| { | |
| "best_metric": 0.8618489515327258, | |
| "best_model_checkpoint": "result/bert-base-uncased-cls_before_pooler-sym_mlp-mlp_bert-bs64-gpu8-gs1-lr5e-5-m=stsb-norm0.05-l32-contra", | |
| "epoch": 3.0, | |
| "global_step": 1617, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.969078540507112e-05, | |
| "loss": 3.0781, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.938157081014224e-05, | |
| "loss": 1.5252, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.907235621521336e-05, | |
| "loss": 1.3452, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.876314162028448e-05, | |
| "loss": 1.1968, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.84539270253556e-05, | |
| "loss": 1.1594, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.814471243042672e-05, | |
| "loss": 1.1094, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.783549783549784e-05, | |
| "loss": 1.0483, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.752628324056896e-05, | |
| "loss": 1.0223, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.721706864564008e-05, | |
| "loss": 1.0104, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.69078540507112e-05, | |
| "loss": 0.9423, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.6598639455782315e-05, | |
| "loss": 0.995, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.628942486085344e-05, | |
| "loss": 0.9824, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "eval_avg_sts": 0.8261830768165319, | |
| "eval_sickr_spearman": 0.8082406697081602, | |
| "eval_stsb_spearman": 0.8441254839249035, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.5980210265924555e-05, | |
| "loss": 0.9722, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.567099567099568e-05, | |
| "loss": 0.9475, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5361781076066796e-05, | |
| "loss": 0.9085, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.505256648113791e-05, | |
| "loss": 0.9555, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.4743351886209036e-05, | |
| "loss": 0.9275, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.4434137291280146e-05, | |
| "loss": 0.9116, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.412492269635127e-05, | |
| "loss": 0.8763, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.381570810142239e-05, | |
| "loss": 0.8723, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.3506493506493503e-05, | |
| "loss": 0.893, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.319727891156463e-05, | |
| "loss": 0.8916, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.2888064316635744e-05, | |
| "loss": 0.9036, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.257884972170687e-05, | |
| "loss": 0.9093, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.2269635126777984e-05, | |
| "loss": 0.8959, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "eval_avg_sts": 0.8354195988673837, | |
| "eval_sickr_spearman": 0.8198259154683373, | |
| "eval_stsb_spearman": 0.8510132822664301, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.19604205318491e-05, | |
| "loss": 0.8563, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.1651205936920225e-05, | |
| "loss": 0.8488, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.134199134199134e-05, | |
| "loss": 0.852, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.1032776747062465e-05, | |
| "loss": 0.84, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.072356215213358e-05, | |
| "loss": 0.8553, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.04143475572047e-05, | |
| "loss": 0.8611, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.010513296227582e-05, | |
| "loss": 0.8427, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.979591836734694e-05, | |
| "loss": 0.8314, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.948670377241806e-05, | |
| "loss": 0.8604, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.917748917748918e-05, | |
| "loss": 0.8528, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.88682745825603e-05, | |
| "loss": 0.8808, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.855905998763142e-05, | |
| "loss": 0.8556, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "eval_avg_sts": 0.8415134058271878, | |
| "eval_sickr_spearman": 0.8248841668229878, | |
| "eval_stsb_spearman": 0.8581426448313877, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.824984539270254e-05, | |
| "loss": 0.8503, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.794063079777366e-05, | |
| "loss": 0.807, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.763141620284478e-05, | |
| "loss": 0.8066, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.7322201607915894e-05, | |
| "loss": 0.8689, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.701298701298702e-05, | |
| "loss": 0.8117, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.6703772418058135e-05, | |
| "loss": 0.8311, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.639455782312925e-05, | |
| "loss": 0.8118, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.6085343228200375e-05, | |
| "loss": 0.7894, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.577612863327149e-05, | |
| "loss": 0.7979, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.5466914038342616e-05, | |
| "loss": 0.8033, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.515769944341373e-05, | |
| "loss": 0.7821, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.484848484848485e-05, | |
| "loss": 0.7713, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.453927025355597e-05, | |
| "loss": 0.7738, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "eval_avg_sts": 0.8368228659864433, | |
| "eval_sickr_spearman": 0.8144479691025146, | |
| "eval_stsb_spearman": 0.859197762870372, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.423005565862709e-05, | |
| "loss": 0.808, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.392084106369821e-05, | |
| "loss": 0.7669, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.3611626468769324e-05, | |
| "loss": 0.7892, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.330241187384045e-05, | |
| "loss": 0.7481, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.2993197278911564e-05, | |
| "loss": 0.6594, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.268398268398268e-05, | |
| "loss": 0.6741, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.2374768089053805e-05, | |
| "loss": 0.659, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.206555349412492e-05, | |
| "loss": 0.6727, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.1756338899196045e-05, | |
| "loss": 0.6712, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.144712430426716e-05, | |
| "loss": 0.6552, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.113790970933828e-05, | |
| "loss": 0.6436, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.08286951144094e-05, | |
| "loss": 0.6747, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "eval_avg_sts": 0.8398383909792182, | |
| "eval_sickr_spearman": 0.8222943778645462, | |
| "eval_stsb_spearman": 0.8573824040938901, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.051948051948052e-05, | |
| "loss": 0.6963, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.021026592455164e-05, | |
| "loss": 0.65, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.990105132962276e-05, | |
| "loss": 0.6749, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.959183673469388e-05, | |
| "loss": 0.6695, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.9282622139764997e-05, | |
| "loss": 0.6984, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.8973407544836117e-05, | |
| "loss": 0.656, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.8664192949907237e-05, | |
| "loss": 0.693, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.8354978354978357e-05, | |
| "loss": 0.669, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.8045763760049478e-05, | |
| "loss": 0.6992, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.7736549165120594e-05, | |
| "loss": 0.6667, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.7427334570191715e-05, | |
| "loss": 0.6588, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.7118119975262835e-05, | |
| "loss": 0.6505, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.6808905380333955e-05, | |
| "loss": 0.6546, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "eval_avg_sts": 0.8437054666296041, | |
| "eval_sickr_spearman": 0.8255619817264824, | |
| "eval_stsb_spearman": 0.8618489515327258, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.6499690785405072e-05, | |
| "loss": 0.645, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.6190476190476192e-05, | |
| "loss": 0.6576, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.5881261595547312e-05, | |
| "loss": 0.6593, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.5572047000618433e-05, | |
| "loss": 0.6423, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.5262832405689553e-05, | |
| "loss": 0.6725, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.495361781076067e-05, | |
| "loss": 0.6694, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.4644403215831787e-05, | |
| "loss": 0.6694, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.4335188620902907e-05, | |
| "loss": 0.6693, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.4025974025974027e-05, | |
| "loss": 0.6596, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.3716759431045147e-05, | |
| "loss": 0.6869, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.3407544836116267e-05, | |
| "loss": 0.6958, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.3098330241187384e-05, | |
| "loss": 0.6389, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "eval_avg_sts": 0.8406547485620587, | |
| "eval_sickr_spearman": 0.8221673836323439, | |
| "eval_stsb_spearman": 0.8591421134917735, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.2789115646258505e-05, | |
| "loss": 0.6295, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.2479901051329625e-05, | |
| "loss": 0.6759, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.2170686456400745e-05, | |
| "loss": 0.635, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.1861471861471862e-05, | |
| "loss": 0.6361, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.1552257266542982e-05, | |
| "loss": 0.6723, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.1243042671614102e-05, | |
| "loss": 0.6441, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.0933828076685223e-05, | |
| "loss": 0.6509, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.062461348175634e-05, | |
| "loss": 0.6572, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.031539888682746e-05, | |
| "loss": 0.6662, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.0006184291898576e-05, | |
| "loss": 0.6425, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.9696969696969697e-05, | |
| "loss": 0.6641, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.9387755102040817e-05, | |
| "loss": 0.6673, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.9078540507111937e-05, | |
| "loss": 0.6863, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "eval_avg_sts": 0.8431467949723008, | |
| "eval_sickr_spearman": 0.8253605873181524, | |
| "eval_stsb_spearman": 0.8609330026264492, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.8769325912183054e-05, | |
| "loss": 0.6352, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.8460111317254174e-05, | |
| "loss": 0.6226, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.8150896722325294e-05, | |
| "loss": 0.627, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.7841682127396415e-05, | |
| "loss": 0.6481, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.7532467532467535e-05, | |
| "loss": 0.6553, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.7223252937538652e-05, | |
| "loss": 0.6345, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.6914038342609772e-05, | |
| "loss": 0.6328, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.6604823747680892e-05, | |
| "loss": 0.6054, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.6295609152752012e-05, | |
| "loss": 0.5763, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.5986394557823133e-05, | |
| "loss": 0.5829, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.567717996289425e-05, | |
| "loss": 0.544, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.5367965367965366e-05, | |
| "loss": 0.5845, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "eval_avg_sts": 0.8400228852384868, | |
| "eval_sickr_spearman": 0.8210134364202931, | |
| "eval_stsb_spearman": 0.8590323340566804, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.5058750773036487e-05, | |
| "loss": 0.5509, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4749536178107607e-05, | |
| "loss": 0.5749, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.4440321583178725e-05, | |
| "loss": 0.5924, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.4131106988249846e-05, | |
| "loss": 0.5857, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.3821892393320964e-05, | |
| "loss": 0.5721, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.3512677798392084e-05, | |
| "loss": 0.5762, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.3203463203463205e-05, | |
| "loss": 0.5654, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.2894248608534323e-05, | |
| "loss": 0.5878, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.2585034013605443e-05, | |
| "loss": 0.5924, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.2275819418676562e-05, | |
| "loss": 0.5856, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.1966604823747682e-05, | |
| "loss": 0.573, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.16573902288188e-05, | |
| "loss": 0.5702, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.1348175633889919e-05, | |
| "loss": 0.5718, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "eval_avg_sts": 0.8425666412908246, | |
| "eval_sickr_spearman": 0.8245988620804454, | |
| "eval_stsb_spearman": 0.8605344205012039, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.103896103896104e-05, | |
| "loss": 0.5801, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.0729746444032158e-05, | |
| "loss": 0.5627, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.0420531849103278e-05, | |
| "loss": 0.5659, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.0111317254174398e-05, | |
| "loss": 0.5404, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.802102659245517e-06, | |
| "loss": 0.5901, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.492888064316637e-06, | |
| "loss": 0.5645, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.183673469387756e-06, | |
| "loss": 0.5713, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 8.874458874458876e-06, | |
| "loss": 0.558, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.565244279529994e-06, | |
| "loss": 0.577, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 8.256029684601113e-06, | |
| "loss": 0.5706, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 7.946815089672233e-06, | |
| "loss": 0.5888, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.637600494743352e-06, | |
| "loss": 0.5671, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "eval_avg_sts": 0.8414059990046279, | |
| "eval_sickr_spearman": 0.8225058588041783, | |
| "eval_stsb_spearman": 0.8603061392050774, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.328385899814472e-06, | |
| "loss": 0.5655, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 7.019171304885591e-06, | |
| "loss": 0.5677, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.709956709956711e-06, | |
| "loss": 0.5451, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 6.40074211502783e-06, | |
| "loss": 0.5552, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.091527520098949e-06, | |
| "loss": 0.5515, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.782312925170069e-06, | |
| "loss": 0.5635, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.473098330241188e-06, | |
| "loss": 0.5935, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.163883735312307e-06, | |
| "loss": 0.5736, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.854669140383426e-06, | |
| "loss": 0.5772, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.5454545454545455e-06, | |
| "loss": 0.5969, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.236239950525665e-06, | |
| "loss": 0.5654, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 3.927025355596784e-06, | |
| "loss": 0.5565, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 3.6178107606679037e-06, | |
| "loss": 0.5586, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "eval_avg_sts": 0.8421757762418902, | |
| "eval_sickr_spearman": 0.8238130732609178, | |
| "eval_stsb_spearman": 0.8605384792228625, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.308596165739023e-06, | |
| "loss": 0.5726, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 2.9993815708101424e-06, | |
| "loss": 0.5556, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.690166975881262e-06, | |
| "loss": 0.58, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.3809523809523808e-06, | |
| "loss": 0.5682, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 2.0717377860235006e-06, | |
| "loss": 0.5636, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.7625231910946195e-06, | |
| "loss": 0.5494, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.4533085961657391e-06, | |
| "loss": 0.57, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.1440940012368585e-06, | |
| "loss": 0.5566, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 8.348794063079778e-07, | |
| "loss": 0.5617, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 5.25664811379097e-07, | |
| "loss": 0.5632, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 2.1645021645021646e-07, | |
| "loss": 0.5804, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 1617, | |
| "train_runtime": 1176.2097, | |
| "train_samples_per_second": 1.375 | |
| } | |
| ], | |
| "max_steps": 1617, | |
| "num_train_epochs": 3, | |
| "total_flos": 52422141990721536, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |