| { | |
| "best_metric": 0.8684313725490196, | |
| "best_model_checkpoint": "vit-base-patch16-224-finetuned-food102/checkpoint-2151", | |
| "epoch": 2.998954339491112, | |
| "eval_steps": 500, | |
| "global_step": 2151, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.013942140118508192, | |
| "grad_norm": 11.970505714416504, | |
| "learning_rate": 2.3148148148148148e-06, | |
| "loss": 19.054, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.027884280237016383, | |
| "grad_norm": 11.977810859680176, | |
| "learning_rate": 4.6296296296296296e-06, | |
| "loss": 19.0411, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.04182642035552457, | |
| "grad_norm": 12.143768310546875, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 19.0188, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.055768560474032766, | |
| "grad_norm": 11.653632164001465, | |
| "learning_rate": 9.259259259259259e-06, | |
| "loss": 18.7585, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.06971070059254096, | |
| "grad_norm": 13.25544261932373, | |
| "learning_rate": 1.1574074074074075e-05, | |
| "loss": 18.4989, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.08365284071104914, | |
| "grad_norm": 12.296609878540039, | |
| "learning_rate": 1.388888888888889e-05, | |
| "loss": 18.199, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.09759498082955734, | |
| "grad_norm": 11.88782024383545, | |
| "learning_rate": 1.6203703703703704e-05, | |
| "loss": 17.9952, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.11153712094806553, | |
| "grad_norm": 11.943421363830566, | |
| "learning_rate": 1.8518518518518518e-05, | |
| "loss": 17.5776, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.12547926106657373, | |
| "grad_norm": 12.558123588562012, | |
| "learning_rate": 2.0833333333333336e-05, | |
| "loss": 17.0623, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.13942140118508192, | |
| "grad_norm": 13.311689376831055, | |
| "learning_rate": 2.314814814814815e-05, | |
| "loss": 16.5161, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1533635413035901, | |
| "grad_norm": 14.52517318725586, | |
| "learning_rate": 2.5462962962962965e-05, | |
| "loss": 15.9671, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.16730568142209828, | |
| "grad_norm": 12.310554504394531, | |
| "learning_rate": 2.777777777777778e-05, | |
| "loss": 15.2662, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.18124782154060648, | |
| "grad_norm": 12.67236042022705, | |
| "learning_rate": 3.0092592592592593e-05, | |
| "loss": 14.4554, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.19518996165911467, | |
| "grad_norm": 12.847444534301758, | |
| "learning_rate": 3.240740740740741e-05, | |
| "loss": 13.7588, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.20913210177762287, | |
| "grad_norm": 12.77316665649414, | |
| "learning_rate": 3.472222222222222e-05, | |
| "loss": 13.0135, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.22307424189613106, | |
| "grad_norm": 13.008152961730957, | |
| "learning_rate": 3.7037037037037037e-05, | |
| "loss": 12.324, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.23701638201463926, | |
| "grad_norm": 13.326498985290527, | |
| "learning_rate": 3.935185185185186e-05, | |
| "loss": 11.2685, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.25095852213314745, | |
| "grad_norm": 13.166327476501465, | |
| "learning_rate": 4.166666666666667e-05, | |
| "loss": 10.491, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.26490066225165565, | |
| "grad_norm": 12.491602897644043, | |
| "learning_rate": 4.3981481481481486e-05, | |
| "loss": 9.8871, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.27884280237016384, | |
| "grad_norm": 12.803274154663086, | |
| "learning_rate": 4.62962962962963e-05, | |
| "loss": 9.0965, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.292784942488672, | |
| "grad_norm": 13.318122863769531, | |
| "learning_rate": 4.8611111111111115e-05, | |
| "loss": 8.2912, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.3067270826071802, | |
| "grad_norm": 12.178855895996094, | |
| "learning_rate": 4.989664082687339e-05, | |
| "loss": 8.0104, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.3206692227256884, | |
| "grad_norm": 11.760112762451172, | |
| "learning_rate": 4.963824289405685e-05, | |
| "loss": 7.2478, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.33461136284419657, | |
| "grad_norm": 12.436824798583984, | |
| "learning_rate": 4.9379844961240315e-05, | |
| "loss": 6.8172, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.34855350296270476, | |
| "grad_norm": 11.301292419433594, | |
| "learning_rate": 4.912144702842377e-05, | |
| "loss": 6.5386, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.36249564308121296, | |
| "grad_norm": 12.099160194396973, | |
| "learning_rate": 4.886304909560724e-05, | |
| "loss": 6.1708, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.37643778319972115, | |
| "grad_norm": 12.640030860900879, | |
| "learning_rate": 4.86046511627907e-05, | |
| "loss": 6.0798, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.39037992331822935, | |
| "grad_norm": 12.914190292358398, | |
| "learning_rate": 4.834625322997416e-05, | |
| "loss": 5.5983, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.40432206343673754, | |
| "grad_norm": 13.290894508361816, | |
| "learning_rate": 4.8087855297157624e-05, | |
| "loss": 5.6591, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.41826420355524574, | |
| "grad_norm": 12.26599407196045, | |
| "learning_rate": 4.782945736434109e-05, | |
| "loss": 5.0549, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.43220634367375393, | |
| "grad_norm": 13.02130126953125, | |
| "learning_rate": 4.757105943152455e-05, | |
| "loss": 4.8798, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.44614848379226213, | |
| "grad_norm": 11.812455177307129, | |
| "learning_rate": 4.7312661498708014e-05, | |
| "loss": 4.8651, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.4600906239107703, | |
| "grad_norm": 15.11469554901123, | |
| "learning_rate": 4.705426356589148e-05, | |
| "loss": 4.8677, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.4740327640292785, | |
| "grad_norm": 13.028066635131836, | |
| "learning_rate": 4.679586563307494e-05, | |
| "loss": 4.7999, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.48797490414778666, | |
| "grad_norm": 14.58755111694336, | |
| "learning_rate": 4.65374677002584e-05, | |
| "loss": 4.7765, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.5019170442662949, | |
| "grad_norm": 11.573223114013672, | |
| "learning_rate": 4.627906976744186e-05, | |
| "loss": 4.6033, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.515859184384803, | |
| "grad_norm": 13.456429481506348, | |
| "learning_rate": 4.602067183462532e-05, | |
| "loss": 4.406, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.5298013245033113, | |
| "grad_norm": 13.06003475189209, | |
| "learning_rate": 4.5762273901808786e-05, | |
| "loss": 4.4271, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.5437434646218194, | |
| "grad_norm": 12.669313430786133, | |
| "learning_rate": 4.550387596899225e-05, | |
| "loss": 4.1339, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.5576856047403277, | |
| "grad_norm": 12.845880508422852, | |
| "learning_rate": 4.524547803617571e-05, | |
| "loss": 4.3616, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.5716277448588358, | |
| "grad_norm": 13.643671035766602, | |
| "learning_rate": 4.4987080103359175e-05, | |
| "loss": 4.03, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.585569884977344, | |
| "grad_norm": 12.903581619262695, | |
| "learning_rate": 4.472868217054264e-05, | |
| "loss": 4.0456, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.5995120250958522, | |
| "grad_norm": 11.996821403503418, | |
| "learning_rate": 4.44702842377261e-05, | |
| "loss": 4.1508, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.6134541652143604, | |
| "grad_norm": 12.505786895751953, | |
| "learning_rate": 4.421188630490956e-05, | |
| "loss": 4.112, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.6273963053328686, | |
| "grad_norm": 12.409116744995117, | |
| "learning_rate": 4.395348837209303e-05, | |
| "loss": 3.8887, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.6413384454513767, | |
| "grad_norm": 12.7697114944458, | |
| "learning_rate": 4.3695090439276484e-05, | |
| "loss": 3.7454, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.655280585569885, | |
| "grad_norm": 13.008294105529785, | |
| "learning_rate": 4.3436692506459954e-05, | |
| "loss": 3.879, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.6692227256883931, | |
| "grad_norm": 12.83389949798584, | |
| "learning_rate": 4.317829457364341e-05, | |
| "loss": 4.032, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.6831648658069014, | |
| "grad_norm": 13.29801082611084, | |
| "learning_rate": 4.2919896640826874e-05, | |
| "loss": 3.9523, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.6971070059254095, | |
| "grad_norm": 13.225083351135254, | |
| "learning_rate": 4.266149870801034e-05, | |
| "loss": 3.7454, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.7110491460439178, | |
| "grad_norm": 13.245360374450684, | |
| "learning_rate": 4.24031007751938e-05, | |
| "loss": 3.5761, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.7249912861624259, | |
| "grad_norm": 12.166769981384277, | |
| "learning_rate": 4.214470284237726e-05, | |
| "loss": 3.8125, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.7389334262809342, | |
| "grad_norm": 13.542282104492188, | |
| "learning_rate": 4.1886304909560726e-05, | |
| "loss": 3.6979, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.7528755663994423, | |
| "grad_norm": 17.77124786376953, | |
| "learning_rate": 4.162790697674418e-05, | |
| "loss": 3.4578, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.7668177065179506, | |
| "grad_norm": 13.022871971130371, | |
| "learning_rate": 4.136950904392765e-05, | |
| "loss": 3.4832, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.7807598466364587, | |
| "grad_norm": 15.890631675720215, | |
| "learning_rate": 4.111111111111111e-05, | |
| "loss": 3.5997, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.7947019867549668, | |
| "grad_norm": 14.40475845336914, | |
| "learning_rate": 4.085271317829458e-05, | |
| "loss": 3.6687, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.8086441268734751, | |
| "grad_norm": 10.683858871459961, | |
| "learning_rate": 4.0594315245478035e-05, | |
| "loss": 3.3821, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.8225862669919832, | |
| "grad_norm": 14.170791625976562, | |
| "learning_rate": 4.0335917312661505e-05, | |
| "loss": 3.4403, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.8365284071104915, | |
| "grad_norm": 12.744939804077148, | |
| "learning_rate": 4.007751937984496e-05, | |
| "loss": 3.1932, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.8504705472289996, | |
| "grad_norm": 12.260272979736328, | |
| "learning_rate": 3.9819121447028425e-05, | |
| "loss": 3.5403, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.8644126873475079, | |
| "grad_norm": 14.6715669631958, | |
| "learning_rate": 3.956072351421189e-05, | |
| "loss": 3.4033, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.878354827466016, | |
| "grad_norm": 12.32928466796875, | |
| "learning_rate": 3.930232558139535e-05, | |
| "loss": 3.2728, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.8922969675845243, | |
| "grad_norm": 13.14637279510498, | |
| "learning_rate": 3.9043927648578814e-05, | |
| "loss": 3.3631, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.9062391077030324, | |
| "grad_norm": 11.95271110534668, | |
| "learning_rate": 3.878552971576228e-05, | |
| "loss": 3.2798, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.9201812478215406, | |
| "grad_norm": 12.834661483764648, | |
| "learning_rate": 3.8527131782945734e-05, | |
| "loss": 3.5264, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.9341233879400488, | |
| "grad_norm": 13.227917671203613, | |
| "learning_rate": 3.8268733850129204e-05, | |
| "loss": 3.5221, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.948065528058557, | |
| "grad_norm": 13.852123260498047, | |
| "learning_rate": 3.801033591731266e-05, | |
| "loss": 3.1419, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.9620076681770652, | |
| "grad_norm": 16.48082160949707, | |
| "learning_rate": 3.775193798449613e-05, | |
| "loss": 3.4653, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.9759498082955733, | |
| "grad_norm": 11.60905933380127, | |
| "learning_rate": 3.7493540051679586e-05, | |
| "loss": 3.1008, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.9898919484140816, | |
| "grad_norm": 25.217321395874023, | |
| "learning_rate": 3.7235142118863056e-05, | |
| "loss": 3.3941, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.9996514464970373, | |
| "eval_accuracy": 0.8350980392156863, | |
| "eval_loss": 0.6625040769577026, | |
| "eval_runtime": 286.4591, | |
| "eval_samples_per_second": 35.607, | |
| "eval_steps_per_second": 1.114, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 1.0038340885325898, | |
| "grad_norm": 10.983098983764648, | |
| "learning_rate": 3.697674418604651e-05, | |
| "loss": 3.2282, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.0177762286510978, | |
| "grad_norm": 12.984674453735352, | |
| "learning_rate": 3.6718346253229976e-05, | |
| "loss": 2.9876, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.031718368769606, | |
| "grad_norm": 11.363933563232422, | |
| "learning_rate": 3.645994832041344e-05, | |
| "loss": 3.1042, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.0456605088881143, | |
| "grad_norm": 12.959346771240234, | |
| "learning_rate": 3.62015503875969e-05, | |
| "loss": 2.8314, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.0596026490066226, | |
| "grad_norm": 13.28182601928711, | |
| "learning_rate": 3.5943152454780365e-05, | |
| "loss": 2.9215, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.0735447891251306, | |
| "grad_norm": 12.953139305114746, | |
| "learning_rate": 3.568475452196383e-05, | |
| "loss": 2.9363, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.0874869292436389, | |
| "grad_norm": 12.209274291992188, | |
| "learning_rate": 3.5426356589147285e-05, | |
| "loss": 2.9331, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.1014290693621471, | |
| "grad_norm": 12.68194580078125, | |
| "learning_rate": 3.5167958656330755e-05, | |
| "loss": 2.9515, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.1153712094806554, | |
| "grad_norm": 11.559549331665039, | |
| "learning_rate": 3.490956072351421e-05, | |
| "loss": 2.7768, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.1293133495991634, | |
| "grad_norm": 13.342930793762207, | |
| "learning_rate": 3.465116279069768e-05, | |
| "loss": 2.8238, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.1432554897176717, | |
| "grad_norm": 14.305107116699219, | |
| "learning_rate": 3.439276485788114e-05, | |
| "loss": 2.9598, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.15719762983618, | |
| "grad_norm": 12.436918258666992, | |
| "learning_rate": 3.41343669250646e-05, | |
| "loss": 2.8277, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.1711397699546882, | |
| "grad_norm": 12.794795036315918, | |
| "learning_rate": 3.3875968992248064e-05, | |
| "loss": 2.7674, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.1850819100731962, | |
| "grad_norm": 12.380959510803223, | |
| "learning_rate": 3.361757105943153e-05, | |
| "loss": 2.6084, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.1990240501917044, | |
| "grad_norm": 16.531293869018555, | |
| "learning_rate": 3.335917312661499e-05, | |
| "loss": 2.6154, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.2129661903102127, | |
| "grad_norm": 12.086883544921875, | |
| "learning_rate": 3.310077519379845e-05, | |
| "loss": 2.9116, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.2269083304287207, | |
| "grad_norm": 11.90904712677002, | |
| "learning_rate": 3.2842377260981916e-05, | |
| "loss": 2.7052, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.240850470547229, | |
| "grad_norm": 13.168009757995605, | |
| "learning_rate": 3.258397932816537e-05, | |
| "loss": 2.5401, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.2547926106657372, | |
| "grad_norm": 13.133540153503418, | |
| "learning_rate": 3.2325581395348836e-05, | |
| "loss": 3.089, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.2687347507842452, | |
| "grad_norm": 10.270415306091309, | |
| "learning_rate": 3.20671834625323e-05, | |
| "loss": 2.5938, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.2826768909027535, | |
| "grad_norm": 13.742632865905762, | |
| "learning_rate": 3.180878552971576e-05, | |
| "loss": 2.6841, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.2966190310212617, | |
| "grad_norm": 10.14291000366211, | |
| "learning_rate": 3.1550387596899225e-05, | |
| "loss": 2.5288, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.31056117113977, | |
| "grad_norm": 11.036438941955566, | |
| "learning_rate": 3.129198966408269e-05, | |
| "loss": 2.8172, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.3245033112582782, | |
| "grad_norm": 11.163543701171875, | |
| "learning_rate": 3.103359173126615e-05, | |
| "loss": 2.8402, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.3384454513767863, | |
| "grad_norm": 16.406246185302734, | |
| "learning_rate": 3.0775193798449615e-05, | |
| "loss": 2.8161, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.3523875914952945, | |
| "grad_norm": 11.284817695617676, | |
| "learning_rate": 3.0516795865633074e-05, | |
| "loss": 2.448, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.3663297316138028, | |
| "grad_norm": 12.130209922790527, | |
| "learning_rate": 3.025839793281654e-05, | |
| "loss": 2.8401, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.3802718717323108, | |
| "grad_norm": 11.599443435668945, | |
| "learning_rate": 3e-05, | |
| "loss": 2.6025, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.394214011850819, | |
| "grad_norm": 11.705489158630371, | |
| "learning_rate": 2.9741602067183467e-05, | |
| "loss": 2.6507, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.4081561519693273, | |
| "grad_norm": 11.031926155090332, | |
| "learning_rate": 2.9483204134366927e-05, | |
| "loss": 2.8557, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.4220982920878356, | |
| "grad_norm": 13.046774864196777, | |
| "learning_rate": 2.9224806201550387e-05, | |
| "loss": 2.972, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.4360404322063438, | |
| "grad_norm": 12.700828552246094, | |
| "learning_rate": 2.8966408268733853e-05, | |
| "loss": 2.7807, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.4499825723248518, | |
| "grad_norm": 15.926816940307617, | |
| "learning_rate": 2.8708010335917313e-05, | |
| "loss": 2.5443, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.46392471244336, | |
| "grad_norm": 14.657876968383789, | |
| "learning_rate": 2.844961240310078e-05, | |
| "loss": 2.8499, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.4778668525618683, | |
| "grad_norm": 12.2283935546875, | |
| "learning_rate": 2.819121447028424e-05, | |
| "loss": 2.6669, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.4918089926803764, | |
| "grad_norm": 14.132109642028809, | |
| "learning_rate": 2.7932816537467703e-05, | |
| "loss": 2.75, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.5057511327988846, | |
| "grad_norm": 15.497398376464844, | |
| "learning_rate": 2.7674418604651166e-05, | |
| "loss": 2.6658, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.5196932729173929, | |
| "grad_norm": 39.895477294921875, | |
| "learning_rate": 2.7416020671834625e-05, | |
| "loss": 2.8124, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.533635413035901, | |
| "grad_norm": 12.836750030517578, | |
| "learning_rate": 2.7157622739018092e-05, | |
| "loss": 2.7151, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.5475775531544091, | |
| "grad_norm": 12.65983772277832, | |
| "learning_rate": 2.6899224806201552e-05, | |
| "loss": 2.5187, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.5615196932729174, | |
| "grad_norm": 12.410466194152832, | |
| "learning_rate": 2.6640826873385015e-05, | |
| "loss": 2.7021, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.5754618333914254, | |
| "grad_norm": 16.45282554626465, | |
| "learning_rate": 2.6382428940568478e-05, | |
| "loss": 2.8277, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.589403973509934, | |
| "grad_norm": 12.401958465576172, | |
| "learning_rate": 2.6124031007751938e-05, | |
| "loss": 2.6465, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.603346113628442, | |
| "grad_norm": 13.451824188232422, | |
| "learning_rate": 2.58656330749354e-05, | |
| "loss": 2.7568, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.6172882537469502, | |
| "grad_norm": 12.007116317749023, | |
| "learning_rate": 2.5607235142118864e-05, | |
| "loss": 2.5732, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.6312303938654584, | |
| "grad_norm": 13.369595527648926, | |
| "learning_rate": 2.5348837209302327e-05, | |
| "loss": 2.5993, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.6451725339839665, | |
| "grad_norm": 13.079323768615723, | |
| "learning_rate": 2.5090439276485787e-05, | |
| "loss": 2.557, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.6591146741024747, | |
| "grad_norm": 12.571920394897461, | |
| "learning_rate": 2.483204134366925e-05, | |
| "loss": 2.5582, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.673056814220983, | |
| "grad_norm": 12.451131820678711, | |
| "learning_rate": 2.4573643410852713e-05, | |
| "loss": 2.485, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.686998954339491, | |
| "grad_norm": 11.426173210144043, | |
| "learning_rate": 2.4315245478036177e-05, | |
| "loss": 2.8228, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.7009410944579995, | |
| "grad_norm": 13.152283668518066, | |
| "learning_rate": 2.405684754521964e-05, | |
| "loss": 2.9743, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.7148832345765075, | |
| "grad_norm": 12.218315124511719, | |
| "learning_rate": 2.37984496124031e-05, | |
| "loss": 2.5617, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.7288253746950157, | |
| "grad_norm": 16.104169845581055, | |
| "learning_rate": 2.3540051679586563e-05, | |
| "loss": 2.5886, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.742767514813524, | |
| "grad_norm": 9.216697692871094, | |
| "learning_rate": 2.3281653746770026e-05, | |
| "loss": 2.2698, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.756709654932032, | |
| "grad_norm": 12.568550109863281, | |
| "learning_rate": 2.302325581395349e-05, | |
| "loss": 2.7008, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.7706517950505403, | |
| "grad_norm": 13.100058555603027, | |
| "learning_rate": 2.2764857881136952e-05, | |
| "loss": 2.5122, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.7845939351690485, | |
| "grad_norm": 14.60105037689209, | |
| "learning_rate": 2.2506459948320415e-05, | |
| "loss": 2.7869, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.7985360752875565, | |
| "grad_norm": 11.809743881225586, | |
| "learning_rate": 2.2248062015503875e-05, | |
| "loss": 3.0241, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.8124782154060648, | |
| "grad_norm": 11.19316291809082, | |
| "learning_rate": 2.1989664082687338e-05, | |
| "loss": 2.3842, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.826420355524573, | |
| "grad_norm": 11.623698234558105, | |
| "learning_rate": 2.17312661498708e-05, | |
| "loss": 2.5165, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.840362495643081, | |
| "grad_norm": 12.553630828857422, | |
| "learning_rate": 2.1472868217054264e-05, | |
| "loss": 2.6636, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.8543046357615895, | |
| "grad_norm": 16.92484474182129, | |
| "learning_rate": 2.1214470284237728e-05, | |
| "loss": 2.8168, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.8682467758800976, | |
| "grad_norm": 11.145873069763184, | |
| "learning_rate": 2.095607235142119e-05, | |
| "loss": 2.498, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.8821889159986058, | |
| "grad_norm": 13.681021690368652, | |
| "learning_rate": 2.069767441860465e-05, | |
| "loss": 2.5061, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.896131056117114, | |
| "grad_norm": 13.941627502441406, | |
| "learning_rate": 2.0439276485788114e-05, | |
| "loss": 2.5961, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.910073196235622, | |
| "grad_norm": 10.970531463623047, | |
| "learning_rate": 2.0180878552971577e-05, | |
| "loss": 2.5375, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.9240153363541304, | |
| "grad_norm": 11.011059761047363, | |
| "learning_rate": 1.992248062015504e-05, | |
| "loss": 2.6054, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.9379574764726386, | |
| "grad_norm": 11.878082275390625, | |
| "learning_rate": 1.9664082687338503e-05, | |
| "loss": 2.487, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.9518996165911466, | |
| "grad_norm": 12.48755931854248, | |
| "learning_rate": 1.9405684754521966e-05, | |
| "loss": 2.8289, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.9658417567096549, | |
| "grad_norm": 14.737470626831055, | |
| "learning_rate": 1.9147286821705426e-05, | |
| "loss": 2.5594, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.9797838968281631, | |
| "grad_norm": 11.880918502807617, | |
| "learning_rate": 1.888888888888889e-05, | |
| "loss": 2.4736, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.9937260369466712, | |
| "grad_norm": 14.554669380187988, | |
| "learning_rate": 1.8630490956072352e-05, | |
| "loss": 2.6442, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.9993028929940746, | |
| "eval_accuracy": 0.8597058823529412, | |
| "eval_loss": 0.5420134663581848, | |
| "eval_runtime": 226.1399, | |
| "eval_samples_per_second": 45.105, | |
| "eval_steps_per_second": 1.411, | |
| "step": 1434 | |
| }, | |
| { | |
| "epoch": 2.0076681770651796, | |
| "grad_norm": 11.852781295776367, | |
| "learning_rate": 1.8372093023255815e-05, | |
| "loss": 2.2307, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.0216103171836877, | |
| "grad_norm": 12.653249740600586, | |
| "learning_rate": 1.811369509043928e-05, | |
| "loss": 2.3448, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.0355524573021957, | |
| "grad_norm": 15.625718116760254, | |
| "learning_rate": 1.7855297157622742e-05, | |
| "loss": 2.4701, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.049494597420704, | |
| "grad_norm": 10.152451515197754, | |
| "learning_rate": 1.75968992248062e-05, | |
| "loss": 2.2279, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.063436737539212, | |
| "grad_norm": 11.112920761108398, | |
| "learning_rate": 1.7338501291989665e-05, | |
| "loss": 2.5778, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.0773788776577207, | |
| "grad_norm": 13.829360008239746, | |
| "learning_rate": 1.7080103359173128e-05, | |
| "loss": 2.2665, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.0913210177762287, | |
| "grad_norm": 12.493084907531738, | |
| "learning_rate": 1.682170542635659e-05, | |
| "loss": 2.5002, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.1052631578947367, | |
| "grad_norm": 11.931866645812988, | |
| "learning_rate": 1.6563307493540054e-05, | |
| "loss": 2.3208, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.119205298013245, | |
| "grad_norm": 10.55782413482666, | |
| "learning_rate": 1.6304909560723514e-05, | |
| "loss": 2.2226, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.1331474381317532, | |
| "grad_norm": 14.521842002868652, | |
| "learning_rate": 1.6046511627906977e-05, | |
| "loss": 2.1835, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.1470895782502613, | |
| "grad_norm": 12.054328918457031, | |
| "learning_rate": 1.578811369509044e-05, | |
| "loss": 2.3775, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.1610317183687697, | |
| "grad_norm": 15.088787078857422, | |
| "learning_rate": 1.5529715762273903e-05, | |
| "loss": 2.4388, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.1749738584872778, | |
| "grad_norm": 13.146328926086426, | |
| "learning_rate": 1.5271317829457366e-05, | |
| "loss": 2.5277, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.188915998605786, | |
| "grad_norm": 12.280598640441895, | |
| "learning_rate": 1.5012919896640828e-05, | |
| "loss": 2.3422, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.2028581387242943, | |
| "grad_norm": 12.004185676574707, | |
| "learning_rate": 1.475452196382429e-05, | |
| "loss": 2.4908, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.2168002788428023, | |
| "grad_norm": 10.80855941772461, | |
| "learning_rate": 1.4496124031007753e-05, | |
| "loss": 2.357, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.2307424189613108, | |
| "grad_norm": 11.332206726074219, | |
| "learning_rate": 1.4237726098191214e-05, | |
| "loss": 2.2628, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.244684559079819, | |
| "grad_norm": 13.836607933044434, | |
| "learning_rate": 1.3979328165374677e-05, | |
| "loss": 2.1193, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 2.258626699198327, | |
| "grad_norm": 11.600666999816895, | |
| "learning_rate": 1.372093023255814e-05, | |
| "loss": 2.4166, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.2725688393168353, | |
| "grad_norm": 14.399880409240723, | |
| "learning_rate": 1.3462532299741603e-05, | |
| "loss": 2.2878, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 2.2865109794353433, | |
| "grad_norm": 16.899677276611328, | |
| "learning_rate": 1.3204134366925063e-05, | |
| "loss": 2.26, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 2.3004531195538513, | |
| "grad_norm": 12.676505088806152, | |
| "learning_rate": 1.2945736434108526e-05, | |
| "loss": 2.5939, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.31439525967236, | |
| "grad_norm": 11.750486373901367, | |
| "learning_rate": 1.268733850129199e-05, | |
| "loss": 2.2246, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 2.328337399790868, | |
| "grad_norm": 14.333374977111816, | |
| "learning_rate": 1.2428940568475453e-05, | |
| "loss": 2.4163, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 2.3422795399093763, | |
| "grad_norm": 12.187148094177246, | |
| "learning_rate": 1.2170542635658916e-05, | |
| "loss": 2.3381, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 2.3562216800278843, | |
| "grad_norm": 10.980198860168457, | |
| "learning_rate": 1.1912144702842377e-05, | |
| "loss": 2.3502, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 2.3701638201463924, | |
| "grad_norm": 13.697635650634766, | |
| "learning_rate": 1.165374677002584e-05, | |
| "loss": 2.2907, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.384105960264901, | |
| "grad_norm": 12.195317268371582, | |
| "learning_rate": 1.1395348837209304e-05, | |
| "loss": 2.2207, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 2.398048100383409, | |
| "grad_norm": 39.64952850341797, | |
| "learning_rate": 1.1136950904392765e-05, | |
| "loss": 2.2268, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 2.411990240501917, | |
| "grad_norm": 11.313281059265137, | |
| "learning_rate": 1.0878552971576228e-05, | |
| "loss": 2.2816, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 2.4259323806204254, | |
| "grad_norm": 13.669633865356445, | |
| "learning_rate": 1.0620155038759691e-05, | |
| "loss": 2.4572, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 2.4398745207389334, | |
| "grad_norm": 13.479631423950195, | |
| "learning_rate": 1.0361757105943153e-05, | |
| "loss": 2.094, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.4538166608574414, | |
| "grad_norm": 10.151293754577637, | |
| "learning_rate": 1.0103359173126616e-05, | |
| "loss": 2.0609, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 2.46775880097595, | |
| "grad_norm": 14.358535766601562, | |
| "learning_rate": 9.844961240310079e-06, | |
| "loss": 2.4335, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 2.481700941094458, | |
| "grad_norm": 14.903388023376465, | |
| "learning_rate": 9.58656330749354e-06, | |
| "loss": 2.2056, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 2.4956430812129664, | |
| "grad_norm": 12.196182250976562, | |
| "learning_rate": 9.328165374677004e-06, | |
| "loss": 2.358, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 2.5095852213314744, | |
| "grad_norm": 10.977519989013672, | |
| "learning_rate": 9.069767441860467e-06, | |
| "loss": 2.1615, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.5235273614499825, | |
| "grad_norm": 10.606942176818848, | |
| "learning_rate": 8.811369509043928e-06, | |
| "loss": 2.1802, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 2.5374695015684905, | |
| "grad_norm": 11.948677062988281, | |
| "learning_rate": 8.552971576227391e-06, | |
| "loss": 2.3221, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 2.551411641686999, | |
| "grad_norm": 8.935708999633789, | |
| "learning_rate": 8.294573643410853e-06, | |
| "loss": 2.0471, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 2.565353781805507, | |
| "grad_norm": 11.44541072845459, | |
| "learning_rate": 8.036175710594316e-06, | |
| "loss": 2.1508, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 2.5792959219240155, | |
| "grad_norm": 12.39538288116455, | |
| "learning_rate": 7.777777777777777e-06, | |
| "loss": 2.253, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.5932380620425235, | |
| "grad_norm": 16.37424087524414, | |
| "learning_rate": 7.5193798449612415e-06, | |
| "loss": 2.2232, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 2.6071802021610315, | |
| "grad_norm": 11.820206642150879, | |
| "learning_rate": 7.260981912144703e-06, | |
| "loss": 2.4425, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 2.62112234227954, | |
| "grad_norm": 12.931533813476562, | |
| "learning_rate": 7.002583979328166e-06, | |
| "loss": 2.2895, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 2.635064482398048, | |
| "grad_norm": 13.43673038482666, | |
| "learning_rate": 6.744186046511629e-06, | |
| "loss": 2.2439, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 2.6490066225165565, | |
| "grad_norm": 10.057219505310059, | |
| "learning_rate": 6.485788113695091e-06, | |
| "loss": 2.2317, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.6629487626350645, | |
| "grad_norm": 15.320812225341797, | |
| "learning_rate": 6.227390180878554e-06, | |
| "loss": 2.3445, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 2.6768909027535726, | |
| "grad_norm": 13.997590065002441, | |
| "learning_rate": 5.968992248062015e-06, | |
| "loss": 2.1295, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 2.690833042872081, | |
| "grad_norm": 10.897136688232422, | |
| "learning_rate": 5.710594315245478e-06, | |
| "loss": 2.0328, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 2.704775182990589, | |
| "grad_norm": 11.251324653625488, | |
| "learning_rate": 5.452196382428941e-06, | |
| "loss": 2.2964, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 2.718717323109097, | |
| "grad_norm": 12.688488960266113, | |
| "learning_rate": 5.193798449612403e-06, | |
| "loss": 2.585, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.7326594632276056, | |
| "grad_norm": 15.246435165405273, | |
| "learning_rate": 4.935400516795865e-06, | |
| "loss": 2.4457, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 2.7466016033461136, | |
| "grad_norm": 11.022492408752441, | |
| "learning_rate": 4.6770025839793285e-06, | |
| "loss": 1.9957, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 2.7605437434646216, | |
| "grad_norm": 15.17203426361084, | |
| "learning_rate": 4.418604651162791e-06, | |
| "loss": 2.3373, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 2.77448588358313, | |
| "grad_norm": 13.542671203613281, | |
| "learning_rate": 4.160206718346253e-06, | |
| "loss": 2.392, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 2.788428023701638, | |
| "grad_norm": 15.951026916503906, | |
| "learning_rate": 3.901808785529716e-06, | |
| "loss": 1.9829, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.8023701638201466, | |
| "grad_norm": 13.995731353759766, | |
| "learning_rate": 3.6434108527131786e-06, | |
| "loss": 2.4399, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 2.8163123039386546, | |
| "grad_norm": 11.442358016967773, | |
| "learning_rate": 3.385012919896641e-06, | |
| "loss": 2.2327, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 2.8302544440571626, | |
| "grad_norm": 13.158218383789062, | |
| "learning_rate": 3.1266149870801036e-06, | |
| "loss": 2.1585, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 2.844196584175671, | |
| "grad_norm": 14.946197509765625, | |
| "learning_rate": 2.868217054263566e-06, | |
| "loss": 2.2202, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 2.858138724294179, | |
| "grad_norm": 13.085625648498535, | |
| "learning_rate": 2.6098191214470286e-06, | |
| "loss": 2.3849, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.8720808644126876, | |
| "grad_norm": 11.152565002441406, | |
| "learning_rate": 2.351421188630491e-06, | |
| "loss": 2.3714, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 2.8860230045311956, | |
| "grad_norm": 13.09467601776123, | |
| "learning_rate": 2.0930232558139536e-06, | |
| "loss": 2.1876, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 2.8999651446497037, | |
| "grad_norm": 17.58858871459961, | |
| "learning_rate": 1.8346253229974162e-06, | |
| "loss": 2.2959, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 2.9139072847682117, | |
| "grad_norm": 16.1025447845459, | |
| "learning_rate": 1.5762273901808785e-06, | |
| "loss": 2.3838, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 2.92784942488672, | |
| "grad_norm": 13.546030044555664, | |
| "learning_rate": 1.3178294573643412e-06, | |
| "loss": 2.14, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.941791565005228, | |
| "grad_norm": 12.100831985473633, | |
| "learning_rate": 1.0594315245478035e-06, | |
| "loss": 2.0447, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 2.9557337051237367, | |
| "grad_norm": 12.555295944213867, | |
| "learning_rate": 8.010335917312662e-07, | |
| "loss": 2.4814, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 2.9696758452422447, | |
| "grad_norm": 11.319598197937012, | |
| "learning_rate": 5.426356589147287e-07, | |
| "loss": 2.3802, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 2.9836179853607527, | |
| "grad_norm": 10.149794578552246, | |
| "learning_rate": 2.842377260981912e-07, | |
| "loss": 2.3564, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 2.997560125479261, | |
| "grad_norm": 13.360361099243164, | |
| "learning_rate": 2.5839793281653747e-08, | |
| "loss": 2.1182, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.998954339491112, | |
| "eval_accuracy": 0.8684313725490196, | |
| "eval_loss": 0.5095656514167786, | |
| "eval_runtime": 245.9211, | |
| "eval_samples_per_second": 41.477, | |
| "eval_steps_per_second": 1.297, | |
| "step": 2151 | |
| }, | |
| { | |
| "epoch": 2.998954339491112, | |
| "step": 2151, | |
| "total_flos": 2.1353596059581743e+19, | |
| "train_loss": 4.144423897296315, | |
| "train_runtime": 8280.2172, | |
| "train_samples_per_second": 33.26, | |
| "train_steps_per_second": 0.26 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2151, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.1353596059581743e+19, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |