| { | |
| "best_metric": 0.8859984697781178, | |
| "best_model_checkpoint": "distilbert-base-multilingual-cased-aoe-hyper/run-0/checkpoint-2616", | |
| "epoch": 4.0, | |
| "eval_steps": 500, | |
| "global_step": 2616, | |
| "is_hyper_param_search": true, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01529051987767584, | |
| "grad_norm": 1.6358972787857056, | |
| "learning_rate": 9.349717786759033e-06, | |
| "loss": 0.6796, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03058103975535168, | |
| "grad_norm": 1.7600407600402832, | |
| "learning_rate": 9.321037670848728e-06, | |
| "loss": 0.5945, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.045871559633027525, | |
| "grad_norm": 2.3243937492370605, | |
| "learning_rate": 9.292357554938425e-06, | |
| "loss": 0.5386, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.06116207951070336, | |
| "grad_norm": 2.1618645191192627, | |
| "learning_rate": 9.263677439028121e-06, | |
| "loss": 0.4348, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0764525993883792, | |
| "grad_norm": 4.011763572692871, | |
| "learning_rate": 9.234997323117817e-06, | |
| "loss": 0.4917, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.09174311926605505, | |
| "grad_norm": 3.0631511211395264, | |
| "learning_rate": 9.206317207207513e-06, | |
| "loss": 0.4419, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.10703363914373089, | |
| "grad_norm": 5.19169807434082, | |
| "learning_rate": 9.177637091297209e-06, | |
| "loss": 0.4885, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.12232415902140673, | |
| "grad_norm": 3.2671306133270264, | |
| "learning_rate": 9.148956975386905e-06, | |
| "loss": 0.3778, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.13761467889908258, | |
| "grad_norm": 2.0801641941070557, | |
| "learning_rate": 9.120276859476602e-06, | |
| "loss": 0.4524, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.1529051987767584, | |
| "grad_norm": 3.2077746391296387, | |
| "learning_rate": 9.091596743566297e-06, | |
| "loss": 0.3107, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.16819571865443425, | |
| "grad_norm": 3.8301804065704346, | |
| "learning_rate": 9.062916627655994e-06, | |
| "loss": 0.4904, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.1834862385321101, | |
| "grad_norm": 2.995657205581665, | |
| "learning_rate": 9.03423651174569e-06, | |
| "loss": 0.3564, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.19877675840978593, | |
| "grad_norm": 5.345709800720215, | |
| "learning_rate": 9.005556395835386e-06, | |
| "loss": 0.498, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.21406727828746178, | |
| "grad_norm": 1.6241793632507324, | |
| "learning_rate": 8.976876279925083e-06, | |
| "loss": 0.4324, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.22935779816513763, | |
| "grad_norm": 6.096868991851807, | |
| "learning_rate": 8.94819616401478e-06, | |
| "loss": 0.4474, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.24464831804281345, | |
| "grad_norm": 4.777408123016357, | |
| "learning_rate": 8.919516048104475e-06, | |
| "loss": 0.3605, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.2599388379204893, | |
| "grad_norm": 4.883125305175781, | |
| "learning_rate": 8.890835932194172e-06, | |
| "loss": 0.3694, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.27522935779816515, | |
| "grad_norm": 2.1104636192321777, | |
| "learning_rate": 8.862155816283867e-06, | |
| "loss": 0.3391, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.290519877675841, | |
| "grad_norm": 3.5193943977355957, | |
| "learning_rate": 8.833475700373564e-06, | |
| "loss": 0.4028, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.3058103975535168, | |
| "grad_norm": 12.695348739624023, | |
| "learning_rate": 8.80479558446326e-06, | |
| "loss": 0.5372, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.3211009174311927, | |
| "grad_norm": 8.019516944885254, | |
| "learning_rate": 8.776115468552956e-06, | |
| "loss": 0.3839, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.3363914373088685, | |
| "grad_norm": 2.723048448562622, | |
| "learning_rate": 8.747435352642653e-06, | |
| "loss": 0.3155, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.3516819571865443, | |
| "grad_norm": 5.75403356552124, | |
| "learning_rate": 8.718755236732348e-06, | |
| "loss": 0.4086, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.3669724770642202, | |
| "grad_norm": 3.325057029724121, | |
| "learning_rate": 8.690075120822045e-06, | |
| "loss": 0.293, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.382262996941896, | |
| "grad_norm": 18.065032958984375, | |
| "learning_rate": 8.66139500491174e-06, | |
| "loss": 0.3126, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.39755351681957185, | |
| "grad_norm": 5.72575569152832, | |
| "learning_rate": 8.632714889001437e-06, | |
| "loss": 0.3087, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.41284403669724773, | |
| "grad_norm": 5.2425055503845215, | |
| "learning_rate": 8.604034773091134e-06, | |
| "loss": 0.474, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.42813455657492355, | |
| "grad_norm": 8.025136947631836, | |
| "learning_rate": 8.57535465718083e-06, | |
| "loss": 0.2461, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.4434250764525994, | |
| "grad_norm": 11.76362133026123, | |
| "learning_rate": 8.546674541270526e-06, | |
| "loss": 0.3092, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.45871559633027525, | |
| "grad_norm": 14.632379531860352, | |
| "learning_rate": 8.517994425360223e-06, | |
| "loss": 0.4511, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.4740061162079511, | |
| "grad_norm": 3.7113358974456787, | |
| "learning_rate": 8.489314309449918e-06, | |
| "loss": 0.3188, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.4892966360856269, | |
| "grad_norm": 10.738109588623047, | |
| "learning_rate": 8.460634193539615e-06, | |
| "loss": 0.4563, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.5045871559633027, | |
| "grad_norm": 13.628973960876465, | |
