{ "best_metric": 0.8233758432892795, "best_model_checkpoint": "result/bert-large-uncased-cls_before_pooler-sym_mlp-mlp_bert-bs64-gpu8-gs1-lr5e-5-m=stsb-norm0.05-l32-wiki-id-1epoch", "epoch": 1.0, "global_step": 1954, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 4.9744114636642786e-05, "loss": 0.7987, "step": 10 }, { "epoch": 0.01, "learning_rate": 4.948822927328557e-05, "loss": 0.0132, "step": 20 }, { "epoch": 0.02, "learning_rate": 4.923234390992835e-05, "loss": 0.0083, "step": 30 }, { "epoch": 0.02, "learning_rate": 4.897645854657114e-05, "loss": 0.0044, "step": 40 }, { "epoch": 0.03, "learning_rate": 4.8720573183213926e-05, "loss": 0.0041, "step": 50 }, { "epoch": 0.03, "learning_rate": 4.846468781985671e-05, "loss": 0.0032, "step": 60 }, { "epoch": 0.04, "learning_rate": 4.820880245649949e-05, "loss": 0.0022, "step": 70 }, { "epoch": 0.04, "learning_rate": 4.795291709314228e-05, "loss": 0.0014, "step": 80 }, { "epoch": 0.05, "learning_rate": 4.769703172978506e-05, "loss": 0.0023, "step": 90 }, { "epoch": 0.05, "learning_rate": 4.7441146366427843e-05, "loss": 0.0012, "step": 100 }, { "epoch": 0.06, "learning_rate": 4.718526100307063e-05, "loss": 0.0013, "step": 110 }, { "epoch": 0.06, "learning_rate": 4.692937563971341e-05, "loss": 0.0017, "step": 120 }, { "epoch": 0.06, "eval_avg_sts": 0.7527941583986333, "eval_sickr_spearman": 0.7351572662264496, "eval_stsb_spearman": 0.770431050570817, "step": 125 }, { "epoch": 0.07, "learning_rate": 4.6673490276356194e-05, "loss": 0.0021, "step": 130 }, { "epoch": 0.07, "learning_rate": 4.641760491299898e-05, "loss": 0.0012, "step": 140 }, { "epoch": 0.08, "learning_rate": 4.616171954964176e-05, "loss": 0.0017, "step": 150 }, { "epoch": 0.08, "learning_rate": 4.5905834186284544e-05, "loss": 0.0011, "step": 160 }, { "epoch": 0.09, "learning_rate": 4.564994882292733e-05, "loss": 0.0004, "step": 170 }, { "epoch": 0.09, "learning_rate": 4.539406345957011e-05, "loss": 0.001, "step": 180 }, { "epoch": 0.1, "learning_rate": 4.5138178096212894e-05, "loss": 0.0013, "step": 190 }, { "epoch": 0.1, "learning_rate": 4.4882292732855685e-05, "loss": 0.0011, "step": 200 }, { "epoch": 0.11, "learning_rate": 4.462640736949847e-05, "loss": 0.0016, "step": 210 }, { "epoch": 0.11, "learning_rate": 4.437052200614125e-05, "loss": 0.0008, "step": 220 }, { "epoch": 0.12, "learning_rate": 4.4114636642784035e-05, "loss": 0.0006, "step": 230 }, { "epoch": 0.12, "learning_rate": 4.385875127942682e-05, "loss": 0.0012, "step": 240 }, { "epoch": 0.13, "learning_rate": 4.36028659160696e-05, "loss": 0.0012, "step": 250 }, { "epoch": 0.13, "eval_avg_sts": 0.7786723498526811, "eval_sickr_spearman": 0.7543037120683187, "eval_stsb_spearman": 0.8030409876370435, "step": 250 }, { "epoch": 0.13, "learning_rate": 4.3346980552712385e-05, "loss": 0.001, "step": 260 }, { "epoch": 0.14, "learning_rate": 