| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "global_step": 960, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.999665344773905e-05, | |
| "loss": 2.2777, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9986614686909146e-05, | |
| "loss": 2.1303, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.996988640512931e-05, | |
| "loss": 2.143, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.994647308096509e-05, | |
| "loss": 2.2047, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.991638098272951e-05, | |
| "loss": 1.9869, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.987961816680492e-05, | |
| "loss": 2.1792, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.983619447548611e-05, | |
| "loss": 2.0121, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.9786121534345265e-05, | |
| "loss": 1.9919, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.972941274911953e-05, | |
| "loss": 1.9453, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.966608330212198e-05, | |
| "loss": 2.0629, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.9596150148176936e-05, | |
| "loss": 1.8407, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.951963201008076e-05, | |
| "loss": 1.9516, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.94365493735893e-05, | |
| "loss": 1.8641, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.934692448193334e-05, | |
| "loss": 1.8404, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.9250781329863606e-05, | |
| "loss": 1.9995, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.914814565722671e-05, | |
| "loss": 1.822, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.903904494207405e-05, | |
| "loss": 1.9633, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.892350839330522e-05, | |
| "loss": 1.8818, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.880156694284811e-05, | |
| "loss": 1.7518, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.867325323737765e-05, | |
| "loss": 1.8622, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.853860162957552e-05, | |
| "loss": 1.8353, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.839764816893315e-05, | |
| "loss": 1.9588, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.8250430592100306e-05, | |
| "loss": 1.9896, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.8096988312782174e-05, | |
| "loss": 1.8516, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.793736241118728e-05, | |
| "loss": 1.7615, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.7771595623029394e-05, | |
| "loss": 1.7356, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.759973232808609e-05, | |
| "loss": 1.9084, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.742181853831721e-05, | |
| "loss": 1.8052, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.7237901885546404e-05, | |
| "loss": 1.9753, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.7048031608708876e-05, | |
| "loss": 1.8283, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.6852258540668974e-05, | |
| "loss": 1.8031, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.665063509461097e-05, | |
| "loss": 1.842, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.6443215250006806e-05, | |
| "loss": 1.8729, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.6230054538164474e-05, | |
| "loss": 1.824, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.601121002736095e-05, | |
| "loss": 1.5677, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.5786740307563636e-05, | |
| "loss": 1.8157, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.555670547474438e-05, | |
| "loss": 1.7408, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.5321167114790385e-05, | |
| "loss": 1.8033, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.508018828701612e-05, | |
| "loss": 1.3534, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.4833833507280884e-05, | |
| "loss": 1.167, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.458216873071626e-05, | |
| "loss": 1.0725, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.4325261334068426e-05, | |
| "loss": 1.0958, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.406318009765971e-05, | |
| "loss": 1.223, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.379599518697444e-05, | |
| "loss": 1.1112, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.352377813387398e-05, | |
| "loss": 1.0701, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.324660181744589e-05, | |
| "loss": 1.1594, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.296454044449245e-05, | |
| "loss": 1.1602, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.267766952966369e-05, | |
| "loss": 1.1109, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.238606587524029e-05, | |
| "loss": 1.0719, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.208980755057178e-05, | |
| "loss": 1.1497, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.178897387117546e-05, | |
| "loss": 1.2594, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.148364537750172e-05, | |
| "loss": 1.1576, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.117390381337144e-05, | |
| "loss": 1.1553, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.085983210409114e-05, | |
| "loss": 1.1414, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.054151433425194e-05, | |
| "loss": 1.241, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.021903572521802e-05, | |
| "loss": 1.1757, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 3.9892482612310836e-05, | |
| "loss": 1.1875, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 3.956194242169506e-05, | |
| "loss": 1.0741, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 3.922750364697246e-05, | |
| "loss": 1.1855, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 3.888925582549006e-05, | |
| "loss": 1.1681, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 3.854728951436879e-05, | |
| "loss": 1.1153, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 3.82016962662592e-05, | |
| "loss": 1.2223, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 3.785256860483054e-05, | |
| "loss": 1.1862, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "loss": 1.2807, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 3.7144084842908505e-05, | |
| "loss": 1.1803, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 3.678491842064995e-05, | |