| "learning_rate": 8.431954077629312e-06, | |
| "loss": 0.3233, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.5198776758409785, | |
| "grad_norm": 17.098241806030273, | |
| "learning_rate": 8.403273961719007e-06, | |
| "loss": 0.4134, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.5351681957186545, | |
| "grad_norm": 4.145214080810547, | |
| "learning_rate": 8.374593845808704e-06, | |
| "loss": 0.3638, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.5504587155963303, | |
| "grad_norm": 21.35369300842285, | |
| "learning_rate": 8.345913729898399e-06, | |
| "loss": 0.3341, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.5657492354740061, | |
| "grad_norm": 6.4279937744140625, | |
| "learning_rate": 8.317233613988096e-06, | |
| "loss": 0.3332, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.581039755351682, | |
| "grad_norm": 6.384445667266846, | |
| "learning_rate": 8.288553498077791e-06, | |
| "loss": 0.3299, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.5963302752293578, | |
| "grad_norm": 9.927398681640625, | |
| "learning_rate": 8.25987338216749e-06, | |
| "loss": 0.3687, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.6116207951070336, | |
| "grad_norm": 2.317152261734009, | |
| "learning_rate": 8.231193266257185e-06, | |
| "loss": 0.2947, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.6269113149847095, | |
| "grad_norm": 6.1938982009887695, | |
| "learning_rate": 8.202513150346881e-06, | |
| "loss": 0.2495, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.6422018348623854, | |
| "grad_norm": 9.2775297164917, | |
| "learning_rate": 8.173833034436577e-06, | |
| "loss": 0.3011, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.6574923547400612, | |
| "grad_norm": 12.099929809570312, | |
| "learning_rate": 8.145152918526273e-06, | |
| "loss": 0.2866, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.672782874617737, | |
| "grad_norm": 15.571517944335938, | |
| "learning_rate": 8.116472802615969e-06, | |
| "loss": 0.2352, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.6880733944954128, | |
| "grad_norm": 2.7042486667633057, | |
| "learning_rate": 8.087792686705666e-06, | |
| "loss": 0.4442, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.7033639143730887, | |
| "grad_norm": 24.800390243530273, | |
| "learning_rate": 8.059112570795362e-06, | |
| "loss": 0.2647, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.7186544342507645, | |
| "grad_norm": 1.1386603116989136, | |
| "learning_rate": 8.030432454885058e-06, | |
| "loss": 0.339, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.7339449541284404, | |
| "grad_norm": 2.4210026264190674, | |
| "learning_rate": 8.001752338974754e-06, | |
| "loss": 0.2963, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.7492354740061162, | |
| "grad_norm": 10.266980171203613, | |
| "learning_rate": 7.97307222306445e-06, | |
| "loss": 0.2177, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.764525993883792, | |
| "grad_norm": 8.048089027404785, | |
| "learning_rate": 7.944392107154146e-06, | |
| "loss": 0.3248, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.7798165137614679, | |
| "grad_norm": 25.07794189453125, | |
| "learning_rate": 7.915711991243843e-06, | |
| "loss": 0.3321, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.7951070336391437, | |
| "grad_norm": 25.960834503173828, | |
| "learning_rate": 7.88703187533354e-06, | |
| "loss": 0.4167, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.8103975535168195, | |
| "grad_norm": 7.0256028175354, | |
| "learning_rate": 7.858351759423235e-06, | |
| "loss": 0.2602, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.8256880733944955, | |
| "grad_norm": 3.3532984256744385, | |
| "learning_rate": 7.829671643512932e-06, | |
| "loss": 0.2569, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.8409785932721713, | |
| "grad_norm": 5.293972969055176, | |
| "learning_rate": 7.800991527602627e-06, | |
| "loss": 0.321, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.8562691131498471, | |
| "grad_norm": 5.642370223999023, | |
| "learning_rate": 7.772311411692324e-06, | |
| "loss": 0.2011, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.8715596330275229, | |
| "grad_norm": 19.925289154052734, | |
| "learning_rate": 7.743631295782021e-06, | |
| "loss": 0.3707, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.8868501529051988, | |
| "grad_norm": 7.974088668823242, | |
| "learning_rate": 7.714951179871716e-06, | |
| "loss": 0.3449, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.9021406727828746, | |
| "grad_norm": 28.756046295166016, | |
| "learning_rate": 7.686271063961413e-06, | |
| "loss": 0.5487, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.9174311926605505, | |
| "grad_norm": 2.515625238418579, | |
| "learning_rate": 7.657590948051108e-06, | |
| "loss": 0.3541, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.9327217125382263, | |
| "grad_norm": 7.654485702514648, | |
| "learning_rate": 7.628910832140805e-06, | |
| "loss": 0.3335, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.9480122324159022, | |
| "grad_norm": 17.905406951904297, | |
| "learning_rate": 7.600230716230501e-06, | |
| "loss": 0.3162, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.963302752293578, | |
| "grad_norm": 16.23491668701172, | |
| "learning_rate": 7.571550600320198e-06, | |
| "loss": 0.3626, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.9785932721712538, | |
| "grad_norm": 0.26539096236228943, | |
| "learning_rate": 7.542870484409894e-06, | |
| "loss": 0.2231, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.9938837920489296, | |
| "grad_norm": 1.7205305099487305, | |
| "learning_rate": 7.51419036849959e-06, | |
| "loss": 0.3941, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8706962509563887, | |
| "eval_f1": 0.7540029112081513, | |
| "eval_loss": 0.29941147565841675, | |
| "eval_precision": 0.6906666666666667, | |
| "eval_recall": 0.8301282051282052, | |
| "eval_runtime": 4.8981, | |
| "eval_samples_per_second": 266.841, | |