4.309109518935517e-05, "loss": 0.0007, "step": 270 }, { "epoch": 0.14, "learning_rate": 4.283520982599796e-05, "loss": 0.0009, "step": 280 }, { "epoch": 0.15, "learning_rate": 4.257932446264074e-05, "loss": 0.0012, "step": 290 }, { "epoch": 0.15, "learning_rate": 4.2323439099283526e-05, "loss": 0.0015, "step": 300 }, { "epoch": 0.16, "learning_rate": 4.206755373592631e-05, "loss": 0.0013, "step": 310 }, { "epoch": 0.16, "learning_rate": 4.181166837256909e-05, "loss": 0.0019, "step": 320 }, { "epoch": 0.17, "learning_rate": 4.1555783009211876e-05, "loss": 0.0013, "step": 330 }, { "epoch": 0.17, "learning_rate": 4.129989764585466e-05, "loss": 0.0007, "step": 340 }, { "epoch": 0.18, "learning_rate": 4.104401228249744e-05, "loss": 0.0004, "step": 350 }, { "epoch": 0.18, "learning_rate": 4.0788126919140226e-05, "loss": 0.0015, "step": 360 }, { "epoch": 0.19, "learning_rate": 4.053224155578301e-05, "loss": 0.0004, "step": 370 }, { "epoch": 0.19, "eval_avg_sts": 0.7913077528900125, "eval_sickr_spearman": 0.7655320387137092, "eval_stsb_spearman": 0.8170834670663158, "step": 375 }, { "epoch": 0.19, "learning_rate": 4.027635619242579e-05, "loss": 0.0014, "step": 380 }, { "epoch": 0.2, "learning_rate": 4.0020470829068577e-05, "loss": 0.0013, "step": 390 }, { "epoch": 0.2, "learning_rate": 3.976458546571136e-05, "loss": 0.0015, "step": 400 }, { "epoch": 0.21, "learning_rate": 3.9508700102354143e-05, "loss": 0.0006, "step": 410 }, { "epoch": 0.21, "learning_rate": 3.925281473899693e-05, "loss": 0.0005, "step": 420 }, { "epoch": 0.22, "learning_rate": 3.899692937563972e-05, "loss": 0.0006, "step": 430 }, { "epoch": 0.23, "learning_rate": 3.87410440122825e-05, "loss": 0.0004, "step": 440 }, { "epoch": 0.23, "learning_rate": 3.8485158648925284e-05, "loss": 0.001, "step": 450 }, { "epoch": 0.24, "learning_rate": 3.822927328556807e-05, "loss": 0.0012, "step": 460 }, { "epoch": 0.24, "learning_rate": 3.797338792221085e-05, "loss": 0.0014, "step": 470 }, { "epoch": 0.25, "learning_rate": 3.7717502558853634e-05, "loss": 0.0007, "step": 480 }, { "epoch": 0.25, "learning_rate": 3.746161719549642e-05, "loss": 0.0011, "step": 490 }, { "epoch": 0.26, "learning_rate": 3.72057318321392e-05, "loss": 0.0011, "step": 500 }, { "epoch": 0.26, "eval_avg_sts": 0.7880331799657297, "eval_sickr_spearman": 0.7571696318289279, "eval_stsb_spearman": 0.8188967281025316, "step": 500 }, { "epoch": 0.26, "learning_rate": 3.694984646878199e-05, "loss": 0.0003, "step": 510 }, { "epoch": 0.27, "learning_rate": 3.6693961105424775e-05, "loss": 0.0007, "step": 520 }, { "epoch": 0.27, "learning_rate": 3.643807574206756e-05, "loss": 0.0007, "step": 530 }, { "epoch": 0.28, "learning_rate": 3.618219037871034e-05, "loss": 0.0021, "step": 540 }, { "epoch": 0.28, "learning_rate": 3.5926305015353125e-05, "loss": 0.0019, "step": 550 }, { "epoch": 0.29, "learning_rate": 