| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "global_step": 376, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.1764705882352942e-06, | |
| "loss": 1.3193, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.3529411764705885e-06, | |
| "loss": 1.2891, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.529411764705883e-06, | |
| "loss": 1.3105, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.705882352941177e-06, | |
| "loss": 1.2383, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.882352941176471e-06, | |
| "loss": 0.8359, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 7.058823529411766e-06, | |
| "loss": 0.49, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.23529411764706e-06, | |
| "loss": 0.1881, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.411764705882354e-06, | |
| "loss": 0.2035, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.0588235294117648e-05, | |
| "loss": 0.1893, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.1764705882352942e-05, | |
| "loss": 0.2448, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.2941176470588238e-05, | |
| "loss": 0.1973, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.4117647058823532e-05, | |
| "loss": 0.2133, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.5294117647058822e-05, | |
| "loss": 0.2087, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.647058823529412e-05, | |
| "loss": 0.2408, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.7647058823529414e-05, | |
| "loss": 0.1686, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.8823529411764708e-05, | |
| "loss": 0.1917, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2e-05, | |
| "loss": 0.2241, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9999835072185805e-05, | |
| "loss": 0.2195, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.999934029418346e-05, | |
| "loss": 0.1847, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9998515682313485e-05, | |
| "loss": 0.1765, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.999736126377618e-05, | |
| "loss": 0.1572, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.999587707665068e-05, | |
| "loss": 0.1677, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.999406316989374e-05, | |
| "loss": 0.1906, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9991919603338088e-05, | |
| "loss": 0.153, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.998944644769048e-05, | |
| "loss": 0.1473, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9986643784529346e-05, | |
| "loss": 0.1345, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9983511706302102e-05, | |
| "loss": 0.1476, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9980050316322118e-05, | |
| "loss": 0.1219, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.997625972876529e-05, | |
| "loss": 0.1336, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.997214006866628e-05, | |
| "loss": 0.1133, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9967691471914392e-05, | |
| "loss": 0.1424, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.99629140852491e-05, | |
| "loss": 0.123, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9957808066255187e-05, | |
| "loss": 0.1199, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9952373583357566e-05, | |
| "loss": 0.1308, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.994661081581571e-05, | |
| "loss": 0.1172, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.9940519953717762e-05, | |
| "loss": 0.1299, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.993410119797422e-05, | |
| "loss": 0.1358, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.9927354760311365e-05, | |
| "loss": 0.121, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.992028086326424e-05, | |
| "loss": 0.1273, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.991287974016932e-05, | |
| "loss": 0.1215, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.9905151635156813e-05, | |
| "loss": 0.1239, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.9897096803142616e-05, | |
| "loss": 0.1178, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.988871550981989e-05, | |
| "loss": 0.1135, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.988000803165032e-05, | |
| "loss": 0.1103, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.9870974655854974e-05, | |
| "loss": 0.124, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.9861615680404833e-05, | |
| "loss": 0.1031, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.985193141401097e-05, | |
| "loss": 0.1032, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.9841922176114366e-05, | |
| "loss": 0.1151, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.9831588296875367e-05, | |
| "loss": 0.1061, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.982093011716279e-05, | |
| "loss": 0.1019, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.9809947988542696e-05, | |
| "loss": 0.1004, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.979864227326678e-05, | |
| "loss": 0.0971, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.9787013344260422e-05, | |
| "loss": 0.1137, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.9775061585110387e-05, | |
| "loss": 0.1052, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.976278739005218e-05, | |
| "loss": 0.1034, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.9750191163957042e-05, | |
| "loss": 0.0911, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.9737273322318565e-05, | |
| "loss": 0.1107, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.972403429123904e-05, | |
| "loss": 0.1055, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.971047450741535e-05, | |
| "loss": 0.106, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.9696594418124598e-05, | |
| "loss": 0.0934, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.9682394481209338e-05, | |
| "loss": 0.0986, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.966787516506249e-05, | |
| "loss": 0.1112, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.9653036948611864e-05, | |