| "eval_steps_per_second": 16.741, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.0091743119266054, | |
| "grad_norm": 13.617008209228516, | |
| "learning_rate": 7.485510252589286e-06, | |
| "loss": 0.3849, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.0244648318042813, | |
| "grad_norm": 6.546511650085449, | |
| "learning_rate": 7.456830136678982e-06, | |
| "loss": 0.2871, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.039755351681957, | |
| "grad_norm": 5.344148635864258, | |
| "learning_rate": 7.428150020768678e-06, | |
| "loss": 0.2358, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.0550458715596331, | |
| "grad_norm": 7.6064348220825195, | |
| "learning_rate": 7.399469904858376e-06, | |
| "loss": 0.2451, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.070336391437309, | |
| "grad_norm": 21.806922912597656, | |
| "learning_rate": 7.370789788948072e-06, | |
| "loss": 0.2107, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.0856269113149848, | |
| "grad_norm": 10.673025131225586, | |
| "learning_rate": 7.342109673037768e-06, | |
| "loss": 0.2411, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.1009174311926606, | |
| "grad_norm": 14.94886589050293, | |
| "learning_rate": 7.313429557127464e-06, | |
| "loss": 0.1456, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.1162079510703364, | |
| "grad_norm": 12.618221282958984, | |
| "learning_rate": 7.28474944121716e-06, | |
| "loss": 0.2997, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.1314984709480123, | |
| "grad_norm": 0.2970161736011505, | |
| "learning_rate": 7.256069325306856e-06, | |
| "loss": 0.2922, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.146788990825688, | |
| "grad_norm": 3.0866048336029053, | |
| "learning_rate": 7.227389209396553e-06, | |
| "loss": 0.1767, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.162079510703364, | |
| "grad_norm": 12.53414535522461, | |
| "learning_rate": 7.198709093486249e-06, | |
| "loss": 0.2948, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.1773700305810397, | |
| "grad_norm": 24.84918975830078, | |
| "learning_rate": 7.170028977575945e-06, | |
| "loss": 0.1902, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.1926605504587156, | |
| "grad_norm": 1.028529167175293, | |
| "learning_rate": 7.141348861665641e-06, | |
| "loss": 0.2246, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.2079510703363914, | |
| "grad_norm": 20.015581130981445, | |
| "learning_rate": 7.112668745755337e-06, | |
| "loss": 0.2278, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.2232415902140672, | |
| "grad_norm": 12.658394813537598, | |
| "learning_rate": 7.083988629845033e-06, | |
| "loss": 0.3021, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.238532110091743, | |
| "grad_norm": 4.727680683135986, | |
| "learning_rate": 7.05530851393473e-06, | |
| "loss": 0.3192, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.2538226299694188, | |
| "grad_norm": 16.0307559967041, | |
| "learning_rate": 7.026628398024426e-06, | |
| "loss": 0.3534, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.2691131498470947, | |
| "grad_norm": 4.940106391906738, | |
| "learning_rate": 6.997948282114122e-06, | |
| "loss": 0.3564, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.2844036697247707, | |
| "grad_norm": 23.759117126464844, | |
| "learning_rate": 6.969268166203818e-06, | |
| "loss": 0.2746, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.2996941896024465, | |
| "grad_norm": 25.81449317932129, | |
| "learning_rate": 6.9405880502935144e-06, | |
| "loss": 0.4967, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.3149847094801224, | |
| "grad_norm": 4.882550239562988, | |
| "learning_rate": 6.9119079343832104e-06, | |
| "loss": 0.2423, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.3302752293577982, | |
| "grad_norm": 12.674756050109863, | |
| "learning_rate": 6.883227818472907e-06, | |
| "loss": 0.3532, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.345565749235474, | |
| "grad_norm": 13.883800506591797, | |
| "learning_rate": 6.854547702562603e-06, | |
| "loss": 0.2396, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.3608562691131498, | |
| "grad_norm": 7.71323299407959, | |
| "learning_rate": 6.825867586652299e-06, | |
| "loss": 0.2591, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.3761467889908257, | |
| "grad_norm": 2.174020767211914, | |
| "learning_rate": 6.797187470741995e-06, | |
| "loss": 0.2171, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.3914373088685015, | |
| "grad_norm": 2.351623773574829, | |
| "learning_rate": 6.768507354831691e-06, | |
| "loss": 0.3057, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.4067278287461773, | |
| "grad_norm": 26.668006896972656, | |
| "learning_rate": 6.739827238921387e-06, | |
| "loss": 0.2594, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.4220183486238533, | |
| "grad_norm": 13.497000694274902, | |
| "learning_rate": 6.711147123011084e-06, | |
| "loss": 0.2886, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.4373088685015292, | |
| "grad_norm": 9.161819458007812, | |
| "learning_rate": 6.682467007100781e-06, | |
| "loss": 0.2983, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.452599388379205, | |
| "grad_norm": 22.162334442138672, | |
| "learning_rate": 6.653786891190477e-06, | |
| "loss": 0.1644, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.4678899082568808, | |
| "grad_norm": 0.5806663632392883, | |
| "learning_rate": 6.625106775280173e-06, | |
| "loss": 0.2643, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.4831804281345566, | |
| "grad_norm": 0.31279271841049194, | |
| "learning_rate": 6.596426659369869e-06, | |
| "loss": 0.2888, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.4984709480122325, | |
| "grad_norm": 0.543879508972168, | |
| "learning_rate": 6.567746543459565e-06, | |
| "loss": 0.2786, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.5137614678899083, | |
| "grad_norm": 0.9113799333572388, | |
| "learning_rate": 6.539066427549262e-06, | |