3.567041965199591e-05, "loss": 0.0015, "step": 560 }, { "epoch": 0.29, "learning_rate": 3.541453428863869e-05, "loss": 0.0002, "step": 570 }, { "epoch": 0.3, "learning_rate": 3.5158648925281475e-05, "loss": 0.0017, "step": 580 }, { "epoch": 0.3, "learning_rate": 3.4902763561924265e-05, "loss": 0.0018, "step": 590 }, { "epoch": 0.31, "learning_rate": 3.464687819856704e-05, "loss": 0.0013, "step": 600 }, { "epoch": 0.31, "learning_rate": 3.4390992835209826e-05, "loss": 0.0008, "step": 610 }, { "epoch": 0.32, "learning_rate": 3.413510747185261e-05, "loss": 0.0022, "step": 620 }, { "epoch": 0.32, "eval_avg_sts": 0.7732885189334773, "eval_sickr_spearman": 0.7409421801147573, "eval_stsb_spearman": 0.8056348577521973, "step": 625 }, { "epoch": 0.32, "learning_rate": 3.387922210849539e-05, "loss": 0.0022, "step": 630 }, { "epoch": 0.33, "learning_rate": 3.3623336745138176e-05, "loss": 0.0006, "step": 640 }, { "epoch": 0.33, "learning_rate": 3.336745138178096e-05, "loss": 0.0011, "step": 650 }, { "epoch": 0.34, "learning_rate": 3.311156601842374e-05, "loss": 0.0097, "step": 660 }, { "epoch": 0.34, "learning_rate": 3.285568065506653e-05, "loss": 0.0065, "step": 670 }, { "epoch": 0.35, "learning_rate": 3.2599795291709316e-05, "loss": 0.0026, "step": 680 }, { "epoch": 0.35, "learning_rate": 3.23439099283521e-05, "loss": 0.002, "step": 690 }, { "epoch": 0.36, "learning_rate": 3.208802456499488e-05, "loss": 0.0029, "step": 700 }, { "epoch": 0.36, "learning_rate": 3.183213920163767e-05, "loss": 0.0025, "step": 710 }, { "epoch": 0.37, "learning_rate": 3.157625383828045e-05, "loss": 0.0024, "step": 720 }, { "epoch": 0.37, "learning_rate": 3.1320368474923234e-05, "loss": 0.001, "step": 730 }, { "epoch": 0.38, "learning_rate": 3.106448311156602e-05, "loss": 0.0016, "step": 740 }, { "epoch": 0.38, "learning_rate": 3.080859774820881e-05, "loss": 0.001, "step": 750 }, { "epoch": 0.38, "eval_avg_sts": 0.7885603958950645, "eval_sickr_spearman": 0.7648866928347867, "eval_stsb_spearman": 0.8122340989553424, "step": 750 }, { "epoch": 0.39, "learning_rate": 3.055271238485159e-05, "loss": 0.0016, "step": 760 }, { "epoch": 0.39, "learning_rate": 3.0296827021494374e-05, "loss": 0.0007, "step": 770 }, { "epoch": 0.4, "learning_rate": 3.0040941658137157e-05, "loss": 0.0015, "step": 780 }, { "epoch": 0.4, "learning_rate": 2.978505629477994e-05, "loss": 0.0017, "step": 790 }, { "epoch": 0.41, "learning_rate": 2.9529170931422724e-05, "loss": 0.001, "step": 800 }, { "epoch": 0.41, "learning_rate": 2.9273285568065504e-05, "loss": 0.0005, "step": 810 }, { "epoch": 0.42, "learning_rate": 2.9017400204708288e-05, "loss": 0.001, "step": 820 }, { "epoch": 0.42, "learning_rate": 2.8761514841351078e-05, "loss": 0.0004, "step": 830 }, { "epoch": 0.43, "learning_rate": 2.850562947799386e-05, "loss": 0.0008, "step": 840 }, { "epoch": 0.44, "learning_rate": 