| "loss": 1.2763, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 3.6422596890760517e-05, | |
| "loss": 1.2246, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.6057217255475034e-05, | |
| "loss": 1.0887, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 3.568887733575706e-05, | |
| "loss": 1.1806, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 3.5317675745109866e-05, | |
| "loss": 1.0842, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.494371186317528e-05, | |
| "loss": 1.0559, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.456708580912725e-05, | |
| "loss": 1.0297, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 3.418789841486759e-05, | |
| "loss": 1.0726, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 3.380625119803084e-05, | |
| "loss": 1.1097, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 3.34222463348055e-05, | |
| "loss": 1.0985, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 3.303598663257904e-05, | |
| "loss": 1.182, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.264757550241384e-05, | |
| "loss": 1.0829, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 3.225711693136156e-05, | |
| "loss": 0.6117, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 3.186471545462331e-05, | |
| "loss": 0.53, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 3.147047612756302e-05, | |
| "loss": 0.6219, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 3.10745044975816e-05, | |
| "loss": 0.528, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 3.0676906575859334e-05, | |
| "loss": 0.5044, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 3.0277788808974128e-05, | |
| "loss": 0.6492, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 2.9877258050403212e-05, | |
| "loss": 0.5539, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 2.9475421531915827e-05, | |
| "loss": 0.6158, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 2.9072386834864724e-05, | |
| "loss": 0.5606, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 2.8668261861384045e-05, | |
| "loss": 0.4933, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 2.8263154805501297e-05, | |
| "loss": 0.5449, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 2.7857174124171165e-05, | |
| "loss": 0.5829, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 2.7450428508239024e-05, | |
| "loss": 0.5175, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.704302685334171e-05, | |
| "loss": 0.5473, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 2.663507823075358e-05, | |
| "loss": 0.6025, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 2.6226691858185454e-05, | |
| "loss": 0.5818, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 2.5817977070544407e-05, | |
| "loss": 0.5363, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 2.5409043290662173e-05, | |
| "loss": 0.5645, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 2.5e-05, | |
| "loss": 0.5195, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 2.459095670933783e-05, | |
| "loss": 0.5496, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 2.41820229294556e-05, | |
| "loss": 0.5588, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 2.3773308141814552e-05, | |
| "loss": 0.5454, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 2.3364921769246423e-05, | |
| "loss": 0.5319, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 2.2956973146658303e-05, | |
| "loss": 0.5125, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 2.2549571491760986e-05, | |
| "loss": 0.5035, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 2.2142825875828838e-05, | |
| "loss": 0.5413, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 2.173684519449872e-05, | |
| "loss": 0.5031, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 2.1331738138615958e-05, | |
| "loss": 0.6249, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 2.0927613165135285e-05, | |
| "loss": 0.5193, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 2.052457846808419e-05, | |
| "loss": 0.5967, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 2.0122741949596797e-05, | |
| "loss": 0.5391, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 1.972221119102587e-05, | |
| "loss": 0.5601, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 1.932309342414067e-05, | |
| "loss": 0.4991, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.8925495502418406e-05, | |
| "loss": 0.6078, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.852952387243698e-05, | |
| "loss": 0.5349, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1.81352845453767e-05, | |
| "loss": 0.4897, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 1.7742883068638447e-05, | |
| "loss": 0.5732, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 1.7352424497586163e-05, | |
| "loss": 0.5019, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 1.6964013367420966e-05, | |
| "loss": 0.3509, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 1.65777536651945e-05, | |
| "loss": 0.2495, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 1.6193748801969163e-05, | |
| "loss": 0.265, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 1.5812101585132417e-05, | |
| "loss": 0.2412, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 1.5432914190872757e-05, | |
| "loss": 0.2662, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 1.5056288136824726e-05, | |
| "loss": 0.241, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 1.4682324254890134e-05, | |
| "loss": 0.2568, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 1.4311122664242954e-05, | |
| "loss": 0.2462, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 1.3942782744524973e-05, | |
| "loss": 0.2399, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 1.3577403109239484e-05, | |
| "loss": 0.2486, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 1.3215081579350058e-05, | |
| "loss": 0.2343, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 1.2855915157091497e-05, | |
| "loss": 0.2531, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 1.2500000000000006e-05, | |
| "loss": 0.2529, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 1.2147431395169459e-05, | |
| "loss": 0.2451, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 1.1798303733740802e-05, | |
| "loss": 0.2397, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 1.1452710485631216e-05, | |