| "loss": 0.4815, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.529051987767584, | |
| "grad_norm": 17.983909606933594, | |
| "learning_rate": 6.510386311638958e-06, | |
| "loss": 0.2022, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.54434250764526, | |
| "grad_norm": 3.3928322792053223, | |
| "learning_rate": 6.481706195728654e-06, | |
| "loss": 0.2222, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.5596330275229358, | |
| "grad_norm": 27.15311050415039, | |
| "learning_rate": 6.45302607981835e-06, | |
| "loss": 0.2879, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.5749235474006116, | |
| "grad_norm": 1.4941473007202148, | |
| "learning_rate": 6.424345963908046e-06, | |
| "loss": 0.3431, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.5902140672782874, | |
| "grad_norm": 26.38034439086914, | |
| "learning_rate": 6.395665847997742e-06, | |
| "loss": 0.2812, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.6055045871559632, | |
| "grad_norm": 4.923500061035156, | |
| "learning_rate": 6.366985732087439e-06, | |
| "loss": 0.2865, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.620795107033639, | |
| "grad_norm": 0.2041570544242859, | |
| "learning_rate": 6.338305616177135e-06, | |
| "loss": 0.4086, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.6360856269113149, | |
| "grad_norm": 1.037766456604004, | |
| "learning_rate": 6.309625500266832e-06, | |
| "loss": 0.1451, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.6513761467889907, | |
| "grad_norm": 9.758125305175781, | |
| "learning_rate": 6.280945384356528e-06, | |
| "loss": 0.3565, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "grad_norm": 0.7395672798156738, | |
| "learning_rate": 6.252265268446224e-06, | |
| "loss": 0.3725, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.6819571865443423, | |
| "grad_norm": 2.0258843898773193, | |
| "learning_rate": 6.223585152535921e-06, | |
| "loss": 0.1929, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.6972477064220184, | |
| "grad_norm": 6.11789083480835, | |
| "learning_rate": 6.194905036625617e-06, | |
| "loss": 0.3753, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.7125382262996942, | |
| "grad_norm": 11.606084823608398, | |
| "learning_rate": 6.166224920715313e-06, | |
| "loss": 0.227, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.72782874617737, | |
| "grad_norm": 17.951644897460938, | |
| "learning_rate": 6.137544804805009e-06, | |
| "loss": 0.4915, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.7431192660550459, | |
| "grad_norm": 0.44653165340423584, | |
| "learning_rate": 6.108864688894705e-06, | |
| "loss": 0.2273, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.7584097859327217, | |
| "grad_norm": 17.01247215270996, | |
| "learning_rate": 6.080184572984401e-06, | |
| "loss": 0.3209, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.7737003058103975, | |
| "grad_norm": 20.699323654174805, | |
| "learning_rate": 6.0515044570740976e-06, | |
| "loss": 0.435, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.7889908256880735, | |
| "grad_norm": 30.34465789794922, | |
| "learning_rate": 6.0228243411637936e-06, | |
| "loss": 0.2472, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.8042813455657494, | |
| "grad_norm": 4.261192798614502, | |
| "learning_rate": 5.99414422525349e-06, | |
| "loss": 0.285, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.8195718654434252, | |
| "grad_norm": 20.993389129638672, | |
| "learning_rate": 5.965464109343186e-06, | |
| "loss": 0.3737, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.834862385321101, | |
| "grad_norm": 0.8759846091270447, | |
| "learning_rate": 5.936783993432882e-06, | |
| "loss": 0.1866, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.8501529051987768, | |
| "grad_norm": 29.079099655151367, | |
| "learning_rate": 5.9081038775225785e-06, | |
| "loss": 0.4163, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.8654434250764527, | |
| "grad_norm": 3.231898546218872, | |
| "learning_rate": 5.879423761612275e-06, | |
| "loss": 0.4359, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.8807339449541285, | |
| "grad_norm": 17.104846954345703, | |
| "learning_rate": 5.850743645701971e-06, | |
| "loss": 0.2606, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.8960244648318043, | |
| "grad_norm": 0.409883588552475, | |
| "learning_rate": 5.822063529791667e-06, | |
| "loss": 0.1581, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.9113149847094801, | |
| "grad_norm": 10.206957817077637, | |
| "learning_rate": 5.793383413881363e-06, | |
| "loss": 0.2835, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.926605504587156, | |
| "grad_norm": 16.08922576904297, | |
| "learning_rate": 5.764703297971059e-06, | |
| "loss": 0.2338, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.9418960244648318, | |
| "grad_norm": 3.5700275897979736, | |
| "learning_rate": 5.736023182060755e-06, | |
| "loss": 0.4643, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.9571865443425076, | |
| "grad_norm": 13.477959632873535, | |
| "learning_rate": 5.707343066150452e-06, | |
| "loss": 0.2526, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.9724770642201834, | |
| "grad_norm": 8.30550479888916, | |
| "learning_rate": 5.678662950240148e-06, | |
| "loss": 0.4456, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.9877675840978593, | |
| "grad_norm": 7.599588871002197, | |
| "learning_rate": 5.649982834329844e-06, | |
| "loss": 0.259, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.881407804131599, | |
| "eval_f1": 0.7589424572317263, | |
| "eval_loss": 0.3265691101551056, | |
| "eval_precision": 0.7371601208459214, | |
| "eval_recall": 0.782051282051282, | |
| "eval_runtime": 4.9078, | |
| "eval_samples_per_second": 266.31, | |
| "eval_steps_per_second": 16.708, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 2.003058103975535, | |
| "grad_norm": 5.8759026527404785, | |
| "learning_rate": 5.62130271841954e-06, | |
| "loss": 0.2964, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.018348623853211, | |
| "grad_norm": 5.663475036621094, | |
| "learning_rate": 5.592622602509236e-06, | |
| "loss": 0.2355, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.0336391437308867, | |
| "grad_norm": 12.831345558166504, | |
| "learning_rate": 5.563942486598932e-06, | |
| "loss": 0.2225, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.0489296636085625, | |
| "grad_norm": 0.6970747709274292, | |
| "learning_rate": 5.53526237068863e-06, | |
| "loss": 0.279, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.0642201834862384, | |
| "grad_norm": 20.61145782470703, | |
| "learning_rate": 5.506582254778326e-06, | |
| "loss": 0.2825, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.079510703363914, | |
| "grad_norm": 4.7170891761779785, | |
| "learning_rate": 5.477902138868022e-06, | |
| "loss": 0.2996, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.09480122324159, | |
| "grad_norm": 8.943619728088379, | |
| "learning_rate": 5.449222022957718e-06, | |
| "loss": 0.1539, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.1100917431192663, | |
| "grad_norm": 0.7099681496620178, | |
| "learning_rate": 5.420541907047414e-06, | |
| "loss": 0.1958, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.1253822629969417, | |
| "grad_norm": 3.903538227081299, | |
| "learning_rate": 5.39186179113711e-06, | |
| "loss": 0.3434, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 2.140672782874618, | |
| "grad_norm": 4.8174920082092285, | |
| "learning_rate": 5.363181675226807e-06, | |
| "loss": 0.266, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.1559633027522938, | |
| "grad_norm": 13.861398696899414, | |
| "learning_rate": 5.334501559316503e-06, | |
| "loss": 0.226, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.1712538226299696, | |
| "grad_norm": 0.26603391766548157, | |
| "learning_rate": 5.305821443406199e-06, | |
| "loss": 0.1668, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 2.1865443425076454, | |
| "grad_norm": 8.830183982849121, | |
| "learning_rate": 5.277141327495895e-06, | |
| "loss": 0.1988, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 2.2018348623853212, | |
| "grad_norm": 0.7192776799201965, | |
| "learning_rate": 5.248461211585591e-06, | |
| "loss": 0.2137, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.217125382262997, | |
| "grad_norm": 24.46476173400879, | |
| "learning_rate": 5.219781095675287e-06, | |
| "loss": 0.3146, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.232415902140673, | |
| "grad_norm": 36.542945861816406, | |
| "learning_rate": 5.191100979764985e-06, | |
| "loss": 0.3024, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.2477064220183487, | |
| "grad_norm": 18.665542602539062, | |
| "learning_rate": 5.162420863854681e-06, | |
| "loss": 0.45, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.2629969418960245, | |
| "grad_norm": 0.5163823366165161, | |
| "learning_rate": 5.133740747944377e-06, | |
| "loss": 0.2063, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.2782874617737003, | |
| "grad_norm": 12.623830795288086, | |
| "learning_rate": 5.105060632034073e-06, | |
| "loss": 0.3537, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.293577981651376, | |
| "grad_norm": 6.660862922668457, | |
| "learning_rate": 5.076380516123769e-06, | |
| "loss": 0.3763, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.308868501529052, | |
| "grad_norm": 18.302440643310547, | |
| "learning_rate": 5.047700400213465e-06, | |
| "loss": 0.2796, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.324159021406728, | |
| "grad_norm": 5.233463764190674, | |
| "learning_rate": 5.019020284303162e-06, | |
| "loss": 0.2484, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.3394495412844036, | |
| "grad_norm": 0.2765680253505707, | |
| "learning_rate": 4.990340168392858e-06, | |
| "loss": 0.197, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.3547400611620795, | |
| "grad_norm": 7.559242248535156, | |
| "learning_rate": 4.961660052482554e-06, | |
| "loss": 0.2913, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.3700305810397553, | |
| "grad_norm": 20.358930587768555, | |
| "learning_rate": 4.93297993657225e-06, | |
| "loss": 0.2762, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.385321100917431, | |
| "grad_norm": 23.49895668029785, | |
| "learning_rate": 4.904299820661946e-06, | |
| "loss": 0.3327, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.400611620795107, | |
| "grad_norm": 9.241127014160156, | |
| "learning_rate": 4.875619704751642e-06, | |
| "loss": 0.2981, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.4159021406727827, | |
| "grad_norm": 9.627830505371094, | |
| "learning_rate": 4.846939588841339e-06, | |
| "loss": 0.3052, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.4311926605504586, | |
| "grad_norm": 6.3946075439453125, | |
| "learning_rate": 4.818259472931035e-06, | |
| "loss": 0.1022, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.4464831804281344, | |
| "grad_norm": 0.6927328109741211, | |
| "learning_rate": 4.789579357020731e-06, | |
| "loss": 0.0922, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.46177370030581, | |
| "grad_norm": 1.498061180114746, | |
| "learning_rate": 4.760899241110427e-06, | |
| "loss": 0.1883, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 2.477064220183486, | |
| "grad_norm": 3.1200311183929443, | |
| "learning_rate": 4.732219125200123e-06, | |
| "loss": 0.1963, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.4923547400611623, | |
| "grad_norm": 5.31001615524292, | |
| "learning_rate": 4.703539009289819e-06, | |
| "loss": 0.3003, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 2.5076452599388377, | |
| "grad_norm": 13.968355178833008, | |
| "learning_rate": 4.674858893379516e-06, | |
| "loss": 0.3723, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 2.522935779816514, | |
| "grad_norm": 30.514612197875977, | |
| "learning_rate": 4.646178777469212e-06, | |
| "loss": 0.2453, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.5382262996941893, | |
| "grad_norm": 18.336116790771484, | |
| "learning_rate": 4.617498661558908e-06, | |
| "loss": 0.2514, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 2.5535168195718656, | |
| "grad_norm": 3.3345324993133545, | |
| "learning_rate": 4.588818545648604e-06, | |
| "loss": 0.1784, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 2.5688073394495414, | |
| "grad_norm": 0.6828510165214539, | |
| "learning_rate": 4.560138429738301e-06, | |
| "loss": 0.1926, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 2.5840978593272173, | |
| "grad_norm": 0.3491632044315338, | |
| "learning_rate": 4.531458313827997e-06, | |
| "loss": 0.3459, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 2.599388379204893, | |
| "grad_norm": 14.196633338928223, | |
| "learning_rate": 4.502778197917693e-06, | |
| "loss": 0.2798, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.614678899082569, | |
| "grad_norm": 3.067326307296753, | |
| "learning_rate": 4.47409808200739e-06, | |
| "loss": 0.2548, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 2.6299694189602447, | |
| "grad_norm": 10.058695793151855, | |
| "learning_rate": 4.445417966097086e-06, | |
| "loss": 0.3088, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 2.6452599388379205, | |
| "grad_norm": 24.748844146728516, | |
| "learning_rate": 4.416737850186782e-06, | |
| "loss": 0.2524, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 2.6605504587155964, | |
| "grad_norm": 9.57737922668457, | |
| "learning_rate": 4.388057734276478e-06, | |
| "loss": 0.1089, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 2.675840978593272, | |
| "grad_norm": 4.255125999450684, | |
| "learning_rate": 4.359377618366174e-06, | |
| "loss": 0.2619, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.691131498470948, | |
| "grad_norm": 12.66395378112793, | |
| "learning_rate": 4.33069750245587e-06, | |
| "loss": 0.113, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 2.706422018348624, | |
| "grad_norm": 27.81719970703125, | |
| "learning_rate": 4.302017386545567e-06, | |
| "loss": 0.1587, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 2.7217125382262997, | |
| "grad_norm": 9.098254203796387, | |
| "learning_rate": 4.273337270635263e-06, | |
| "loss": 0.1325, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 2.7370030581039755, | |
| "grad_norm": 7.441411018371582, | |
| "learning_rate": 4.244657154724959e-06, | |
| "loss": 0.4097, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 2.7522935779816513, | |
| "grad_norm": 28.30503273010254, | |
| "learning_rate": 4.215977038814656e-06, | |
| "loss": 0.2826, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.767584097859327, | |
| "grad_norm": 47.75388717651367, | |
| "learning_rate": 4.187296922904352e-06, | |
| "loss": 0.2992, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 2.782874617737003, | |
| "grad_norm": 5.235270023345947, | |
| "learning_rate": 4.158616806994048e-06, | |
| "loss": 0.1981, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 2.7981651376146788, | |
| "grad_norm": 0.08484458923339844, | |
| "learning_rate": 4.129936691083745e-06, | |
| "loss": 0.3315, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 2.8134556574923546, | |
| "grad_norm": 17.93990707397461, | |
| "learning_rate": 4.101256575173441e-06, | |
| "loss": 0.3326, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 2.8287461773700304, | |
| "grad_norm": 0.8848928213119507, | |
| "learning_rate": 4.072576459263137e-06, | |
| "loss": 0.1852, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.8440366972477067, | |
| "grad_norm": 1.0666664838790894, | |
| "learning_rate": 4.043896343352833e-06, | |
| "loss": 0.1235, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 2.859327217125382, | |
| "grad_norm": 29.053184509277344, | |
| "learning_rate": 4.015216227442529e-06, | |
| "loss": 0.309, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 2.8746177370030583, | |
| "grad_norm": 30.479965209960938, | |
| "learning_rate": 3.986536111532225e-06, | |
| "loss": 0.2986, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 2.8899082568807337, | |
| "grad_norm": 23.085195541381836, | |
| "learning_rate": 3.957855995621922e-06, | |
| "loss": 0.1596, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 2.90519877675841, | |
| "grad_norm": 1.5042694807052612, | |
| "learning_rate": 3.929175879711618e-06, | |
| "loss": 0.2213, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.9204892966360854, | |
| "grad_norm": 22.314062118530273, | |
| "learning_rate": 3.900495763801314e-06, | |
| "loss": 0.1854, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 2.9357798165137616, | |
| "grad_norm": 3.0091068744659424, | |
| "learning_rate": 3.8718156478910105e-06, | |
| "loss": 0.1231, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 2.9510703363914375, | |
| "grad_norm": 0.4003874361515045, | |
| "learning_rate": 3.8431355319807065e-06, | |
| "loss": 0.1013, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 2.9663608562691133, | |
| "grad_norm": 36.91960906982422, | |
| "learning_rate": 3.8144554160704025e-06, | |
| "loss": 0.1974, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 2.981651376146789, | |
| "grad_norm": 10.349553108215332, | |
| "learning_rate": 3.785775300160099e-06, | |
| "loss": 0.3974, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.996941896024465, | |
| "grad_norm": 4.602114677429199, | |
| "learning_rate": 3.757095184249795e-06, | |
| "loss": 0.1764, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8837031369548585, | |
| "eval_f1": 0.7661538461538462, | |
| "eval_loss": 0.416372686624527, | |
| "eval_precision": 0.7366863905325444, | |
| "eval_recall": 0.7980769230769231, | |
| "eval_runtime": 4.9368, | |
| "eval_samples_per_second": 264.749, | |
| "eval_steps_per_second": 16.61, | |
| "step": 1962 | |
| }, | |
| { | |
| "epoch": 3.0122324159021407, | |
| "grad_norm": 0.9072074890136719, | |
| "learning_rate": 3.728415068339491e-06, | |
| "loss": 0.0342, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 3.0275229357798166, | |
| "grad_norm": 11.377812385559082, | |
| "learning_rate": 3.699734952429188e-06, | |
| "loss": 0.1661, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 3.0428134556574924, | |
| "grad_norm": 17.510143280029297, | |
| "learning_rate": 3.671054836518884e-06, | |
| "loss": 0.1267, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 3.058103975535168, | |
| "grad_norm": 0.12311755865812302, | |
| "learning_rate": 3.64237472060858e-06, | |
| "loss": 0.272, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.073394495412844, | |
| "grad_norm": 7.080625534057617, | |
| "learning_rate": 3.6136946046982763e-06, | |
| "loss": 0.3724, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 3.08868501529052, | |
| "grad_norm": 15.893166542053223, | |
| "learning_rate": 3.5850144887879723e-06, | |
| "loss": 0.211, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 3.1039755351681957, | |
| "grad_norm": 70.26548767089844, | |
| "learning_rate": 3.5563343728776683e-06, | |
| "loss": 0.1047, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 3.1192660550458715, | |
| "grad_norm": 1.035912036895752, | |
| "learning_rate": 3.527654256967365e-06, | |
| "loss": 0.0586, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 3.1345565749235473, | |
| "grad_norm": 21.898548126220703, | |
| "learning_rate": 3.498974141057061e-06, | |
| "loss": 0.2289, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.149847094801223, | |
| "grad_norm": 9.897711753845215, | |
| "learning_rate": 3.4702940251467572e-06, | |
| "loss": 0.2056, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 3.165137614678899, | |
| "grad_norm": 0.49520930647850037, | |
| "learning_rate": 3.4416139092364537e-06, | |
| "loss": 0.2293, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 3.180428134556575, | |
| "grad_norm": 0.14918214082717896, | |
| "learning_rate": 3.4129337933261497e-06, | |
| "loss": 0.1414, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 3.1957186544342506, | |
| "grad_norm": 0.4814736247062683, | |
| "learning_rate": 3.3842536774158457e-06, | |
| "loss": 0.2611, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 3.2110091743119265, | |
| "grad_norm": 0.19969391822814941, | |
| "learning_rate": 3.355573561505542e-06, | |
| "loss": 0.1482, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.2262996941896023, | |
| "grad_norm": 0.16247302293777466, | |
| "learning_rate": 3.3268934455952385e-06, | |
| "loss": 0.4237, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 3.241590214067278, | |
| "grad_norm": 17.90416717529297, | |
| "learning_rate": 3.2982133296849346e-06, | |
| "loss": 0.127, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 3.2568807339449544, | |
| "grad_norm": 34.14628219604492, | |
| "learning_rate": 3.269533213774631e-06, | |
| "loss": 0.1481, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 3.2721712538226297, | |
| "grad_norm": 0.18314845860004425, | |
| "learning_rate": 3.240853097864327e-06, | |
| "loss": 0.1436, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 3.287461773700306, | |
| "grad_norm": 0.13257867097854614, | |
| "learning_rate": 3.212172981954023e-06, | |
| "loss": 0.1203, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.302752293577982, | |
| "grad_norm": 42.67006301879883, | |
| "learning_rate": 3.1834928660437194e-06, | |
| "loss": 0.2017, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 3.3180428134556577, | |
| "grad_norm": 0.3060612082481384, | |
| "learning_rate": 3.154812750133416e-06, | |
| "loss": 0.1398, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 3.3333333333333335, | |
| "grad_norm": 6.72953987121582, | |
| "learning_rate": 3.126132634223112e-06, | |
| "loss": 0.1634, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 3.3486238532110093, | |
| "grad_norm": 0.4008305072784424, | |
| "learning_rate": 3.0974525183128083e-06, | |
| "loss": 0.0838, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 3.363914373088685, | |
| "grad_norm": 0.2475535124540329, | |
| "learning_rate": 3.0687724024025043e-06, | |
| "loss": 0.2601, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.379204892966361, | |
| "grad_norm": 0.6882356405258179, | |
| "learning_rate": 3.0400922864922004e-06, | |
| "loss": 0.2079, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 3.3944954128440368, | |
| "grad_norm": 0.11873319745063782, | |
| "learning_rate": 3.0114121705818968e-06, | |
| "loss": 0.1479, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 3.4097859327217126, | |
| "grad_norm": 2.4591195583343506, | |
| "learning_rate": 2.982732054671593e-06, | |
| "loss": 0.0835, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 3.4250764525993884, | |
| "grad_norm": 33.77700424194336, | |
| "learning_rate": 2.9540519387612892e-06, | |
| "loss": 0.325, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 3.4403669724770642, | |
| "grad_norm": 0.19437803328037262, | |
| "learning_rate": 2.9253718228509857e-06, | |
| "loss": 0.147, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.45565749235474, | |
| "grad_norm": 5.968377113342285, | |
| "learning_rate": 2.8966917069406817e-06, | |
| "loss": 0.3554, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 3.470948012232416, | |
| "grad_norm": 12.378327369689941, | |
| "learning_rate": 2.8680115910303777e-06, | |
| "loss": 0.2007, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 3.4862385321100917, | |
| "grad_norm": 0.21296323835849762, | |
| "learning_rate": 2.839331475120074e-06, | |
| "loss": 0.2296, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 3.5015290519877675, | |
| "grad_norm": 13.099735260009766, | |
| "learning_rate": 2.81065135920977e-06, | |
| "loss": 0.1264, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 3.5168195718654434, | |
| "grad_norm": 0.31538063287734985, | |
| "learning_rate": 2.781971243299466e-06, | |
| "loss": 0.4666, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.532110091743119, | |
| "grad_norm": 12.299838066101074, | |
| "learning_rate": 2.753291127389163e-06, | |
| "loss": 0.2505, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 3.547400611620795, | |
| "grad_norm": 16.644168853759766, | |
| "learning_rate": 2.724611011478859e-06, | |
| "loss": 0.4427, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 3.562691131498471, | |
| "grad_norm": 39.950279235839844, | |
| "learning_rate": 2.695930895568555e-06, | |
| "loss": 0.3459, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 3.5779816513761467, | |
| "grad_norm": 5.1495256423950195, | |
| "learning_rate": 2.6672507796582515e-06, | |
| "loss": 0.3096, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 3.5932721712538225, | |
| "grad_norm": 35.178226470947266, | |
| "learning_rate": 2.6385706637479475e-06, | |
| "loss": 0.2467, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.6085626911314987, | |
| "grad_norm": 6.108614921569824, | |
| "learning_rate": 2.6098905478376435e-06, | |
| "loss": 0.0156, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 3.623853211009174, | |
| "grad_norm": 0.34197351336479187, | |
| "learning_rate": 2.5812104319273403e-06, | |
| "loss": 0.1566, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 3.6391437308868504, | |
| "grad_norm": 5.665993690490723, | |
| "learning_rate": 2.5525303160170364e-06, | |
| "loss": 0.268, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 3.6544342507645258, | |
| "grad_norm": 3.0416011810302734, | |
| "learning_rate": 2.5238502001067324e-06, | |
| "loss": 0.1028, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 3.669724770642202, | |
| "grad_norm": 5.079551696777344, | |
| "learning_rate": 2.495170084196429e-06, | |
| "loss": 0.2225, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.6850152905198774, | |
| "grad_norm": 0.19564682245254517, | |
| "learning_rate": 2.466489968286125e-06, | |
| "loss": 0.1261, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 3.7003058103975537, | |
| "grad_norm": 15.542204856872559, | |
| "learning_rate": 2.437809852375821e-06, | |
| "loss": 0.1688, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 3.7155963302752295, | |
| "grad_norm": 0.636468231678009, | |
| "learning_rate": 2.4091297364655177e-06, | |
| "loss": 0.3542, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 3.7308868501529053, | |
| "grad_norm": 0.26796048879623413, | |
| "learning_rate": 2.3804496205552137e-06, | |
| "loss": 0.0689, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 3.746177370030581, | |
| "grad_norm": 9.50084400177002, | |
| "learning_rate": 2.3517695046449097e-06, | |
| "loss": 0.3902, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.761467889908257, | |
| "grad_norm": 1.7848141193389893, | |
| "learning_rate": 2.323089388734606e-06, | |
| "loss": 0.0613, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 3.776758409785933, | |
| "grad_norm": 20.41553497314453, | |
| "learning_rate": 2.294409272824302e-06, | |
| "loss": 0.2541, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 3.7920489296636086, | |
| "grad_norm": 1.6424323320388794, | |
| "learning_rate": 2.2657291569139986e-06, | |
| "loss": 0.2198, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 3.8073394495412844, | |
| "grad_norm": 0.24343428015708923, | |
| "learning_rate": 2.237049041003695e-06, | |
| "loss": 0.2029, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 3.8226299694189603, | |
| "grad_norm": 44.71451950073242, | |
| "learning_rate": 2.208368925093391e-06, | |
| "loss": 0.3573, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.837920489296636, | |
| "grad_norm": 30.519176483154297, | |
| "learning_rate": 2.179688809183087e-06, | |
| "loss": 0.3304, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 3.853211009174312, | |
| "grad_norm": 25.292348861694336, | |
| "learning_rate": 2.1510086932727835e-06, | |
| "loss": 0.1079, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 3.8685015290519877, | |
| "grad_norm": 35.308006286621094, | |
| "learning_rate": 2.1223285773624795e-06, | |
| "loss": 0.1774, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 3.8837920489296636, | |
| "grad_norm": 0.16474059224128723, | |
| "learning_rate": 2.093648461452176e-06, | |
| "loss": 0.0328, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 3.8990825688073394, | |
| "grad_norm": 0.09989167749881744, | |
| "learning_rate": 2.0649683455418724e-06, | |
| "loss": 0.1969, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.914373088685015, | |
| "grad_norm": 18.91350746154785, | |
| "learning_rate": 2.0362882296315684e-06, | |
| "loss": 0.1445, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 3.929663608562691, | |
| "grad_norm": 0.08901531994342804, | |
| "learning_rate": 2.0076081137212644e-06, | |
| "loss": 0.3773, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 3.944954128440367, | |
| "grad_norm": 46.0722770690918, | |
| "learning_rate": 1.978927997810961e-06, | |
| "loss": 0.3817, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 3.9602446483180427, | |
| "grad_norm": 7.898420810699463, | |
| "learning_rate": 1.950247881900657e-06, | |
| "loss": 0.1688, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 3.9755351681957185, | |
| "grad_norm": 51.515220642089844, | |
| "learning_rate": 1.9215677659903533e-06, | |
| "loss": 0.1521, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.9908256880733948, | |
| "grad_norm": 0.9271498918533325, | |
| "learning_rate": 1.8928876500800495e-06, | |
| "loss": 0.1549, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.8859984697781178, | |
| "eval_f1": 0.7585089141004863, | |
| "eval_loss": 0.44139671325683594, | |
| "eval_precision": 0.7672131147540984, | |
| "eval_recall": 0.75, | |
| "eval_runtime": 4.9043, | |
| "eval_samples_per_second": 266.502, | |
| "eval_steps_per_second": 16.72, | |
| "step": 2616 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 3270, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2764329675153408.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": { | |
| "learning_rate": 9.378397902669336e-06, | |
| "num_train_epochs": 5, | |
| "per_device_train_batch_size": 8, | |
| "seed": 28 | |
| } | |
| } | |