2.8249744114636645e-05, "loss": 0.0008, "step": 850 }, { "epoch": 0.44, "learning_rate": 2.7993858751279428e-05, "loss": 0.0018, "step": 860 }, { "epoch": 0.45, "learning_rate": 2.7737973387922212e-05, "loss": 0.0004, "step": 870 }, { "epoch": 0.45, "eval_avg_sts": 0.7926162200321307, "eval_sickr_spearman": 0.7705847664916943, "eval_stsb_spearman": 0.8146476735725671, "step": 875 }, { "epoch": 0.45, "learning_rate": 2.7482088024564995e-05, "loss": 0.0009, "step": 880 }, { "epoch": 0.46, "learning_rate": 2.722620266120778e-05, "loss": 0.0008, "step": 890 }, { "epoch": 0.46, "learning_rate": 2.6970317297850565e-05, "loss": 0.0011, "step": 900 }, { "epoch": 0.47, "learning_rate": 2.671443193449335e-05, "loss": 0.0009, "step": 910 }, { "epoch": 0.47, "learning_rate": 2.6458546571136132e-05, "loss": 0.0012, "step": 920 }, { "epoch": 0.48, "learning_rate": 2.6202661207778916e-05, "loss": 0.0006, "step": 930 }, { "epoch": 0.48, "learning_rate": 2.59467758444217e-05, "loss": 0.0006, "step": 940 }, { "epoch": 0.49, "learning_rate": 2.5690890481064483e-05, "loss": 0.0019, "step": 950 }, { "epoch": 0.49, "learning_rate": 2.5435005117707266e-05, "loss": 0.0003, "step": 960 }, { "epoch": 0.5, "learning_rate": 2.517911975435005e-05, "loss": 0.0015, "step": 970 }, { "epoch": 0.5, "learning_rate": 2.4923234390992836e-05, "loss": 0.0007, "step": 980 }, { "epoch": 0.51, "learning_rate": 2.466734902763562e-05, "loss": 0.0009, "step": 990 }, { "epoch": 0.51, "learning_rate": 2.4411463664278406e-05, "loss": 0.0007, "step": 1000 }, { "epoch": 0.51, "eval_avg_sts": 0.7959315624689012, "eval_sickr_spearman": 0.774320769654855, "eval_stsb_spearman": 0.8175423552829473, "step": 1000 }, { "epoch": 0.52, "learning_rate": 2.415557830092119e-05, "loss": 0.0008, "step": 1010 }, { "epoch": 0.52, "learning_rate": 2.3899692937563973e-05, "loss": 0.0007, "step": 1020 }, { "epoch": 0.53, "learning_rate": 2.3643807574206757e-05, "loss": 0.0008, "step": 1030 }, { "epoch": 0.53, "learning_rate": 2.338792221084954e-05, "loss": 0.0017, "step": 1040 }, { "epoch": 0.54, "learning_rate": 2.3132036847492324e-05, "loss": 0.0008, "step": 1050 }, { "epoch": 0.54, "learning_rate": 2.2876151484135107e-05, "loss": 0.0013, "step": 1060 }, { "epoch": 0.55, "learning_rate": 2.262026612077789e-05, "loss": 0.0003, "step": 1070 }, { "epoch": 0.55, "learning_rate": 2.2364380757420677e-05, "loss": 0.0007, "step": 1080 }, { "epoch": 0.56, "learning_rate": 2.210849539406346e-05, "loss": 0.0014, "step": 1090 }, { "epoch": 0.56, "learning_rate": 2.1852610030706244e-05, "loss": 0.001, "step": 1100 }, { "epoch": 0.57, "learning_rate": 2.1596724667349028e-05, "loss": 0.0005, "step": 1110 }, { "epoch": 0.57, "learning_rate": 2.1340839303991814e-05, "loss": 0.0005, "step": 1120 }, { "epoch": 0.58, "eval_avg_sts": 0.795896362310046, "eval_sickr_spearman": 0.7711842426687433, "eval_stsb_spearman": 