| "loss": 0.2397, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 1.1110744174509952e-05, | |
| "loss": 0.2515, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 1.0772496353027538e-05, | |
| "loss": 0.2551, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 1.043805757830495e-05, | |
| "loss": 0.293, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 1.0107517387689166e-05, | |
| "loss": 0.2861, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 9.780964274781984e-06, | |
| "loss": 0.2599, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 9.45848566574807e-06, | |
| "loss": 0.2521, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 9.140167895908867e-06, | |
| "loss": 0.264, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 8.82609618662857e-06, | |
| "loss": 0.257, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 8.51635462249828e-06, | |
| "loss": 0.2621, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 8.211026128824539e-06, | |
| "loss": 0.2529, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 7.910192449428217e-06, | |
| "loss": 0.25, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 7.613934124759711e-06, | |
| "loss": 0.2652, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 7.3223304703363135e-06, | |
| "loss": 0.2448, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 7.035459555507548e-06, | |
| "loss": 0.238, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 6.753398182554116e-06, | |
| "loss": 0.2335, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 6.476221866126029e-06, | |
| "loss": 0.2498, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 6.204004813025568e-06, | |
| "loss": 0.2657, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 5.936819902340299e-06, | |
| "loss": 0.2439, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 5.674738665931575e-06, | |
| "loss": 0.2455, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 5.417831269283741e-06, | |
| "loss": 0.2569, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 5.166166492719124e-06, | |
| "loss": 0.2566, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 4.91981171298388e-06, | |
| "loss": 0.2502, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 4.678832885209622e-06, | |
| "loss": 0.2081, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 4.4432945252556284e-06, | |
| "loss": 0.1486, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.213259692436367e-06, | |
| "loss": 0.1647, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.988789972639045e-06, | |
| "loss": 0.152, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.769945461835531e-06, | |
| "loss": 0.1594, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 3.5567847499932e-06, | |
| "loss": 0.1576, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 3.3493649053890326e-06, | |
| "loss": 0.1622, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 3.147741459331033e-06, | |
| "loss": 0.1564, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 2.9519683912911266e-06, | |
| "loss": 0.1668, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 2.7620981144536013e-06, | |
| "loss": 0.1479, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 2.578181461682794e-06, | |
| "loss": 0.1558, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 2.4002676719139166e-06, | |
| "loss": 0.1483, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 2.2284043769706027e-06, | |
| "loss": 0.1556, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 2.0626375888127187e-06, | |
| "loss": 0.1634, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 1.9030116872178316e-06, | |
| "loss": 0.1428, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 1.7495694078996983e-06, | |
| "loss": 0.1491, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 1.6023518310668618e-06, | |
| "loss": 0.1633, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 1.4613983704244826e-06, | |
| "loss": 0.1471, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 1.3267467626223606e-06, | |
| "loss": 0.1481, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 1.1984330571518931e-06, | |
| "loss": 0.1464, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 1.0764916066947794e-06, | |
| "loss": 0.1658, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 9.609550579259496e-07, | |
| "loss": 0.1535, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 8.51854342773295e-07, | |
| "loss": 0.1548, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 7.492186701364007e-07, | |
| "loss": 0.1452, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 6.530755180666592e-07, | |
| "loss": 0.1518, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 5.634506264107053e-07, | |
| "loss": 0.153, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 4.803679899192392e-07, | |
| "loss": 0.1568, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 4.0384985182306265e-07, | |
| "loss": 0.1527, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 3.339166978780256e-07, | |
| "loss": 0.144, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 2.705872508804747e-07, | |
| "loss": 0.1509, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 2.1387846565474045e-07, | |
| "loss": 0.1418, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 1.6380552451389086e-07, | |
| "loss": 0.168, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 1.2038183319507955e-07, | |
| "loss": 0.156, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 8.361901727049204e-08, | |
| "loss": 0.1364, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 5.352691903491303e-08, | |
| "loss": 0.1482, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 3.011359487068987e-08, | |
| "loss": 0.1574, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 1.3385313090857887e-08, | |
| "loss": 0.1496, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 3.346552260954927e-09, | |
| "loss": 0.1475, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.1461, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 960, | |
| "total_flos": 1778473878159360.0, | |
| "train_loss": 0.8035268472507596, | |
| "train_runtime": 399.0968, | |
| "train_samples_per_second": 4.798, | |
| "train_steps_per_second": 2.405 | |
| } | |
| ], | |
| "max_steps": 960, | |
| "num_train_epochs": 5, | |
| "total_flos": 1778473878159360.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |