-06, |
|
"loss": 0.522, |
|
"step": 385 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.748511958343076e-06, |
|
"loss": 0.5446, |
|
"step": 386 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.702417748382384e-06, |
|
"loss": 0.5351, |
|
"step": 387 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.656351563089185e-06, |
|
"loss": 0.5184, |
|
"step": 388 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 8.610314397382701e-06, |
|
"loss": 0.4872, |
|
"step": 389 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.564307245555403e-06, |
|
"loss": 0.535, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.51833110125153e-06, |
|
"loss": 0.5197, |
|
"step": 391 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.472386957445635e-06, |
|
"loss": 0.5015, |
|
"step": 392 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.426475806421139e-06, |
|
"loss": 0.5493, |
|
"step": 393 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.380598639748898e-06, |
|
"loss": 0.565, |
|
"step": 394 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.334756448265782e-06, |
|
"loss": 0.4995, |
|
"step": 395 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.288950222053287e-06, |
|
"loss": 0.5379, |
|
"step": 396 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 8.243180950416142e-06, |
|
"loss": 0.5056, |
|
"step": 397 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 8.243180950416142e-06, |
|
"loss": 0.5248, |
|
"step": 398 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 8.197449621860944e-06, |
|
"loss": 0.5495, |
|
"step": 399 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 8.151757224074815e-06, |
|
"loss": 0.5626, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 8.106104743904065e-06, |
|
"loss": 0.5502, |
|
"step": 401 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 8.060493167332874e-06, |
|
"loss": 0.5708, |
|
"step": 402 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 8.014923479462007e-06, |
|
"loss": 0.5454, |
|
"step": 403 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 7.969396664487534e-06, |
|
"loss": 0.4988, |
|
"step": 404 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 7.923913705679573e-06, |
|
"loss": 0.5065, |
|
"step": 405 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 7.878475585361045e-06, |
|
"loss": 0.5241, |
|
"step": 406 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 7.833083284886484e-06, |
|
"loss": 0.5259, |
|
"step": 407 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 7.787737784620803e-06, |
|
"loss": 0.5474, |
|
"step": 408 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 7.74244006391817e-06, |
|
"loss": 0.5107, |
|
"step": 409 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 7.697191101100802e-06, |
|
"loss": 0.5166, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 7.651991873437884e-06, |
|
"loss": 0.5427, |
|
"step": 411 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 7.606843357124426e-06, |
|
"loss": 0.5187, |
|
"step": 412 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 7.561746527260197e-06, |
|
"loss": 0.5332, |
|
"step": 413 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 7.516702357828672e-06, |
|
"loss": 0.493, |
|
"step": 414 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 7.471711821675973e-06, |
|
"loss": 0.5354, |
|
"step": 415 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 7.42677589048989e-06, |
|
"loss": 0.544, |
|
"step": 416 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 7.381895534778852e-06, |
|
"loss": 0.5487, |
|
"step": 417 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 7.337071723851018e-06, |
|
"loss": 0.5368, |
|
"step": 418 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 7.292305425793291e-06, |
|
"loss": 0.5248, |
|
"step": 419 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 7.247597607450446e-06, |
|
"loss": 0.5321, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 7.2029492344042395e-06, |
|
"loss": 0.5125, |
|
"step": 421 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 7.1583612709525405e-06, |
|
"loss": 0.535, |
|
"step": 422 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 7.113834680088527e-06, |
|
"loss": 0.537, |
|
"step": 423 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 7.06937042347987e-06, |
|
"loss": 0.5298, |
|
"step": 424 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 7.024969461447973e-06, |
|
"loss": 0.5386, |
|
"step": 425 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 6.980632752947221e-06, |
|
"loss": 0.5347, |
|
"step": 426 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 6.936361255544288e-06, |
|
"loss": 0.5235, |
|
"step": 427 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 6.892155925397437e-06, |
|
"loss": 0.4912, |
|
"step": 428 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 6.848017717235878e-06, |
|
"loss": 0.5415, |
|
"step": 429 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 6.803947584339148e-06, |
|
"loss": 0.5037, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 6.7599464785165195e-06, |
|
"loss": 0.5229, |
|
"step": 431 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 6.716015350086449e-06, |
|
"loss": 0.5535, |
|
"step": 432 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 6.672155147856046e-06, |
|
"loss": 0.487, |
|
"step": 433 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 6.628366819100586e-06, |
|
"loss": 0.5187, |
|
"step": 434 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 6.584651309543049e-06, |
|
"loss": 0.5232, |
|
"step": 435 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 6.54100956333369e-06, |
|
"loss": 0.5217, |
|
"step": 436 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 6.497442523029663e-06, |
|
"loss": 0.5125, |
|
"step": 437 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 6.453951129574644e-06, |
|
"loss": 0.5418, |
|
"step": 438 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 6.410536322278523e-06, |
|
"loss": 0.5503, |
|
"step": 439 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 6.3671990387971096e-06, |
|
"loss": 0.5442, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 6.32394021511189e-06, |
|
"loss": 0.5457, |
|
"step": 441 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 6.280760785509802e-06, |
|
"loss": 0.5364, |
|
"step": 442 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 6.237661682563055e-06, |
|
"loss": 0.5616, |
|
"step": 443 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 6.194643837109015e-06, |
|
"loss": 0.5177, |
|
"step": 444 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 6.151708178230057e-06, |
|
"loss": 0.5068, |
|
"step": 445 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 6.108855633233546e-06, |
|
"loss": 0.4901, |
|
"step": 446 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 6.066087127631761e-06, |
|
"loss": 0.4956, |
|
"step": 447 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 6.0234035851219604e-06, |
|
"loss": 0.5101, |
|
"step": 448 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 6.0234035851219604e-06, |
|
"loss": 0.5244, |
|
"step": 449 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 5.9808059275663775e-06, |
|
"loss": 0.4935, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.93829507497235e-06, |
|
"loss": 0.544, |
|
"step": 451 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.895871945472434e-06, |
|
"loss": 0.5254, |
|
"step": 452 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.853537455304575e-06, |
|
"loss": 0.551, |
|
"step": 453 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.811292518792324e-06, |
|
"loss": 0.5516, |
|
"step": 454 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.769138048325087e-06, |
|
"loss": 0.4987, |
|
"step": 455 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.72707495433842e-06, |
|
"loss": 0.5088, |
|
"step": 456 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 5.685104145294364e-06, |
|
"loss": 0.5168, |
|
"step": 457 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 5.643226527661825e-06, |
|
"loss": 0.5045, |
|
"step": 458 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 5.601443005897012e-06, |
|
"loss": 0.5314, |
|
"step": 459 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 5.55975448242386e-06, |
|
"loss": 0.4984, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 5.5181618576146e-06, |
|
"loss": 0.5256, |
|
"step": 461 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 5.476666029770259e-06, |
|
"loss": 0.529, |
|
"step": 462 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 5.435267895101303e-06, |
|
"loss": 0.4984, |
|
"step": 463 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 5.393968347708233e-06, |
|
"loss": 0.5337, |
|
"step": 464 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 5.352768279562315e-06, |
|
"loss": 0.5347, |
|
"step": 465 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 5.352768279562315e-06, |
|
"loss": 0.513, |
|
"step": 466 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 5.311668580486311e-06, |
|
"loss": 0.5144, |
|
"step": 467 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 5.270670138135234e-06, |
|
"loss": 0.5493, |
|
"step": 468 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 5.229773837977208e-06, |
|
"loss": 0.513, |
|
"step": 469 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 5.188980563274315e-06, |
|
"loss": 0.5349, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 5.148291195063555e-06, |
|
"loss": 0.5332, |
|
"step": 471 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 5.107706612137776e-06, |
|
"loss": 0.519, |
|
"step": 472 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 5.067227691026737e-06, |
|
"loss": 0.5071, |
|
"step": 473 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 5.026855305978129e-06, |
|
"loss": 0.5515, |
|
"step": 474 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 4.986590328938734e-06, |
|
"loss": 0.4935, |
|
"step": 475 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 4.946433629535585e-06, |
|
"loss": 0.5037, |
|
"step": 476 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 4.906386075057164e-06, |
|
"loss": 0.5446, |
|
"step": 477 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.866448530434692e-06, |
|
"loss": 0.5124, |
|
"step": 478 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.826621858223431e-06, |
|
"loss": 0.5134, |
|
"step": 479 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.786906918584083e-06, |
|
"loss": 0.5448, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.747304569264179e-06, |
|
"loss": 0.5001, |
|
"step": 481 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.707815665579569e-06, |
|
"loss": 0.5464, |
|
"step": 482 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.668441060395958e-06, |
|
"loss": 0.5254, |
|
"step": 483 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.629181604110464e-06, |
|
"loss": 0.5247, |
|
"step": 484 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 4.5900381446332854e-06, |
|
"loss": 0.5378, |
|
"step": 485 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 4.551011527369348e-06, |
|
"loss": 0.4988, |
|
"step": 486 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 4.512102595200073e-06, |
|
"loss": 0.5605, |
|
"step": 487 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 4.4733121884651665e-06, |
|
"loss": 0.5165, |
|
"step": 488 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 4.434641144944464e-06, |
|
"loss": 0.5223, |
|
"step": 489 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 4.3960902998398524e-06, |
|
"loss": 0.5199, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 4.357660485757211e-06, |
|
"loss": 0.5419, |
|
"step": 491 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 4.319352532688444e-06, |
|
"loss": 0.5601, |
|
"step": 492 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 4.281167267993541e-06, |
|
"loss": 0.5347, |
|
"step": 493 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 4.243105516382732e-06, |
|
"loss": 0.5401, |
|
"step": 494 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 4.205168099898652e-06, |
|
"loss": 0.5231, |
|
"step": 495 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 4.167355837898585e-06, |
|
"loss": 0.5185, |
|
"step": 496 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 4.129669547036798e-06, |
|
"loss": 0.5064, |
|
"step": 497 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 4.092110041246865e-06, |
|
"loss": 0.5138, |
|
"step": 498 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 4.054678131724128e-06, |
|
"loss": 0.5255, |
|
"step": 499 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 4.017374626908125e-06, |
|
"loss": 0.5259, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.980200332465189e-06, |
|
"loss": 0.5563, |
|
"step": 501 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.943156051271003e-06, |
|
"loss": 0.495, |
|
"step": 502 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.906242583393279e-06, |
|
"loss": 0.5451, |
|
"step": 503 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.8694607260744745e-06, |
|
"loss": 0.5441, |
|
"step": 504 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.832811273714569e-06, |
|
"loss": 0.5035, |
|
"step": 505 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.7962950178539282e-06, |
|
"loss": 0.547, |
|
"step": 506 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.7599127471561746e-06, |
|
"loss": 0.5315, |
|
"step": 507 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.7236652473911817e-06, |
|
"loss": 0.4933, |
|
"step": 508 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.687553301418092e-06, |
|
"loss": 0.5406, |
|
"step": 509 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.651577689168405e-06, |
|
"loss": 0.5649, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 3.6157391876291535e-06, |
|
"loss": 0.5097, |
|
"step": 511 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 3.580038570826093e-06, |
|
"loss": 0.5512, |
|
"step": 512 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 3.5444766098070037e-06, |
|
"loss": 0.5081, |
|
"step": 513 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 3.509054072625031e-06, |
|
"loss": 0.4807, |
|
"step": 514 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 3.473771724322108e-06, |
|
"loss": 0.5423, |
|
"step": 515 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 3.4386303269124142e-06, |
|
"loss": 0.5105, |
|
"step": 516 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 3.4036306393659324e-06, |
|
"loss": 0.5116, |
|
"step": 517 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 3.3687734175920505e-06, |
|
"loss": 0.5197, |
|
"step": 518 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 3.334059414423233e-06, |
|
"loss": 0.5344, |
|
"step": 519 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 3.299489379598777e-06, |
|
"loss": 0.4957, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 3.265064059748597e-06, |
|
"loss": 0.5288, |
|
"step": 521 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 3.2307841983771182e-06, |
|
"loss": 0.5743, |
|
"step": 522 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 3.196650535847201e-06, |
|
"loss": 0.5134, |
|
"step": 523 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 3.162663809364178e-06, |
|
"loss": 0.5275, |
|
"step": 524 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 3.1288247529599035e-06, |
|
"loss": 0.5193, |
|
"step": 525 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 3.095134097476904e-06, |
|
"loss": 0.5404, |
|
"step": 526 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 3.061592570552623e-06, |
|
"loss": 0.5008, |
|
"step": 527 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 3.0282008966036647e-06, |
|
"loss": 0.4978, |
|
"step": 528 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 2.9949597968101883e-06, |
|
"loss": 0.514, |
|
"step": 529 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 2.9618699891002843e-06, |
|
"loss": 0.5425, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 2.9289321881345257e-06, |
|
"loss": 0.5084, |
|
"step": 531 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 2.8961471052904855e-06, |
|
"loss": 0.5268, |
|
"step": 532 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.8635154486473935e-06, |
|
"loss": 0.5086, |
|
"step": 533 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.831037922970855e-06, |
|
"loss": 0.5416, |
|
"step": 534 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.798715229697592e-06, |
|
"loss": 0.5179, |
|
"step": 535 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.7665480669203383e-06, |
|
"loss": 0.506, |
|
"step": 536 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.7345371293727306e-06, |
|
"loss": 0.5367, |
|
"step": 537 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.702683108414326e-06, |
|
"loss": 0.5344, |
|
"step": 538 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.6709866920156434e-06, |
|
"loss": 0.5257, |
|
"step": 539 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.639448564743328e-06, |
|
"loss": 0.5092, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.6080694077453663e-06, |
|
"loss": 0.5196, |
|
"step": 541 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.57684989873636e-06, |
|
"loss": 0.5124, |
|
"step": 542 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.545790711982897e-06, |
|
"loss": 0.5225, |
|
"step": 543 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.514892518288988e-06, |
|
"loss": 0.533, |
|
"step": 544 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.484155984981588e-06, |
|
"loss": 0.5101, |
|
"step": 545 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 2.4535817758961644e-06, |
|
"loss": 0.5121, |
|
"step": 546 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 2.423170551362375e-06, |
|
"loss": 0.518, |
|
"step": 547 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 2.3929229681898005e-06, |
|
"loss": 0.5423, |
|
"step": 548 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 2.3628396796537588e-06, |
|
"loss": 0.486, |
|
"step": 549 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 2.332921335481205e-06, |
|
"loss": 0.5357, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 2.3031685818366844e-06, |
|
"loss": 0.5072, |
|
"step": 551 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 2.2735820613083837e-06, |
|
"loss": 0.541, |
|
"step": 552 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 2.2441624128942496e-06, |
|
"loss": 0.5146, |
|
"step": 553 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 2.2149102719882044e-06, |
|
"loss": 0.532, |
|
"step": 554 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 2.185826270366395e-06, |
|
"loss": 0.4899, |
|
"step": 555 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 2.156911036173568e-06, |
|
"loss": 0.5348, |
|
"step": 556 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 2.1281651939094996e-06, |
|
"loss": 0.5124, |
|
"step": 557 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 2.0995893644155007e-06, |
|
"loss": 0.5052, |
|
"step": 558 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 2.0711841648610254e-06, |
|
"loss": 0.5208, |
|
"step": 559 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 2.0429502087303164e-06, |
|
"loss": 0.5154, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 2.0148881058091764e-06, |
|
"loss": 0.5214, |
|
"step": 561 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 1.9869984621717888e-06, |
|
"loss": 0.5241, |
|
"step": 562 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 1.959281880167626e-06, |
|
"loss": 0.5545, |
|
"step": 563 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 1.931738958408457e-06, |
|
"loss": 0.539, |
|
"step": 564 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 1.904370291755383e-06, |
|
"loss": 0.504, |
|
"step": 565 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 1.8771764713060359e-06, |
|
"loss": 0.5185, |
|
"step": 566 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.8501580843817724e-06, |
|
"loss": 0.5339, |
|
"step": 567 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.8233157145150183e-06, |
|
"loss": 0.5617, |
|
"step": 568 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.796649941436638e-06, |
|
"loss": 0.5167, |
|
"step": 569 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.7701613410634367e-06, |
|
"loss": 0.5188, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.7438504854857164e-06, |
|
"loss": 0.5493, |
|
"step": 571 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.717717942954914e-06, |
|
"loss": 0.5308, |
|
"step": 572 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.6917642778713294e-06, |
|
"loss": 0.5231, |
|
"step": 573 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.6659900507719406e-06, |
|
"loss": 0.5274, |
|
"step": 574 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.6403958183182977e-06, |
|
"loss": 0.5454, |
|
"step": 575 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.614982133284495e-06, |
|
"loss": 0.5175, |
|
"step": 576 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.5897495445452338e-06, |
|
"loss": 0.5067, |
|
"step": 577 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.5646985970639717e-06, |
|
"loss": 0.5192, |
|
"step": 578 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.5398298318811467e-06, |
|
"loss": 0.5074, |
|
"step": 579 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.5151437861025032e-06, |
|
"loss": 0.5183, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.4906409928874788e-06, |
|
"loss": 0.5311, |
|
"step": 581 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.466321981437694e-06, |
|
"loss": 0.5085, |
|
"step": 582 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.4421872769855262e-06, |
|
"loss": 0.534, |
|
"step": 583 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.4182374007827605e-06, |
|
"loss": 0.5273, |
|
"step": 584 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.3944728700893395e-06, |
|
"loss": 0.515, |
|
"step": 585 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.3708941981621814e-06, |
|
"loss": 0.5524, |
|
"step": 586 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 1.3475018942441055e-06, |
|
"loss": 0.518, |
|
"step": 587 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 1.324296463552821e-06, |
|
"loss": 0.5227, |
|
"step": 588 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 1.3012784072700335e-06, |
|
"loss": 0.5033, |
|
"step": 589 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 1.2784482225306061e-06, |
|
"loss": 0.495, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 1.2558064024118177e-06, |
|
"loss": 0.5146, |
|
"step": 591 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 1.2333534359227383e-06, |
|
"loss": 0.5102, |
|
"step": 592 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 1.2110898079936394e-06, |
|
"loss": 0.5031, |
|
"step": 593 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 1.1890159994655425e-06, |
|
"loss": 0.5057, |
|
"step": 594 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 1.1671324870798195e-06, |
|
"loss": 0.54, |
|
"step": 595 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 1.1454397434679022e-06, |
|
"loss": 0.5218, |
|
"step": 596 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 1.1239382371410713e-06, |
|
"loss": 0.5069, |
|
"step": 597 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 1.1026284324803493e-06, |
|
"loss": 0.5505, |
|
"step": 598 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 1.0815107897264555e-06, |
|
"loss": 0.5026, |
|
"step": 599 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 1.060585764969867e-06, |
|
"loss": 0.5239, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 1.039853810140987e-06, |
|
"loss": 0.509, |
|
"step": 601 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 1.0193153730003603e-06, |
|
"loss": 0.5112, |
|
"step": 602 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 9.989708971290212e-07, |
|
"loss": 0.5269, |
|
"step": 603 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 9.788208219188932e-07, |
|
"loss": 0.4989, |
|
"step": 604 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 9.588655825633274e-07, |
|
"loss": 0.5488, |
|
"step": 605 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 9.391056100476736e-07, |
|
"loss": 0.5047, |
|
"step": 606 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 9.195413311399948e-07, |
|
"loss": 0.518, |
|
"step": 607 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 9.001731683818338e-07, |
|
"loss": 0.513, |
|
"step": 608 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 8.810015400790994e-07, |
|
"loss": 0.5383, |
|
"step": 609 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 8.620268602930271e-07, |
|
"loss": 0.5455, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 8.432495388312345e-07, |
|
"loss": 0.538, |
|
"step": 611 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 8.246699812388714e-07, |
|
"loss": 0.5231, |
|
"step": 612 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 8.062885887898642e-07, |
|
"loss": 0.5171, |
|
"step": 613 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 7.881057584782448e-07, |
|
"loss": 0.5394, |
|
"step": 614 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 7.701218830095825e-07, |
|
"loss": 0.5486, |
|
"step": 615 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 7.523373507924947e-07, |
|
"loss": 0.5167, |
|
"step": 616 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 7.34752545930264e-07, |
|
"loss": 0.5282, |
|
"step": 617 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 7.17367848212539e-07, |
|
"loss": 0.5135, |
|
"step": 618 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 7.001836331071365e-07, |
|
"loss": 0.5359, |
|
"step": 619 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 6.83200271751927e-07, |
|
"loss": 0.571, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 6.664181309468232e-07, |
|
"loss": 0.4974, |
|
"step": 621 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 6.498375731458529e-07, |
|
"loss": 0.5407, |
|
"step": 622 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 6.334589564493343e-07, |
|
"loss": 0.5143, |
|
"step": 623 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 6.17282634596148e-07, |
|
"loss": 0.5116, |
|
"step": 624 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 6.013089569560826e-07, |
|
"loss": 0.5419, |
|
"step": 625 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 5.85538268522301e-07, |
|
"loss": 0.5208, |
|
"step": 626 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 5.699709099038852e-07, |
|
"loss": 0.5154, |
|
"step": 627 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 5.546072173184791e-07, |
|
"loss": 0.5256, |
|
"step": 628 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 5.394475225850338e-07, |
|
"loss": 0.538, |
|
"step": 629 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 5.244921531166247e-07, |
|
"loss": 0.5175, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 5.097414319134042e-07, |
|
"loss": 0.5592, |
|
"step": 631 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 4.951956775556e-07, |
|
"loss": 0.509, |
|
"step": 632 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 4.808552041966608e-07, |
|
"loss": 0.5273, |
|
"step": 633 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 4.667203215564431e-07, |
|
"loss": 0.4977, |
|
"step": 634 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 4.5279133491454406e-07, |
|
"loss": 0.5225, |
|
"step": 635 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 4.3906854510370245e-07, |
|
"loss": 0.5204, |
|
"step": 636 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 4.255522485032959e-07, |
|
"loss": 0.5008, |
|
"step": 637 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 4.1224273703294515e-07, |
|
"loss": 0.5133, |
|
"step": 638 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 3.991402981462045e-07, |
|
"loss": 0.4845, |
|
"step": 639 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 3.862452148243623e-07, |
|
"loss": 0.4926, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 3.735577655703171e-07, |
|
"loss": 0.5518, |
|
"step": 641 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 3.610782244025768e-07, |
|
"loss": 0.4797, |
|
"step": 642 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 3.488068608493245e-07, |
|
"loss": 0.5197, |
|
"step": 643 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 3.367439399426087e-07, |
|
"loss": 0.505, |
|
"step": 644 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 3.2488972221262125e-07, |
|
"loss": 0.5488, |
|
"step": 645 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 3.132444636820575e-07, |
|
"loss": 0.5387, |
|
"step": 646 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 3.018084158606005e-07, |
|
"loss": 0.5521, |
|
"step": 647 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 2.905818257394799e-07, |
|
"loss": 0.5214, |
|
"step": 648 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 2.79564935786143e-07, |
|
"loss": 0.5068, |
|
"step": 649 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 2.687579839390153e-07, |
|
"loss": 0.4791, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 2.5816120360236176e-07, |
|
"loss": 0.4895, |
|
"step": 651 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 2.4777482364124695e-07, |
|
"loss": 0.5501, |
|
"step": 652 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 2.375990683765894e-07, |
|
"loss": 0.4985, |
|
"step": 653 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 2.2763415758032316e-07, |
|
"loss": 0.5852, |
|
"step": 654 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 2.1788030647064363e-07, |
|
"loss": 0.5335, |
|
"step": 655 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 2.0833772570736376e-07, |
|
"loss": 0.5512, |
|
"step": 656 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 1.9900662138736204e-07, |
|
"loss": 0.5202, |
|
"step": 657 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 1.8988719504013375e-07, |
|
"loss": 0.4981, |
|
"step": 658 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 1.809796436234379e-07, |
|
"loss": 0.5497, |
|
"step": 659 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 1.7228415951904165e-07, |
|
"loss": 0.5066, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 1.6380093052856482e-07, |
|
"loss": 0.5474, |
|
"step": 661 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 1.5553013986942645e-07, |
|
"loss": 0.5365, |
|
"step": 662 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 1.474719661708901e-07, |
|
"loss": 0.5291, |
|
"step": 663 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 1.3962658347019819e-07, |
|
"loss": 0.5479, |
|
"step": 664 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 1.3199416120881714e-07, |
|
"loss": 0.5541, |
|
"step": 665 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 1.245748642287814e-07, |
|
"loss": 0.5235, |
|
"step": 666 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 1.1736885276912747e-07, |
|
"loss": 0.536, |
|
"step": 667 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 1.103762824624377e-07, |
|
"loss": 0.5359, |
|
"step": 668 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 1.0359730433147308e-07, |
|
"loss": 0.485, |
|
"step": 669 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 9.70320647859213e-08, |
|
"loss": 0.5196, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 9.068070561922382e-08, |
|
"loss": 0.5191, |
|
"step": 671 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 8.454336400552154e-08, |
|
"loss": 0.5022, |
|
"step": 672 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 7.862017249668507e-08, |
|
"loss": 0.539, |
|
"step": 673 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 7.291125901946027e-08, |
|
"loss": 0.5083, |
|
"step": 674 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 6.741674687269828e-08, |
|
"loss": 0.5325, |
|
"step": 675 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 6.21367547246976e-08, |
|
"loss": 0.4958, |
|
"step": 676 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 5.70713966106351e-08, |
|
"loss": 0.512, |
|
"step": 677 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"step": 677, |
|
"total_flos": 6.630352487616348e+17, |
|
"train_loss": 0.6011118755030668, |
|
"train_runtime": 14577.6167, |
|
"train_samples_per_second": 1.488, |
|
"train_steps_per_second": 0.046 |
|
} |
|
], |
|
"max_steps": 677, |
|
"num_train_epochs": 1, |
|
"total_flos": 6.630352487616348e+17, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|