0.8206084819513487, "step": 1125 }, { "epoch": 0.58, "learning_rate": 2.1084953940634598e-05, "loss": 0.0009, "step": 1130 }, { "epoch": 0.58, "learning_rate": 2.082906857727738e-05, "loss": 0.0015, "step": 1140 }, { "epoch": 0.59, "learning_rate": 2.0573183213920165e-05, "loss": 0.0009, "step": 1150 }, { "epoch": 0.59, "learning_rate": 2.0317297850562948e-05, "loss": 0.0006, "step": 1160 }, { "epoch": 0.6, "learning_rate": 2.006141248720573e-05, "loss": 0.0004, "step": 1170 }, { "epoch": 0.6, "learning_rate": 1.9805527123848515e-05, "loss": 0.0008, "step": 1180 }, { "epoch": 0.61, "learning_rate": 1.95496417604913e-05, "loss": 0.0005, "step": 1190 }, { "epoch": 0.61, "learning_rate": 1.9293756397134085e-05, "loss": 0.0014, "step": 1200 }, { "epoch": 0.62, "learning_rate": 1.903787103377687e-05, "loss": 0.0008, "step": 1210 }, { "epoch": 0.62, "learning_rate": 1.8781985670419652e-05, "loss": 0.0014, "step": 1220 }, { "epoch": 0.63, "learning_rate": 1.8526100307062436e-05, "loss": 0.0006, "step": 1230 }, { "epoch": 0.63, "learning_rate": 1.8270214943705222e-05, "loss": 0.0009, "step": 1240 }, { "epoch": 0.64, "learning_rate": 1.8014329580348006e-05, "loss": 0.0006, "step": 1250 }, { "epoch": 0.64, "eval_avg_sts": 0.7963228217070192, "eval_sickr_spearman": 0.772572149375899, "eval_stsb_spearman": 0.8200734940381393, "step": 1250 }, { "epoch": 0.64, "learning_rate": 1.775844421699079e-05, "loss": 0.0009, "step": 1260 }, { "epoch": 0.65, "learning_rate": 1.7502558853633573e-05, "loss": 0.0002, "step": 1270 }, { "epoch": 0.66, "learning_rate": 1.7246673490276356e-05, "loss": 0.0007, "step": 1280 }, { "epoch": 0.66, "learning_rate": 1.699078812691914e-05, "loss": 0.0007, "step": 1290 }, { "epoch": 0.67, "learning_rate": 1.6734902763561923e-05, "loss": 0.0014, "step": 1300 }, { "epoch": 0.67, "learning_rate": 1.647901740020471e-05, "loss": 0.0011, "step": 1310 }, { "epoch": 0.68, "learning_rate": 1.6223132036847493e-05, "loss": 0.0011, "step": 1320 }, { "epoch": 0.68, "learning_rate": 1.5967246673490277e-05, "loss": 0.0008, "step": 1330 }, { "epoch": 0.69, "learning_rate": 1.571136131013306e-05, "loss": 0.001, "step": 1340 }, { "epoch": 0.69, "learning_rate": 1.5455475946775847e-05, "loss": 0.0017, "step": 1350 }, { "epoch": 0.7, "learning_rate": 1.519959058341863e-05, "loss": 0.0004, "step": 1360 }, { "epoch": 0.7, "learning_rate": 1.4943705220061414e-05, "loss": 0.0009, "step": 1370 }, { "epoch": 0.7, "eval_avg_sts": 0.794838346221856, "eval_sickr_spearman": 0.7716415467855355, "eval_stsb_spearman": 0.8180351456581764, "step": 1375 }, { "epoch": 0.71, "learning_rate": 1.4687819856704197e-05, "loss": 0.0007, "step": 1380 }, { "epoch": 0.71, "learning_rate": 1.4431934493346982e-05, "loss": 0.0014, "step": 1390 }, { "epoch": 0.72, "learning_rate": 1.4176049129989766e-05, "loss": 0.0007, "step": 1400 }, { "epoch": 0.72, "learning_rate": 1.392016376663255e-05, "loss": 0.0007, "step": 1410 }, { "epoch": 0.73, "learning_rate": 1.3664278403275333e-05, "loss": 0.0009, "step": 1420 }, { "epoch": 0.73, "learning_rate": 1.3408393039918118e-05, "loss": 0.0002, "step": 1430 }, { "epoch": 0.74, "learning_rate": 1.3152507676560901e-05, "loss": 0.0024, "step": 1440 }, { "epoch": 0.74, "learning_rate": 1.2896622313203685e-05, "loss": 0.0005, "step": 1450 }, { "epoch": 0.75, "learning_rate": 1.2640736949846468e-05, "loss": 0.0016, "step": 1460 }, { "epoch": 0.75, "learning_rate": 1.2384851586489253e-05, "loss": 0.0007, "step": 1470 }, { "epoch": 0.76, "learning_rate": 1.2128966223132038e-05, "loss": 0.0002, "step": 1480 }, { "epoch": 0.76, "learning_rate": 1.1873080859774822e-05, "loss": 0.0004, "step": 1490 }, { "epoch": 0.77, "learning_rate": 1.1617195496417605e-05, "loss": 0.0005, "step": 1500 }, { "epoch": 0.77, "eval_avg_sts": 0.7997389471860341, "eval_sickr_spearman": 0.7761020510827888, "eval_stsb_spearman": 0.8233758432892795, "step": 1500 }, { "epoch": 0.77, "learning_rate": 1.1361310133060389e-05, "loss": 0.0005, "step": 1510 }, { "epoch": 0.78, "learning_rate": 1.1105424769703174e-05, "loss": 0.0006, "step": 1520 }, { "epoch": 0.78, "learning_rate": 1.0849539406345957e-05, "loss": 0.0011, "step": 1530 }, { "epoch": 0.79, "learning_rate": 1.0593654042988742e-05, "loss": 0.0006, "step": 1540 }, { "epoch": 0.79, "learning_rate": 1.0337768679631526e-05, "loss": 0.001, "step": 1550 }, { "epoch": 0.8, "learning_rate": 1.0081883316274309e-05, "loss": 0.0011, "step": 1560 }, { "epoch": 0.8, "learning_rate": 9.825997952917093e-06, "loss": 0.001, "step": 1570 }, { "epoch": 0.81, "learning_rate": 9.570112589559878e-06, "loss": 0.0012, "step": 1580 }, { "epoch": 0.81, "learning_rate": 9.314227226202661e-06, "loss": 0.0007, "step": 1590 }, { "epoch": 0.82, "learning_rate": 9.058341862845446e-06, "loss": 0.001, "step": 1600 }, { "epoch": 0.82, "learning_rate": 8.80245649948823e-06, "loss": 0.0005, "step": 1610 }, { "epoch": 0.83, "learning_rate": 8.546571136131013e-06, "loss": 0.0012, "step": 1620 }, { "epoch": 0.83, "eval_avg_sts": 0.7990492556729138, "eval_sickr_spearman": 0.7761147312935684, "eval_stsb_spearman": 0.8219837800522592, "step": 1625 }, { "epoch": 0.83, "learning_rate": 8.290685772773797e-06, "loss": 0.0015, "step": 1630 }, { "epoch": 0.84, "learning_rate": 8.034800409416582e-06, "loss": 0.0015, "step": 1640 }, { "epoch": 0.84, "learning_rate": 7.778915046059365e-06, "loss": 0.0005, "step": 1650 }, { "epoch": 0.85, "learning_rate": 7.52302968270215e-06, "loss": 0.0006, "step": 1660 }, { "epoch": 0.85, "learning_rate": 7.267144319344933e-06, "loss": 0.0006, "step": 1670 }, { "epoch": 0.86, "learning_rate": 7.011258955987718e-06, "loss": 0.0016, "step": 1680 }, { "epoch": 0.86, "learning_rate": 6.755373592630501e-06, "loss": 0.0016, "step": 1690 }, { "epoch": 0.87, "learning_rate": 6.499488229273286e-06, "loss": 0.0017, "step": 1700 }, { "epoch": 0.88, "learning_rate": 6.24360286591607e-06, "loss": 0.0013, "step": 1710 }, { "epoch": 0.88, "learning_rate": 5.987717502558854e-06, "loss": 0.0009, "step": 1720 }, { "epoch": 0.89, "learning_rate": 5.731832139201638e-06, "loss": 0.0007, "step": 1730 }, { "epoch": 0.89, "learning_rate": 5.475946775844422e-06, "loss": 0.0002, "step": 1740 }, { "epoch": 0.9, "learning_rate": 5.220061412487206e-06, "loss": 0.0004, "step": 1750 }, { "epoch": 0.9, "eval_avg_sts": 0.7986109641804153, "eval_sickr_spearman": 0.7760801969316344, "eval_stsb_spearman": 0.8211417314291962, "step": 1750 }, { "epoch": 0.9, "learning_rate": 4.96417604912999e-06, "loss": 0.0008, "step": 1760 }, { "epoch": 0.91, "learning_rate": 4.708290685772774e-06, "loss": 0.0014, "step": 1770 }, { "epoch": 0.91, "learning_rate": 4.452405322415558e-06, "loss": 0.0007, "step": 1780 }, { "epoch": 0.92, "learning_rate": 4.1965199590583416e-06, "loss": 0.0002, "step": 1790 }, { "epoch": 0.92, "learning_rate": 3.940634595701126e-06, "loss": 0.001, "step": 1800 }, { "epoch": 0.93, "learning_rate": 3.68474923234391e-06, "loss": 0.0008, "step": 1810 }, { "epoch": 0.93, "learning_rate": 3.428863868986694e-06, "loss": 0.0015, "step": 1820 }, { "epoch": 0.94, "learning_rate": 3.172978505629478e-06, "loss": 0.0002, "step": 1830 }, { "epoch": 0.94, "learning_rate": 2.917093142272262e-06, "loss": 0.0011, "step": 1840 }, { "epoch": 0.95, "learning_rate": 2.6612077789150464e-06, "loss": 0.0005, "step": 1850 }, { "epoch": 0.95, "learning_rate": 2.4053224155578302e-06, "loss": 0.0006, "step": 1860 }, { "epoch": 0.96, "learning_rate": 2.149437052200614e-06, "loss": 0.0012, "step": 1870 }, { "epoch": 0.96, "eval_avg_sts": 0.7985301863887153, "eval_sickr_spearman": 0.7759786591831943, "eval_stsb_spearman": 0.8210817135942363, "step": 1875 }, { "epoch": 0.96, "learning_rate": 1.8935516888433984e-06, "loss": 0.0004, "step": 1880 }, { "epoch": 0.97, "learning_rate": 1.6376663254861822e-06, "loss": 0.0009, "step": 1890 }, { "epoch": 0.97, "learning_rate": 1.3817809621289663e-06, "loss": 0.0008, "step": 1900 }, { "epoch": 0.98, "learning_rate": 1.1258955987717504e-06, "loss": 0.001, "step": 1910 }, { "epoch": 0.98, "learning_rate": 8.700102354145343e-07, "loss": 0.0005, "step": 1920 }, { "epoch": 0.99, "learning_rate": 6.141248720573183e-07, "loss": 0.0002, "step": 1930 }, { "epoch": 0.99, "learning_rate": 3.5823950870010235e-07, "loss": 0.0005, "step": 1940 }, { "epoch": 1.0, "learning_rate": 1.0235414534288639e-07, "loss": 0.0004, "step": 1950 }, { "epoch": 1.0, "step": 1954, "train_runtime": 3451.5882, "train_samples_per_second": 0.566 } ], "max_steps": 1954, "num_train_epochs": 1, "total_flos": 129098564172251136, "trial_name": null, "trial_params": null }