| "loss": 0.1075, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.9637880321304387e-05, | |
| "loss": 0.1089, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.962240578308993e-05, | |
| "loss": 0.0972, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.9606613844404853e-05, | |
| "loss": 0.0898, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.9590505026155146e-05, | |
| "loss": 0.088, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.9574079859699236e-05, | |
| "loss": 0.1041, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.955733888683049e-05, | |
| "loss": 0.1028, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.9540282659759317e-05, | |
| "loss": 0.0876, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.9522911741094966e-05, | |
| "loss": 0.088, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.9505226703826973e-05, | |
| "loss": 0.0905, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.948722813130624e-05, | |
| "loss": 0.0945, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.9468916617225814e-05, | |
| "loss": 0.0978, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.9450292765601287e-05, | |
| "loss": 0.1132, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.94313571907509e-05, | |
| "loss": 0.0826, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.941211051727524e-05, | |
| "loss": 0.1053, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.939255338003666e-05, | |
| "loss": 0.1049, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.937268642413835e-05, | |
| "loss": 0.1079, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.9352510304903017e-05, | |
| "loss": 0.1078, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.9332025687851325e-05, | |
| "loss": 0.0937, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.931123324867989e-05, | |
| "loss": 0.0822, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.929013367323902e-05, | |
| "loss": 0.1162, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.926872765751009e-05, | |
| "loss": 0.0896, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.9247015907582574e-05, | |
| "loss": 0.1021, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.9224999139630766e-05, | |
| "loss": 0.1036, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.920267807989015e-05, | |
| "loss": 0.0952, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.918005346463344e-05, | |
| "loss": 0.0869, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.9157126040146307e-05, | |
| "loss": 0.0923, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.9133896562702746e-05, | |
| "loss": 0.0983, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.911036579854016e-05, | |
| "loss": 0.1044, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.9086534523834032e-05, | |
| "loss": 0.0901, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.906240352467238e-05, | |
| "loss": 0.0774, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.9037973597029796e-05, | |
| "loss": 0.0863, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.901324554674119e-05, | |
| "loss": 0.0906, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.8988220189475216e-05, | |
| "loss": 0.11, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.896289835070737e-05, | |
| "loss": 0.1056, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.893728086569276e-05, | |
| "loss": 0.0897, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.891136857943854e-05, | |
| "loss": 0.0954, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.8885162346676063e-05, | |
| "loss": 0.095, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.8858663031832665e-05, | |
| "loss": 0.0786, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.8831871509003164e-05, | |
| "loss": 0.0933, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.8804788661921012e-05, | |
| "loss": 0.0833, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.877741538392917e-05, | |
| "loss": 0.0784, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.8749752577950614e-05, | |
| "loss": 0.0851, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.8721801156458573e-05, | |
| "loss": 0.0891, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.869356204144642e-05, | |
| "loss": 0.0812, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.866503616439725e-05, | |
| "loss": 0.1109, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.8636224466253177e-05, | |
| "loss": 0.0948, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.860712789738428e-05, | |
| "loss": 0.0933, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.857774741755726e-05, | |
| "loss": 0.0872, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.854808399590378e-05, | |
| "loss": 0.0995, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.8518138610888505e-05, | |
| "loss": 0.0862, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.8487912250276805e-05, | |
| "loss": 0.0716, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.8457405911102202e-05, | |
| "loss": 0.0977, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.8426620599633464e-05, | |
| "loss": 0.0782, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.8395557331341413e-05, | |
| "loss": 0.0891, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.836421713086544e-05, | |
| "loss": 0.096, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.83326010319797e-05, | |
| "loss": 0.0912, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.830071007755901e-05, | |
| "loss": 0.0807, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.8268545319544443e-05, | |
| "loss": 0.092, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.823610781890865e-05, | |
| "loss": 0.0931, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.820339864562085e-05, | |
| "loss": 0.0859, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.817041887861153e-05, | |
| "loss": 0.089, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.8137169605736867e-05, | |
| "loss": 0.0991, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.8103651923742846e-05, | |
| "loss": 0.0741, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.8069866938229066e-05, | |
| "loss": 0.087, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.8035815763612293e-05, | |
| "loss": 0.0969, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.8001499523089683e-05, | |
| "loss": 0.071, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.7966919348601754e-05, | |
| "loss": 0.0804, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.7932076380795017e-05, | |
| "loss": 0.0895, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.7896971768984373e-05, | |
| "loss": 0.085, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.7861606671115207e-05, | |
| "loss": 0.1048, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.7825982253725175e-05, | |
| "loss": 0.0858, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.7790099691905736e-05, | |
| "loss": 0.1055, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.7753960169263387e-05, | |
| "loss": 0.0835, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.7717564877880623e-05, | |
| "loss": 0.0967, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.7680915018276613e-05, | |
| "loss": 0.0972, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.764401179936761e-05, | |
| "loss": 0.0823, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.7606856438427054e-05, | |
| "loss": 0.0803, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.7569450161045444e-05, | |
| "loss": 0.0926, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.7531794201089888e-05, | |
| "loss": 0.11, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.749388980066342e-05, | |
| "loss": 0.0889, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.745573821006403e-05, | |
| "loss": 0.093, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.7417340687743393e-05, | |
| "loss": 0.0857, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.7378698500265402e-05, | |
| "loss": 0.0871, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.7339812922264366e-05, | |
| "loss": 0.0911, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.730068523640295e-05, | |
| "loss": 0.0938, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.72613167333299e-05, | |
| "loss": 0.0883, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.7221708711637455e-05, | |
| "loss": 0.0897, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.718186247781849e-05, | |
| "loss": 0.0814, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.7141779346223465e-05, | |
| "loss": 0.0793, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.7101460639017034e-05, | |
| "loss": 0.0717, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.7060907686134445e-05, | |
| "loss": 0.0847, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.7020121825237672e-05, | |
| "loss": 0.0798, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.6979104401671296e-05, | |
| "loss": 0.0865, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.693785676841812e-05, | |
| "loss": 0.0914, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.6896380286054537e-05, | |
| "loss": 0.0797, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.6854676322705673e-05, | |
| "loss": 0.0736, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.6812746254000222e-05, | |
| "loss": 0.0765, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.67705914630251e-05, | |
| "loss": 0.0843, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.6728213340279822e-05, | |
| "loss": 0.0772, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.668561328363061e-05, | |
| "loss": 0.0944, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.6642792698264313e-05, | |
| "loss": 0.0915, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.6599752996642044e-05, | |
| "loss": 0.0805, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.655649559845258e-05, | |
| "loss": 0.0925, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.651302193056555e-05, | |
| "loss": 0.0811, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.6469333426984357e-05, | |
| "loss": 0.0787, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.6425431528798883e-05, | |
| "loss": 0.0858, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.6381317684137946e-05, | |
| "loss": 0.0768, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.6336993348121543e-05, | |
| "loss": 0.0818, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.6292459982812845e-05, | |
| "loss": 0.0868, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.624771905716997e-05, | |
| "loss": 0.0828, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.620277204699754e-05, | |
| "loss": 0.0821, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.615762043489797e-05, | |
| "loss": 0.0858, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.611226571022261e-05, | |
| "loss": 0.0711, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.6066709369022576e-05, | |
| "loss": 0.0771, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.6020952913999423e-05, | |
| "loss": 0.0788, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.5974997854455575e-05, | |
| "loss": 0.0716, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.5928845706244537e-05, | |
| "loss": 0.0894, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.588249799172089e-05, | |
| "loss": 0.085, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.583595623969009e-05, | |
| "loss": 0.0832, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.5789221985358017e-05, | |
| "loss": 0.0787, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.574229677028036e-05, | |
| "loss": 0.0792, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.5695182142311743e-05, | |
| "loss": 0.0798, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.564787965555469e-05, | |
| "loss": 0.0978, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.560039087030836e-05, | |
| "loss": 0.0909, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.5552717353017045e-05, | |
| "loss": 0.0749, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.5504860676218557e-05, | |
| "loss": 0.0788, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.5456822418492312e-05, | |
| "loss": 0.07, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.540860416440728e-05, | |
| "loss": 0.073, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.5360207504469715e-05, | |
| "loss": 0.0621, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.5311634035070678e-05, | |
| "loss": 0.0738, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.5262885358433404e-05, | |
| "loss": 0.0698, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.5213963082560424e-05, | |
| "loss": 0.0666, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.5164868821180538e-05, | |
| "loss": 0.0663, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.5115604193695599e-05, | |
| "loss": 0.0686, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.5066170825127069e-05, | |
| "loss": 0.0612, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.5016570346062432e-05, | |
| "loss": 0.0657, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.496680439260141e-05, | |
| "loss": 0.0626, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.4916874606301989e-05, | |
| "loss": 0.068, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.4866782634126266e-05, | |
| "loss": 0.0637, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.4816530128386144e-05, | |
| "loss": 0.0698, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.4766118746688805e-05, | |
| "loss": 0.0714, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.471555015188205e-05, | |
| "loss": 0.061, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.4664826011999436e-05, | |
| "loss": 0.0645, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.4613948000205272e-05, | |
| "loss": 0.0714, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.4562917794739412e-05, | |
| "loss": 0.0583, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.4511737078861903e-05, | |
| "loss": 0.0609, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1.4460407540797467e-05, | |
| "loss": 0.0695, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1.4408930873679805e-05, | |
| "loss": 0.0824, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.4357308775495757e-05, | |
| "loss": 0.0541, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.4305542949029286e-05, | |
| "loss": 0.0706, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.4253635101805313e-05, | |
| "loss": 0.0812, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.4201586946033397e-05, | |
| "loss": 0.0605, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.4149400198551247e-05, | |
| "loss": 0.0674, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.4097076580768103e-05, | |
| "loss": 0.0709, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.4044617818607949e-05, | |
| "loss": 0.0727, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.3992025642452579e-05, | |
| "loss": 0.0734, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.3939301787084522e-05, | |
| "loss": 0.0593, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.3886447991629828e-05, | |
| "loss": 0.0762, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.3833465999500689e-05, | |
| "loss": 0.0668, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.3780357558337927e-05, | |
| "loss": 0.0658, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.372712441995337e-05, | |
| "loss": 0.0823, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.3673768340272053e-05, | |
| "loss": 0.066, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.362029107927429e-05, | |
| "loss": 0.0658, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.3566694400937635e-05, | |
| "loss": 0.0618, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.3512980073178693e-05, | |
| "loss": 0.0536, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.3459149867794794e-05, | |
| "loss": 0.0676, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.3405205560405558e-05, | |
| "loss": 0.0597, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.3351148930394333e-05, | |
| "loss": 0.0618, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.329698176084948e-05, | |
| "loss": 0.0642, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.3242705838505577e-05, | |
| "loss": 0.0508, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.3188322953684467e-05, | |
| "loss": 0.0688, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.3133834900236217e-05, | |
| "loss": 0.0589, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.3079243475479942e-05, | |
| "loss": 0.0569, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.3024550480144506e-05, | |
| "loss": 0.0825, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.296975771830915e-05, | |
| "loss": 0.0652, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.2914866997343957e-05, | |
| "loss": 0.0654, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.2859880127850258e-05, | |
| "loss": 0.0629, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.2804798923600888e-05, | |
| "loss": 0.0704, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.2749625201480375e-05, | |
| "loss": 0.0688, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.2694360781424994e-05, | |
| "loss": 0.0623, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.2639007486362745e-05, | |
| "loss": 0.0948, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.2583567142153224e-05, | |
| "loss": 0.0744, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.2528041577527384e-05, | |
| "loss": 0.0526, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.2472432624027228e-05, | |
| "loss": 0.0624, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.2416742115945391e-05, | |
| "loss": 0.0732, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.2360971890264621e-05, | |
| "loss": 0.0645, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.2305123786597202e-05, | |
| "loss": 0.0678, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.224919964712427e-05, | |
| "loss": 0.0802, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.219320131653504e-05, | |
| "loss": 0.0568, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.2137130641965964e-05, | |
| "loss": 0.0526, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.20809894729398e-05, | |
| "loss": 0.0632, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.2024779661304614e-05, | |
| "loss": 0.0667, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.1968503061172674e-05, | |
| "loss": 0.0727, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.1912161528859308e-05, | |
| "loss": 0.0641, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.1855756922821675e-05, | |
| "loss": 0.0522, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.179929110359745e-05, | |
| "loss": 0.0599, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.1742765933743459e-05, | |
| "loss": 0.0681, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.168618327777425e-05, | |
| "loss": 0.0576, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.1629545002100573e-05, | |
| "loss": 0.0647, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.157285297496783e-05, | |
| "loss": 0.0646, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.1516109066394445e-05, | |
| "loss": 0.0587, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.1459315148110179e-05, | |
| "loss": 0.064, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.1402473093494395e-05, | |
| "loss": 0.0677, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.1345584777514253e-05, | |
| "loss": 0.0599, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.1288652076662878e-05, | |
| "loss": 0.0674, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.1231676868897452e-05, | |
| "loss": 0.0716, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.1174661033577267e-05, | |
| "loss": 0.0797, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.1117606451401745e-05, | |
| "loss": 0.0577, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.1060515004348394e-05, | |
| "loss": 0.0643, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.1003388575610724e-05, | |
| "loss": 0.0735, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.0946229049536136e-05, | |
| "loss": 0.0619, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.088903831156378e-05, | |
| "loss": 0.0667, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.0831818248162328e-05, | |
| "loss": 0.0721, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.0774570746767785e-05, | |
| "loss": 0.0646, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.0717297695721199e-05, | |
| "loss": 0.0604, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.0660000984206395e-05, | |
| "loss": 0.0623, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.0602682502187655e-05, | |
| "loss": 0.0648, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.0545344140347365e-05, | |
| "loss": 0.0605, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.0487987790023665e-05, | |
| "loss": 0.0667, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.0430615343148054e-05, | |
| "loss": 0.0638, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.0373228692182982e-05, | |
| "loss": 0.0598, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.031582973005943e-05, | |
| "loss": 0.0805, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.0258420350114473e-05, | |
| "loss": 0.0682, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.0201002446028815e-05, | |
| "loss": 0.059, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.0143577911764341e-05, | |
| "loss": 0.0558, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.008614864150164e-05, | |
| "loss": 0.0685, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.002871652957751e-05, | |
| "loss": 0.0539, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 9.97128347042249e-06, | |
| "loss": 0.0588, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 9.91385135849836e-06, | |
| "loss": 0.0513, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 9.85642208823566e-06, | |
| "loss": 0.0616, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 9.79899755397119e-06, | |
| "loss": 0.0588, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 9.741579649885532e-06, | |
| "loss": 0.0645, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 9.684170269940573e-06, | |
| "loss": 0.0644, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 9.62677130781702e-06, | |
| "loss": 0.0638, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 9.569384656851948e-06, | |
| "loss": 0.085, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 9.512012209976335e-06, | |
| "loss": 0.0567, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.454655859652637e-06, | |
| "loss": 0.0639, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.39731749781235e-06, | |
| "loss": 0.073, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.339999015793606e-06, | |
| "loss": 0.0703, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.282702304278806e-06, | |
| "loss": 0.0711, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.225429253232218e-06, | |
| "loss": 0.0641, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.168181751837673e-06, | |
| "loss": 0.0591, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 9.110961688436222e-06, | |
| "loss": 0.0684, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 9.053770950463865e-06, | |
| "loss": 0.0775, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 8.996611424389283e-06, | |
| "loss": 0.0688, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 8.93948499565161e-06, | |
| "loss": 0.0785, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 8.882393548598258e-06, | |
| "loss": 0.0737, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 8.825338966422735e-06, | |
| "loss": 0.062, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.768323131102552e-06, | |
| "loss": 0.0685, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.711347923337122e-06, | |
| "loss": 0.0618, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 8.65441522248575e-06, | |
| "loss": 0.0728, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 8.59752690650561e-06, | |
| "loss": 0.0562, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 8.540684851889823e-06, | |
| "loss": 0.0772, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 8.483890933605558e-06, | |
| "loss": 0.0711, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 8.427147025032171e-06, | |
| "loss": 0.0579, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.37045499789943e-06, | |
| "loss": 0.0607, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.313816722225751e-06, | |
| "loss": 0.0692, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 8.257234066256543e-06, | |
| "loss": 0.0632, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 8.200708896402557e-06, | |
| "loss": 0.0706, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 8.144243077178329e-06, | |
| "loss": 0.0612, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 8.087838471140696e-06, | |
| "loss": 0.0684, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 8.031496938827329e-06, | |
| "loss": 0.0712, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 7.97522033869539e-06, | |
| "loss": 0.0676, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 7.9190105270602e-06, | |
| "loss": 0.0643, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 7.86286935803404e-06, | |
| "loss": 0.0602, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 7.806798683464965e-06, | |
| "loss": 0.0594, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 7.750800352875734e-06, | |
| "loss": 0.0677, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 7.694876213402801e-06, | |
| "loss": 0.0684, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 7.63902810973538e-06, | |
| "loss": 0.0599, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 7.583257884054613e-06, | |
| "loss": 0.0711, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 7.527567375972772e-06, | |
| "loss": 0.0724, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 7.471958422472618e-06, | |
| "loss": 0.0579, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 7.416432857846783e-06, | |
| "loss": 0.0544, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 7.360992513637257e-06, | |
| "loss": 0.0559, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 7.305639218575009e-06, | |
| "loss": 0.0681, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 7.250374798519626e-06, | |
| "loss": 0.0668, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 7.1952010763991146e-06, | |
| "loss": 0.0787, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 7.140119872149743e-06, | |
| "loss": 0.0739, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 7.085133002656044e-06, | |
| "loss": 0.0616, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 7.030242281690856e-06, | |
| "loss": 0.052, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 6.975449519855495e-06, | |
| "loss": 0.0676, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 6.9207565245200614e-06, | |
| "loss": 0.0576, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 6.866165099763782e-06, | |
| "loss": 0.0547, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 6.811677046315535e-06, | |
| "loss": 0.0615, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 6.757294161494426e-06, | |
| "loss": 0.0533, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 6.70301823915052e-06, | |
| "loss": 0.0654, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 6.64885106960567e-06, | |
| "loss": 0.0616, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 6.594794439594443e-06, | |
| "loss": 0.0602, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 6.54085013220521e-06, | |
| "loss": 0.0565, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 6.48701992682131e-06, | |
| "loss": 0.0481, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 6.4333055990623674e-06, | |
| "loss": 0.0583, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 6.379708920725713e-06, | |
| "loss": 0.0598, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 6.3262316597279506e-06, | |
| "loss": 0.0565, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 6.272875580046633e-06, | |
| "loss": 0.0684, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 6.219642441662077e-06, | |
| "loss": 0.0593, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 6.1665340004993164e-06, | |
| "loss": 0.0567, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 6.113552008370172e-06, | |
| "loss": 0.064, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 6.06069821291548e-06, | |
| "loss": 0.0613, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 6.007974357547424e-06, | |
| "loss": 0.0671, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 5.9553821813920545e-06, | |
| "loss": 0.0572, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 5.902923419231902e-06, | |
| "loss": 0.0432, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 5.850599801448757e-06, | |
| "loss": 0.0624, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 5.798413053966607e-06, | |
| "loss": 0.0717, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 5.74636489819469e-06, | |
| "loss": 0.064, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 5.6944570509707185e-06, | |
| "loss": 0.0546, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 5.6426912245042435e-06, | |
| "loss": 0.0653, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 5.5910691263201985e-06, | |
| "loss": 0.0492, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 5.5395924592025384e-06, | |
| "loss": 0.0595, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 5.488262921138098e-06, | |
| "loss": 0.0607, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 5.437082205260593e-06, | |
| "loss": 0.0571, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 5.3860519997947295e-06, | |
| "loss": 0.0818, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 5.335173988000566e-06, | |
| "loss": 0.0599, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 5.284449848117954e-06, | |
| "loss": 0.052, | |
| "step": 376 | |
| } | |
| ], | |
| "max_steps": 564, | |
| "num_train_epochs": 3, | |
| "total_flos": 20423908786176.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |