| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.8008673653776652, | |
| "global_step": 7750, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 13.863856315612793, | |
| "epoch": 0.0, | |
| "learning_rate": 1.5178894109143475e-07, | |
| "loss": 14.3579, | |
| "step": 10, | |
| "task_loss": 5.79541015625 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 15.529892921447754, | |
| "epoch": 0.01, | |
| "learning_rate": 3.6863028550777017e-07, | |
| "loss": 14.5251, | |
| "step": 20, | |
| "task_loss": 5.82769775390625 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 15.667880058288574, | |
| "epoch": 0.01, | |
| "learning_rate": 5.637874954824719e-07, | |
| "loss": 14.2874, | |
| "step": 30, | |
| "task_loss": 5.782501220703125 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 16.32543182373047, | |
| "epoch": 0.01, | |
| "learning_rate": 7.806288398988074e-07, | |
| "loss": 14.482, | |
| "step": 40, | |
| "task_loss": 5.93621826171875 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 15.878634452819824, | |
| "epoch": 0.02, | |
| "learning_rate": 9.974701843151429e-07, | |
| "loss": 14.0581, | |
| "step": 50, | |
| "task_loss": 5.87786865234375 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 15.618017196655273, | |
| "epoch": 0.02, | |
| "learning_rate": 1.214311528731478e-06, | |
| "loss": 14.4649, | |
| "step": 60, | |
| "task_loss": 5.95428466796875 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 14.689329147338867, | |
| "epoch": 0.03, | |
| "learning_rate": 1.4311528731478136e-06, | |
| "loss": 14.1966, | |
| "step": 70, | |
| "task_loss": 5.9617919921875 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 15.500862121582031, | |
| "epoch": 0.03, | |
| "learning_rate": 1.647994217564149e-06, | |
| "loss": 14.1659, | |
| "step": 80, | |
| "task_loss": 5.9317626953125 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 15.836941719055176, | |
| "epoch": 0.03, | |
| "learning_rate": 1.8648355619804843e-06, | |
| "loss": 14.0619, | |
| "step": 90, | |
| "task_loss": 5.79229736328125 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 15.45178508758545, | |
| "epoch": 0.04, | |
| "learning_rate": 2.08167690639682e-06, | |
| "loss": 13.9798, | |
| "step": 100, | |
| "task_loss": 5.859405517578125 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 14.564159393310547, | |
| "epoch": 0.04, | |
| "learning_rate": 2.298518250813155e-06, | |
| "loss": 14.0794, | |
| "step": 110, | |
| "task_loss": 5.789794921875 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 13.552111625671387, | |
| "epoch": 0.04, | |
| "learning_rate": 2.5153595952294906e-06, | |
| "loss": 13.8458, | |
| "step": 120, | |
| "task_loss": 5.688568115234375 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 14.749038696289062, | |
| "epoch": 0.05, | |
| "learning_rate": 2.732200939645826e-06, | |
| "loss": 13.8899, | |
| "step": 130, | |
| "task_loss": 5.49835205078125 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 14.9246826171875, | |
| "epoch": 0.05, | |
| "learning_rate": 2.9490422840621613e-06, | |
| "loss": 13.7831, | |
| "step": 140, | |
| "task_loss": 5.456787109375 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 14.531848907470703, | |
| "epoch": 0.05, | |
| "learning_rate": 3.1658836284784965e-06, | |
| "loss": 13.4898, | |
| "step": 150, | |
| "task_loss": 5.613800048828125 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 15.012535095214844, | |
| "epoch": 0.06, | |
| "learning_rate": 3.382724972894832e-06, | |
| "loss": 13.6949, | |
| "step": 160, | |
| "task_loss": 5.2825927734375 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 14.039583206176758, | |
| "epoch": 0.06, | |
| "learning_rate": 3.5995663173111676e-06, | |
| "loss": 13.182, | |
| "step": 170, | |
| "task_loss": 5.353515625 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 13.0057373046875, | |
| "epoch": 0.07, | |
| "learning_rate": 3.816407661727503e-06, | |
| "loss": 12.6947, | |
| "step": 180, | |
| "task_loss": 5.157745361328125 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 14.364778518676758, | |
| "epoch": 0.07, | |
| "learning_rate": 4.033249006143838e-06, | |
| "loss": 12.8377, | |
| "step": 190, | |
| "task_loss": 5.151580810546875 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 12.603987693786621, | |
| "epoch": 0.07, | |
| "learning_rate": 4.2500903505601735e-06, | |
| "loss": 12.5114, | |
| "step": 200, | |
| "task_loss": 5.133880615234375 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 13.770363807678223, | |
| "epoch": 0.08, | |
| "learning_rate": 4.466931694976509e-06, | |
| "loss": 12.571, | |
| "step": 210, | |
| "task_loss": 4.792236328125 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 12.380254745483398, | |
| "epoch": 0.08, | |
| "learning_rate": 4.683773039392844e-06, | |
| "loss": 12.1934, | |
| "step": 220, | |
| "task_loss": 4.9013671875 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 12.504887580871582, | |
| "epoch": 0.08, | |
| "learning_rate": 4.900614383809179e-06, | |
| "loss": 12.2029, | |
| "step": 230, | |
| "task_loss": 4.988494873046875 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 12.855850219726562, | |
| "epoch": 0.09, | |
| "learning_rate": 5.117455728225516e-06, | |
| "loss": 11.5489, | |
| "step": 240, | |
| "task_loss": 4.76690673828125 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 11.811483383178711, | |
| "epoch": 0.09, | |
| "learning_rate": 5.3342970726418505e-06, | |
| "loss": 11.297, | |
| "step": 250, | |
| "task_loss": 4.54803466796875 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "eval_exact_match": 2.7341532639545885, | |
| "eval_f1": 12.629112049670779, | |
| "step": 250 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 11.374210357666016, | |
| "epoch": 0.09, | |
| "learning_rate": 5.551138417058186e-06, | |
| "loss": 10.9207, | |
| "step": 260, | |
| "task_loss": 4.5309906005859375 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 10.759222030639648, | |
| "epoch": 0.1, | |
| "learning_rate": 5.767979761474522e-06, | |
| "loss": 10.4359, | |
| "step": 270, | |
| "task_loss": 4.0685882568359375 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 10.636564254760742, | |
| "epoch": 0.1, | |
| "learning_rate": 5.984821105890856e-06, | |
| "loss": 9.9935, | |
| "step": 280, | |
| "task_loss": 4.374786376953125 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 10.036893844604492, | |
| "epoch": 0.1, | |
| "learning_rate": 6.201662450307192e-06, | |
| "loss": 9.4053, | |
| "step": 290, | |
| "task_loss": 3.8459930419921875 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 9.233875274658203, | |
| "epoch": 0.11, | |
| "learning_rate": 6.4185037947235275e-06, | |
| "loss": 9.0732, | |
| "step": 300, | |
| "task_loss": 3.669677734375 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 8.439445495605469, | |
| "epoch": 0.11, | |
| "learning_rate": 6.635345139139862e-06, | |
| "loss": 8.6626, | |
| "step": 310, | |
| "task_loss": 3.1441650390625 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 8.574735641479492, | |
| "epoch": 0.12, | |
| "learning_rate": 6.852186483556198e-06, | |
| "loss": 8.297, | |
| "step": 320, | |
| "task_loss": 3.4480514526367188 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 8.304073333740234, | |
| "epoch": 0.12, | |
| "learning_rate": 7.069027827972534e-06, | |
| "loss": 8.0954, | |
| "step": 330, | |
| "task_loss": 3.2958412170410156 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 7.721191883087158, | |
| "epoch": 0.12, | |
| "learning_rate": 7.285869172388869e-06, | |
| "loss": 7.3317, | |
| "step": 340, | |
| "task_loss": 3.191631317138672 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 7.107563495635986, | |
| "epoch": 0.13, | |
| "learning_rate": 7.5027105168052046e-06, | |
| "loss": 7.0177, | |
| "step": 350, | |
| "task_loss": 3.102968215942383 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 7.762242794036865, | |
| "epoch": 0.13, | |
| "learning_rate": 7.71955186122154e-06, | |
| "loss": 6.7813, | |
| "step": 360, | |
| "task_loss": 4.048648834228516 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 6.884188652038574, | |
| "epoch": 0.13, | |
| "learning_rate": 7.936393205637875e-06, | |
| "loss": 6.6125, | |
| "step": 370, | |
| "task_loss": 2.772972583770752 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 6.817141056060791, | |
| "epoch": 0.14, | |
| "learning_rate": 8.15323455005421e-06, | |
| "loss": 6.1752, | |
| "step": 380, | |
| "task_loss": 3.29545259475708 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 6.7478532791137695, | |
| "epoch": 0.14, | |
| "learning_rate": 8.370075894470546e-06, | |
| "loss": 5.9283, | |
| "step": 390, | |
| "task_loss": 2.940948486328125 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 6.400359153747559, | |
| "epoch": 0.14, | |
| "learning_rate": 8.586917238886882e-06, | |
| "loss": 5.7384, | |
| "step": 400, | |
| "task_loss": 3.4309067726135254 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 5.582571983337402, | |
| "epoch": 0.15, | |
| "learning_rate": 8.803758583303217e-06, | |
| "loss": 5.6473, | |
| "step": 410, | |
| "task_loss": 2.701523780822754 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 5.448908805847168, | |
| "epoch": 0.15, | |
| "learning_rate": 9.020599927719551e-06, | |
| "loss": 5.7486, | |
| "step": 420, | |
| "task_loss": 3.1738014221191406 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 6.087796211242676, | |
| "epoch": 0.16, | |
| "learning_rate": 9.237441272135887e-06, | |
| "loss": 5.4961, | |
| "step": 430, | |
| "task_loss": 3.124908447265625 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 5.375993728637695, | |
| "epoch": 0.16, | |
| "learning_rate": 9.454282616552222e-06, | |
| "loss": 5.2309, | |
| "step": 440, | |
| "task_loss": 2.6864736080169678 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 5.898089408874512, | |
| "epoch": 0.16, | |
| "learning_rate": 9.671123960968558e-06, | |
| "loss": 5.3107, | |
| "step": 450, | |
| "task_loss": 2.6333131790161133 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 5.690750598907471, | |
| "epoch": 0.17, | |
| "learning_rate": 9.887965305384893e-06, | |
| "loss": 5.0992, | |
| "step": 460, | |
| "task_loss": 2.886949062347412 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 5.3047871589660645, | |
| "epoch": 0.17, | |
| "learning_rate": 1.010480664980123e-05, | |
| "loss": 5.1496, | |
| "step": 470, | |
| "task_loss": 2.6398158073425293 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 5.096426010131836, | |
| "epoch": 0.17, | |
| "learning_rate": 1.0321647994217564e-05, | |
| "loss": 4.888, | |
| "step": 480, | |
| "task_loss": 3.002800941467285 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 5.322364807128906, | |
| "epoch": 0.18, | |
| "learning_rate": 1.05384893386339e-05, | |
| "loss": 4.926, | |
| "step": 490, | |
| "task_loss": 2.943061590194702 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 5.079683303833008, | |
| "epoch": 0.18, | |
| "learning_rate": 1.0755330683050236e-05, | |
| "loss": 4.7843, | |
| "step": 500, | |
| "task_loss": 2.4856796264648438 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "eval_exact_match": 47.6631977294229, | |
| "eval_f1": 61.84689508471956, | |
| "step": 500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.29311466217041, | |
| "epoch": 0.18, | |
| "learning_rate": 1.0972172027466571e-05, | |
| "loss": 4.6627, | |
| "step": 510, | |
| "task_loss": 2.2312231063842773 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.374029159545898, | |
| "epoch": 0.19, | |
| "learning_rate": 1.1189013371882907e-05, | |
| "loss": 4.7554, | |
| "step": 520, | |
| "task_loss": 1.571474313735962 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 5.325689315795898, | |
| "epoch": 0.19, | |
| "learning_rate": 1.1405854716299242e-05, | |
| "loss": 4.5999, | |
| "step": 530, | |
| "task_loss": 2.791367530822754 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.6214494705200195, | |
| "epoch": 0.2, | |
| "learning_rate": 1.1622696060715576e-05, | |
| "loss": 4.3316, | |
| "step": 540, | |
| "task_loss": 2.6126604080200195 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.899660110473633, | |
| "epoch": 0.2, | |
| "learning_rate": 1.1839537405131912e-05, | |
| "loss": 4.5563, | |
| "step": 550, | |
| "task_loss": 2.572834014892578 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.496711730957031, | |
| "epoch": 0.2, | |
| "learning_rate": 1.2056378749548247e-05, | |
| "loss": 4.277, | |
| "step": 560, | |
| "task_loss": 2.6678249835968018 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.555987358093262, | |
| "epoch": 0.21, | |
| "learning_rate": 1.2273220093964583e-05, | |
| "loss": 4.3194, | |
| "step": 570, | |
| "task_loss": 2.6166036128997803 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.9610090255737305, | |
| "epoch": 0.21, | |
| "learning_rate": 1.2490061438380918e-05, | |
| "loss": 4.2434, | |
| "step": 580, | |
| "task_loss": 2.6043477058410645 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.213617324829102, | |
| "epoch": 0.21, | |
| "learning_rate": 1.2706902782797254e-05, | |
| "loss": 4.273, | |
| "step": 590, | |
| "task_loss": 1.856445550918579 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.79631233215332, | |
| "epoch": 0.22, | |
| "learning_rate": 1.2923744127213588e-05, | |
| "loss": 4.3605, | |
| "step": 600, | |
| "task_loss": 2.3547120094299316 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.130844593048096, | |
| "epoch": 0.22, | |
| "learning_rate": 1.3140585471629924e-05, | |
| "loss": 4.1307, | |
| "step": 610, | |
| "task_loss": 2.737905979156494 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.68433952331543, | |
| "epoch": 0.22, | |
| "learning_rate": 1.3357426816046259e-05, | |
| "loss": 4.228, | |
| "step": 620, | |
| "task_loss": 2.0046231746673584 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.9693734645843506, | |
| "epoch": 0.23, | |
| "learning_rate": 1.3574268160462595e-05, | |
| "loss": 4.0317, | |
| "step": 630, | |
| "task_loss": 1.786435604095459 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.474964618682861, | |
| "epoch": 0.23, | |
| "learning_rate": 1.379110950487893e-05, | |
| "loss": 3.9845, | |
| "step": 640, | |
| "task_loss": 2.3881354331970215 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.220601558685303, | |
| "epoch": 0.23, | |
| "learning_rate": 1.4007950849295268e-05, | |
| "loss": 4.126, | |
| "step": 650, | |
| "task_loss": 2.3482580184936523 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.762381553649902, | |
| "epoch": 0.24, | |
| "learning_rate": 1.4224792193711601e-05, | |
| "loss": 3.9768, | |
| "step": 660, | |
| "task_loss": 2.4734535217285156 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.62750244140625, | |
| "epoch": 0.24, | |
| "learning_rate": 1.4441633538127937e-05, | |
| "loss": 3.7916, | |
| "step": 670, | |
| "task_loss": 1.8247270584106445 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.762209415435791, | |
| "epoch": 0.25, | |
| "learning_rate": 1.4658474882544273e-05, | |
| "loss": 3.7097, | |
| "step": 680, | |
| "task_loss": 1.700040578842163 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.5018386840820312, | |
| "epoch": 0.25, | |
| "learning_rate": 1.4875316226960608e-05, | |
| "loss": 3.5434, | |
| "step": 690, | |
| "task_loss": 1.8874092102050781 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.7156856060028076, | |
| "epoch": 0.25, | |
| "learning_rate": 1.5092157571376944e-05, | |
| "loss": 3.6905, | |
| "step": 700, | |
| "task_loss": 1.9769742488861084 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.2249040603637695, | |
| "epoch": 0.26, | |
| "learning_rate": 1.530899891579328e-05, | |
| "loss": 3.6165, | |
| "step": 710, | |
| "task_loss": 1.8299891948699951 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.8128280639648438, | |
| "epoch": 0.26, | |
| "learning_rate": 1.5525840260209615e-05, | |
| "loss": 3.6727, | |
| "step": 720, | |
| "task_loss": 1.8565778732299805 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.894437789916992, | |
| "epoch": 0.26, | |
| "learning_rate": 1.574268160462595e-05, | |
| "loss": 3.8419, | |
| "step": 730, | |
| "task_loss": 1.9490731954574585 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.6588613986968994, | |
| "epoch": 0.27, | |
| "learning_rate": 1.5959522949042286e-05, | |
| "loss": 3.719, | |
| "step": 740, | |
| "task_loss": 2.055631399154663 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.815298557281494, | |
| "epoch": 0.27, | |
| "learning_rate": 1.6176364293458618e-05, | |
| "loss": 3.5813, | |
| "step": 750, | |
| "task_loss": 1.935587763786316 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "eval_exact_match": 58.55250709555345, | |
| "eval_f1": 70.94607775185224, | |
| "step": 750 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.76840877532959, | |
| "epoch": 0.27, | |
| "learning_rate": 1.6393205637874954e-05, | |
| "loss": 3.5721, | |
| "step": 760, | |
| "task_loss": 2.5942468643188477 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.8197884559631348, | |
| "epoch": 0.28, | |
| "learning_rate": 1.661004698229129e-05, | |
| "loss": 3.406, | |
| "step": 770, | |
| "task_loss": 2.399030923843384 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.773465633392334, | |
| "epoch": 0.28, | |
| "learning_rate": 1.6826888326707625e-05, | |
| "loss": 3.6862, | |
| "step": 780, | |
| "task_loss": 2.0786447525024414 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.5512259006500244, | |
| "epoch": 0.29, | |
| "learning_rate": 1.704372967112396e-05, | |
| "loss": 3.4023, | |
| "step": 790, | |
| "task_loss": 1.7031829357147217 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.154162406921387, | |
| "epoch": 0.29, | |
| "learning_rate": 1.7260571015540296e-05, | |
| "loss": 3.5814, | |
| "step": 800, | |
| "task_loss": 2.033867597579956 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.1742305755615234, | |
| "epoch": 0.29, | |
| "learning_rate": 1.747741235995663e-05, | |
| "loss": 3.5524, | |
| "step": 810, | |
| "task_loss": 1.7235503196716309 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.667600154876709, | |
| "epoch": 0.3, | |
| "learning_rate": 1.7694253704372967e-05, | |
| "loss": 3.1374, | |
| "step": 820, | |
| "task_loss": 2.0208334922790527 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.541199207305908, | |
| "epoch": 0.3, | |
| "learning_rate": 1.7911095048789303e-05, | |
| "loss": 3.5502, | |
| "step": 830, | |
| "task_loss": 2.0909945964813232 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.158015012741089, | |
| "epoch": 0.3, | |
| "learning_rate": 1.812793639320564e-05, | |
| "loss": 3.1089, | |
| "step": 840, | |
| "task_loss": 1.5247281789779663 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.0832254886627197, | |
| "epoch": 0.31, | |
| "learning_rate": 1.8344777737621974e-05, | |
| "loss": 3.3842, | |
| "step": 850, | |
| "task_loss": 1.5567718744277954 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.231353521347046, | |
| "epoch": 0.31, | |
| "learning_rate": 1.856161908203831e-05, | |
| "loss": 3.2243, | |
| "step": 860, | |
| "task_loss": 2.2011852264404297 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.409111261367798, | |
| "epoch": 0.31, | |
| "learning_rate": 1.8778460426454645e-05, | |
| "loss": 3.1738, | |
| "step": 870, | |
| "task_loss": 1.6380929946899414 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.968132495880127, | |
| "epoch": 0.32, | |
| "learning_rate": 1.8995301770870977e-05, | |
| "loss": 3.184, | |
| "step": 880, | |
| "task_loss": 1.0741721391677856 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.7900476455688477, | |
| "epoch": 0.32, | |
| "learning_rate": 1.9212143115287313e-05, | |
| "loss": 3.3852, | |
| "step": 890, | |
| "task_loss": 2.300794839859009 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.6072299480438232, | |
| "epoch": 0.33, | |
| "learning_rate": 1.942898445970365e-05, | |
| "loss": 3.0648, | |
| "step": 900, | |
| "task_loss": 1.9764083623886108 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.560149908065796, | |
| "epoch": 0.33, | |
| "learning_rate": 1.9645825804119984e-05, | |
| "loss": 3.1941, | |
| "step": 910, | |
| "task_loss": 2.037846088409424 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.539365768432617, | |
| "epoch": 0.33, | |
| "learning_rate": 1.986266714853632e-05, | |
| "loss": 3.1535, | |
| "step": 920, | |
| "task_loss": 1.5650080442428589 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.952303409576416, | |
| "epoch": 0.34, | |
| "learning_rate": 2.007950849295266e-05, | |
| "loss": 3.0291, | |
| "step": 930, | |
| "task_loss": 2.1716508865356445 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 4.5406928062438965, | |
| "epoch": 0.34, | |
| "learning_rate": 2.0296349837368994e-05, | |
| "loss": 3.2595, | |
| "step": 940, | |
| "task_loss": 2.8979098796844482 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.1165611743927, | |
| "epoch": 0.34, | |
| "learning_rate": 2.051319118178533e-05, | |
| "loss": 3.0601, | |
| "step": 950, | |
| "task_loss": 1.5383003950119019 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.121710777282715, | |
| "epoch": 0.35, | |
| "learning_rate": 2.0730032526201665e-05, | |
| "loss": 3.1225, | |
| "step": 960, | |
| "task_loss": 1.6968241930007935 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.3333802223205566, | |
| "epoch": 0.35, | |
| "learning_rate": 2.0946873870618e-05, | |
| "loss": 3.0869, | |
| "step": 970, | |
| "task_loss": 2.1477713584899902 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.538954257965088, | |
| "epoch": 0.35, | |
| "learning_rate": 2.1163715215034336e-05, | |
| "loss": 2.8604, | |
| "step": 980, | |
| "task_loss": 1.5192737579345703 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.781994581222534, | |
| "epoch": 0.36, | |
| "learning_rate": 2.1380556559450672e-05, | |
| "loss": 2.6585, | |
| "step": 990, | |
| "task_loss": 1.9368581771850586 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.892576217651367, | |
| "epoch": 0.36, | |
| "learning_rate": 2.1597397903867004e-05, | |
| "loss": 3.2002, | |
| "step": 1000, | |
| "task_loss": 2.37716007232666 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "eval_exact_match": 64.34247871333964, | |
| "eval_f1": 76.12832427154721, | |
| "step": 1000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.0241000652313232, | |
| "epoch": 0.37, | |
| "learning_rate": 2.181423924828334e-05, | |
| "loss": 3.118, | |
| "step": 1010, | |
| "task_loss": 1.4347963333129883 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.0104122161865234, | |
| "epoch": 0.37, | |
| "learning_rate": 2.2031080592699675e-05, | |
| "loss": 2.8552, | |
| "step": 1020, | |
| "task_loss": 1.7555286884307861 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.568011999130249, | |
| "epoch": 0.37, | |
| "learning_rate": 2.224792193711601e-05, | |
| "loss": 2.918, | |
| "step": 1030, | |
| "task_loss": 1.1540263891220093 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.7906365394592285, | |
| "epoch": 0.38, | |
| "learning_rate": 2.2464763281532346e-05, | |
| "loss": 3.0068, | |
| "step": 1040, | |
| "task_loss": 2.4095299243927 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.682779312133789, | |
| "epoch": 0.38, | |
| "learning_rate": 2.2681604625948682e-05, | |
| "loss": 2.6718, | |
| "step": 1050, | |
| "task_loss": 1.969969391822815 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.5220847129821777, | |
| "epoch": 0.38, | |
| "learning_rate": 2.2898445970365018e-05, | |
| "loss": 2.6947, | |
| "step": 1060, | |
| "task_loss": 1.0111178159713745 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.6033337116241455, | |
| "epoch": 0.39, | |
| "learning_rate": 2.3115287314781353e-05, | |
| "loss": 2.8889, | |
| "step": 1070, | |
| "task_loss": 1.8982610702514648 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.147583484649658, | |
| "epoch": 0.39, | |
| "learning_rate": 2.333212865919769e-05, | |
| "loss": 2.8495, | |
| "step": 1080, | |
| "task_loss": 1.5683348178863525 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.5625667572021484, | |
| "epoch": 0.39, | |
| "learning_rate": 2.3548970003614024e-05, | |
| "loss": 2.9222, | |
| "step": 1090, | |
| "task_loss": 1.891990065574646 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.2590126991271973, | |
| "epoch": 0.4, | |
| "learning_rate": 2.376581134803036e-05, | |
| "loss": 2.8199, | |
| "step": 1100, | |
| "task_loss": 1.4266440868377686 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.4584269523620605, | |
| "epoch": 0.4, | |
| "learning_rate": 2.3982652692446695e-05, | |
| "loss": 2.856, | |
| "step": 1110, | |
| "task_loss": 1.945828914642334 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.8447415828704834, | |
| "epoch": 0.4, | |
| "learning_rate": 2.4199494036863028e-05, | |
| "loss": 2.5436, | |
| "step": 1120, | |
| "task_loss": 1.3307209014892578 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.670109510421753, | |
| "epoch": 0.41, | |
| "learning_rate": 2.4416335381279363e-05, | |
| "loss": 2.7268, | |
| "step": 1130, | |
| "task_loss": 1.9327623844146729 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.741763114929199, | |
| "epoch": 0.41, | |
| "learning_rate": 2.46331767256957e-05, | |
| "loss": 2.8081, | |
| "step": 1140, | |
| "task_loss": 1.631276249885559 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.837188482284546, | |
| "epoch": 0.42, | |
| "learning_rate": 2.4850018070112034e-05, | |
| "loss": 2.6292, | |
| "step": 1150, | |
| "task_loss": 1.422114372253418 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.2892446517944336, | |
| "epoch": 0.42, | |
| "learning_rate": 2.506685941452837e-05, | |
| "loss": 2.7805, | |
| "step": 1160, | |
| "task_loss": 1.5163651704788208 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.5814714431762695, | |
| "epoch": 0.42, | |
| "learning_rate": 2.5283700758944705e-05, | |
| "loss": 2.7588, | |
| "step": 1170, | |
| "task_loss": 1.4490866661071777 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.4073421955108643, | |
| "epoch": 0.43, | |
| "learning_rate": 2.550054210336104e-05, | |
| "loss": 2.5651, | |
| "step": 1180, | |
| "task_loss": 1.9651927947998047 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.1154234409332275, | |
| "epoch": 0.43, | |
| "learning_rate": 2.5717383447777377e-05, | |
| "loss": 2.5451, | |
| "step": 1190, | |
| "task_loss": 1.5143002271652222 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.7987661361694336, | |
| "epoch": 0.43, | |
| "learning_rate": 2.5934224792193712e-05, | |
| "loss": 2.6821, | |
| "step": 1200, | |
| "task_loss": 1.6177781820297241 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.1852757930755615, | |
| "epoch": 0.44, | |
| "learning_rate": 2.6151066136610048e-05, | |
| "loss": 2.4611, | |
| "step": 1210, | |
| "task_loss": 1.2211247682571411 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.9138824939727783, | |
| "epoch": 0.44, | |
| "learning_rate": 2.6367907481026383e-05, | |
| "loss": 2.4774, | |
| "step": 1220, | |
| "task_loss": 1.7106552124023438 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.2626287937164307, | |
| "epoch": 0.44, | |
| "learning_rate": 2.658474882544272e-05, | |
| "loss": 2.5318, | |
| "step": 1230, | |
| "task_loss": 1.307836890220642 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.6250033378601074, | |
| "epoch": 0.45, | |
| "learning_rate": 2.680159016985905e-05, | |
| "loss": 2.5726, | |
| "step": 1240, | |
| "task_loss": 1.559349536895752 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.5571208000183105, | |
| "epoch": 0.45, | |
| "learning_rate": 2.7018431514275387e-05, | |
| "loss": 2.489, | |
| "step": 1250, | |
| "task_loss": 1.1737449169158936 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "eval_exact_match": 67.50236518448439, | |
| "eval_f1": 78.53072236477003, | |
| "step": 1250 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.1527514457702637, | |
| "epoch": 0.46, | |
| "learning_rate": 2.7235272858691722e-05, | |
| "loss": 2.6094, | |
| "step": 1260, | |
| "task_loss": 1.478395700454712 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.001819610595703, | |
| "epoch": 0.46, | |
| "learning_rate": 2.7452114203108058e-05, | |
| "loss": 2.7699, | |
| "step": 1270, | |
| "task_loss": 1.6132569313049316 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.6055197715759277, | |
| "epoch": 0.46, | |
| "learning_rate": 2.7668955547524393e-05, | |
| "loss": 2.4624, | |
| "step": 1280, | |
| "task_loss": 1.5703206062316895 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.221977472305298, | |
| "epoch": 0.47, | |
| "learning_rate": 2.788579689194073e-05, | |
| "loss": 2.4796, | |
| "step": 1290, | |
| "task_loss": 1.2309448719024658 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.17022705078125, | |
| "epoch": 0.47, | |
| "learning_rate": 2.8102638236357068e-05, | |
| "loss": 2.4461, | |
| "step": 1300, | |
| "task_loss": 1.4810001850128174 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.275075912475586, | |
| "epoch": 0.47, | |
| "learning_rate": 2.8319479580773403e-05, | |
| "loss": 2.6478, | |
| "step": 1310, | |
| "task_loss": 2.2076168060302734 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.1733217239379883, | |
| "epoch": 0.48, | |
| "learning_rate": 2.853632092518974e-05, | |
| "loss": 2.3909, | |
| "step": 1320, | |
| "task_loss": 1.4496859312057495 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.3826065063476562, | |
| "epoch": 0.48, | |
| "learning_rate": 2.8753162269606075e-05, | |
| "loss": 2.3423, | |
| "step": 1330, | |
| "task_loss": 1.4332823753356934 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.4819350242614746, | |
| "epoch": 0.48, | |
| "learning_rate": 2.897000361402241e-05, | |
| "loss": 2.3712, | |
| "step": 1340, | |
| "task_loss": 1.7626259326934814 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.2148211002349854, | |
| "epoch": 0.49, | |
| "learning_rate": 2.9186844958438746e-05, | |
| "loss": 2.3867, | |
| "step": 1350, | |
| "task_loss": 1.9342608451843262 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.4345703125, | |
| "epoch": 0.49, | |
| "learning_rate": 2.9403686302855078e-05, | |
| "loss": 2.3923, | |
| "step": 1360, | |
| "task_loss": 1.5897960662841797 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.2300848960876465, | |
| "epoch": 0.5, | |
| "learning_rate": 2.9620527647271414e-05, | |
| "loss": 2.225, | |
| "step": 1370, | |
| "task_loss": 1.2050740718841553 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.686323881149292, | |
| "epoch": 0.5, | |
| "learning_rate": 2.983736899168775e-05, | |
| "loss": 2.3038, | |
| "step": 1380, | |
| "task_loss": 1.9944977760314941 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.4140007495880127, | |
| "epoch": 0.5, | |
| "learning_rate": 3.005421033610408e-05, | |
| "loss": 2.3103, | |
| "step": 1390, | |
| "task_loss": 1.4956495761871338 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.1493027210235596, | |
| "epoch": 0.51, | |
| "learning_rate": 3.027105168052042e-05, | |
| "loss": 2.278, | |
| "step": 1400, | |
| "task_loss": 1.2530300617218018 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.3148369789123535, | |
| "epoch": 0.51, | |
| "learning_rate": 3.0487893024936752e-05, | |
| "loss": 2.0819, | |
| "step": 1410, | |
| "task_loss": 1.466301441192627 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.3211307525634766, | |
| "epoch": 0.51, | |
| "learning_rate": 3.070473436935309e-05, | |
| "loss": 2.2652, | |
| "step": 1420, | |
| "task_loss": 1.6483142375946045 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.6457409858703613, | |
| "epoch": 0.52, | |
| "learning_rate": 3.092157571376942e-05, | |
| "loss": 2.425, | |
| "step": 1430, | |
| "task_loss": 1.5462552309036255 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.4564802646636963, | |
| "epoch": 0.52, | |
| "learning_rate": 3.113841705818576e-05, | |
| "loss": 2.2011, | |
| "step": 1440, | |
| "task_loss": 1.7669343948364258 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.213484764099121, | |
| "epoch": 0.52, | |
| "learning_rate": 3.135525840260209e-05, | |
| "loss": 2.3324, | |
| "step": 1450, | |
| "task_loss": 1.2010239362716675 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.1059863567352295, | |
| "epoch": 0.53, | |
| "learning_rate": 3.1572099747018434e-05, | |
| "loss": 2.3631, | |
| "step": 1460, | |
| "task_loss": 1.0316790342330933 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.007906913757324, | |
| "epoch": 0.53, | |
| "learning_rate": 3.178894109143476e-05, | |
| "loss": 2.2032, | |
| "step": 1470, | |
| "task_loss": 1.1084401607513428 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.6057040691375732, | |
| "epoch": 0.53, | |
| "learning_rate": 3.2005782435851105e-05, | |
| "loss": 2.3162, | |
| "step": 1480, | |
| "task_loss": 1.4449225664138794 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.419894218444824, | |
| "epoch": 0.54, | |
| "learning_rate": 3.222262378026744e-05, | |
| "loss": 2.1313, | |
| "step": 1490, | |
| "task_loss": 1.390395164489746 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.433657646179199, | |
| "epoch": 0.54, | |
| "learning_rate": 3.2439465124683776e-05, | |
| "loss": 2.3071, | |
| "step": 1500, | |
| "task_loss": 1.726881504058838 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "eval_exact_match": 70.34058656575213, | |
| "eval_f1": 81.00798952513871, | |
| "step": 1500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.9914134740829468, | |
| "epoch": 0.55, | |
| "learning_rate": 3.265630646910011e-05, | |
| "loss": 2.4203, | |
| "step": 1510, | |
| "task_loss": 1.3354167938232422 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.2558791637420654, | |
| "epoch": 0.55, | |
| "learning_rate": 3.287314781351645e-05, | |
| "loss": 2.3477, | |
| "step": 1520, | |
| "task_loss": 1.4666078090667725 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.4174277782440186, | |
| "epoch": 0.55, | |
| "learning_rate": 3.308998915793278e-05, | |
| "loss": 2.226, | |
| "step": 1530, | |
| "task_loss": 1.690213918685913 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.663379430770874, | |
| "epoch": 0.56, | |
| "learning_rate": 3.330683050234912e-05, | |
| "loss": 2.2701, | |
| "step": 1540, | |
| "task_loss": 1.8850646018981934 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.476119041442871, | |
| "epoch": 0.56, | |
| "learning_rate": 3.3523671846765454e-05, | |
| "loss": 2.2315, | |
| "step": 1550, | |
| "task_loss": 1.7790265083312988 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.5638389587402344, | |
| "epoch": 0.56, | |
| "learning_rate": 3.374051319118178e-05, | |
| "loss": 2.2065, | |
| "step": 1560, | |
| "task_loss": 2.4445137977600098 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.102419376373291, | |
| "epoch": 0.57, | |
| "learning_rate": 3.3957354535598125e-05, | |
| "loss": 2.0694, | |
| "step": 1570, | |
| "task_loss": 0.849961519241333 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.377769947052002, | |
| "epoch": 0.57, | |
| "learning_rate": 3.4174195880014454e-05, | |
| "loss": 2.3494, | |
| "step": 1580, | |
| "task_loss": 1.6015727519989014 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.992417812347412, | |
| "epoch": 0.57, | |
| "learning_rate": 3.4391037224430796e-05, | |
| "loss": 2.2862, | |
| "step": 1590, | |
| "task_loss": 1.7169113159179688 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.1633100509643555, | |
| "epoch": 0.58, | |
| "learning_rate": 3.4607878568847125e-05, | |
| "loss": 2.2645, | |
| "step": 1600, | |
| "task_loss": 1.168750524520874 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.846243977546692, | |
| "epoch": 0.58, | |
| "learning_rate": 3.482471991326347e-05, | |
| "loss": 2.1987, | |
| "step": 1610, | |
| "task_loss": 1.0060114860534668 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.866645097732544, | |
| "epoch": 0.59, | |
| "learning_rate": 3.5041561257679796e-05, | |
| "loss": 2.1537, | |
| "step": 1620, | |
| "task_loss": 1.2443784475326538 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.573518753051758, | |
| "epoch": 0.59, | |
| "learning_rate": 3.525840260209614e-05, | |
| "loss": 2.126, | |
| "step": 1630, | |
| "task_loss": 2.0661087036132812 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.5710911750793457, | |
| "epoch": 0.59, | |
| "learning_rate": 3.547524394651247e-05, | |
| "loss": 2.2293, | |
| "step": 1640, | |
| "task_loss": 2.133748769760132 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.9118821620941162, | |
| "epoch": 0.6, | |
| "learning_rate": 3.569208529092881e-05, | |
| "loss": 2.1774, | |
| "step": 1650, | |
| "task_loss": 0.9966872334480286 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.572476863861084, | |
| "epoch": 0.6, | |
| "learning_rate": 3.590892663534514e-05, | |
| "loss": 2.057, | |
| "step": 1660, | |
| "task_loss": 1.7666373252868652 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.8772165775299072, | |
| "epoch": 0.6, | |
| "learning_rate": 3.6125767979761474e-05, | |
| "loss": 2.1655, | |
| "step": 1670, | |
| "task_loss": 1.931375503540039 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.8401541709899902, | |
| "epoch": 0.61, | |
| "learning_rate": 3.634260932417781e-05, | |
| "loss": 1.9505, | |
| "step": 1680, | |
| "task_loss": 1.2190885543823242 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.8582175970077515, | |
| "epoch": 0.61, | |
| "learning_rate": 3.6559450668594145e-05, | |
| "loss": 2.076, | |
| "step": 1690, | |
| "task_loss": 1.3368051052093506 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.614755153656006, | |
| "epoch": 0.61, | |
| "learning_rate": 3.677629201301048e-05, | |
| "loss": 2.3466, | |
| "step": 1700, | |
| "task_loss": 1.8578553199768066 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.2779784202575684, | |
| "epoch": 0.62, | |
| "learning_rate": 3.6993133357426816e-05, | |
| "loss": 2.1923, | |
| "step": 1710, | |
| "task_loss": 1.5653705596923828 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.77517032623291, | |
| "epoch": 0.62, | |
| "learning_rate": 3.720997470184315e-05, | |
| "loss": 2.0672, | |
| "step": 1720, | |
| "task_loss": 1.7407598495483398 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.3437767028808594, | |
| "epoch": 0.63, | |
| "learning_rate": 3.742681604625949e-05, | |
| "loss": 2.2019, | |
| "step": 1730, | |
| "task_loss": 1.8152300119400024 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.2300639152526855, | |
| "epoch": 0.63, | |
| "learning_rate": 3.764365739067582e-05, | |
| "loss": 1.9345, | |
| "step": 1740, | |
| "task_loss": 1.5370993614196777 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.157289981842041, | |
| "epoch": 0.63, | |
| "learning_rate": 3.786049873509216e-05, | |
| "loss": 1.8016, | |
| "step": 1750, | |
| "task_loss": 1.4095211029052734 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "eval_exact_match": 71.9205298013245, | |
| "eval_f1": 82.04179007964369, | |
| "step": 1750 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6568679809570312, | |
| "epoch": 0.64, | |
| "learning_rate": 3.8077340079508494e-05, | |
| "loss": 2.1419, | |
| "step": 1760, | |
| "task_loss": 1.0038526058197021 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.557297706604004, | |
| "epoch": 0.64, | |
| "learning_rate": 3.829418142392483e-05, | |
| "loss": 1.8646, | |
| "step": 1770, | |
| "task_loss": 1.6112699508666992 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.2386903762817383, | |
| "epoch": 0.64, | |
| "learning_rate": 3.8511022768341165e-05, | |
| "loss": 2.1456, | |
| "step": 1780, | |
| "task_loss": 1.4335222244262695 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.2769885063171387, | |
| "epoch": 0.65, | |
| "learning_rate": 3.87278641127575e-05, | |
| "loss": 2.0989, | |
| "step": 1790, | |
| "task_loss": 1.8141961097717285 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.8742246627807617, | |
| "epoch": 0.65, | |
| "learning_rate": 3.894470545717383e-05, | |
| "loss": 2.258, | |
| "step": 1800, | |
| "task_loss": 1.5317169427871704 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.2876062393188477, | |
| "epoch": 0.65, | |
| "learning_rate": 3.916154680159017e-05, | |
| "loss": 1.9455, | |
| "step": 1810, | |
| "task_loss": 1.7473857402801514 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.066962718963623, | |
| "epoch": 0.66, | |
| "learning_rate": 3.93783881460065e-05, | |
| "loss": 1.8893, | |
| "step": 1820, | |
| "task_loss": 1.0266292095184326 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6544253826141357, | |
| "epoch": 0.66, | |
| "learning_rate": 3.959522949042284e-05, | |
| "loss": 2.0365, | |
| "step": 1830, | |
| "task_loss": 1.080996036529541 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.3459901809692383, | |
| "epoch": 0.66, | |
| "learning_rate": 3.981207083483917e-05, | |
| "loss": 1.9773, | |
| "step": 1840, | |
| "task_loss": 1.5108176469802856 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.3785009384155273, | |
| "epoch": 0.67, | |
| "learning_rate": 4.0028912179255514e-05, | |
| "loss": 2.0769, | |
| "step": 1850, | |
| "task_loss": 2.11570405960083 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.0890772342681885, | |
| "epoch": 0.67, | |
| "learning_rate": 4.024575352367185e-05, | |
| "loss": 1.9434, | |
| "step": 1860, | |
| "task_loss": 1.0435796976089478 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.71674382686615, | |
| "epoch": 0.68, | |
| "learning_rate": 4.0462594868088185e-05, | |
| "loss": 1.9389, | |
| "step": 1870, | |
| "task_loss": 1.1821942329406738 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.101773500442505, | |
| "epoch": 0.68, | |
| "learning_rate": 4.067943621250452e-05, | |
| "loss": 2.0819, | |
| "step": 1880, | |
| "task_loss": 1.3899006843566895 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.432915687561035, | |
| "epoch": 0.68, | |
| "learning_rate": 4.0896277556920857e-05, | |
| "loss": 1.9742, | |
| "step": 1890, | |
| "task_loss": 1.5817327499389648 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6859402656555176, | |
| "epoch": 0.69, | |
| "learning_rate": 4.111311890133719e-05, | |
| "loss": 2.0378, | |
| "step": 1900, | |
| "task_loss": 0.9370235800743103 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6273126602172852, | |
| "epoch": 0.69, | |
| "learning_rate": 4.132996024575352e-05, | |
| "loss": 1.9215, | |
| "step": 1910, | |
| "task_loss": 1.2056360244750977 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.7799351215362549, | |
| "epoch": 0.69, | |
| "learning_rate": 4.154680159016986e-05, | |
| "loss": 1.9668, | |
| "step": 1920, | |
| "task_loss": 0.8631846904754639 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.9689221382141113, | |
| "epoch": 0.7, | |
| "learning_rate": 4.176364293458619e-05, | |
| "loss": 1.9988, | |
| "step": 1930, | |
| "task_loss": 1.3810503482818604 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.826797366142273, | |
| "epoch": 0.7, | |
| "learning_rate": 4.1980484279002534e-05, | |
| "loss": 2.0241, | |
| "step": 1940, | |
| "task_loss": 1.1184316873550415 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.8849117755889893, | |
| "epoch": 0.7, | |
| "learning_rate": 4.219732562341886e-05, | |
| "loss": 2.0848, | |
| "step": 1950, | |
| "task_loss": 1.642095923423767 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6249830722808838, | |
| "epoch": 0.71, | |
| "learning_rate": 4.2414166967835206e-05, | |
| "loss": 1.9932, | |
| "step": 1960, | |
| "task_loss": 1.0573899745941162 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3523120880126953, | |
| "epoch": 0.71, | |
| "learning_rate": 4.2631008312251534e-05, | |
| "loss": 1.7783, | |
| "step": 1970, | |
| "task_loss": 0.67606520652771 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.1321749687194824, | |
| "epoch": 0.72, | |
| "learning_rate": 4.284784965666788e-05, | |
| "loss": 1.7797, | |
| "step": 1980, | |
| "task_loss": 1.814510464668274 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4253501892089844, | |
| "epoch": 0.72, | |
| "learning_rate": 4.3064691001084205e-05, | |
| "loss": 1.9431, | |
| "step": 1990, | |
| "task_loss": 0.946668803691864 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.3535895347595215, | |
| "epoch": 0.72, | |
| "learning_rate": 4.328153234550055e-05, | |
| "loss": 1.9957, | |
| "step": 2000, | |
| "task_loss": 1.8486053943634033 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "eval_exact_match": 73.1693472090823, | |
| "eval_f1": 83.02894145240745, | |
| "step": 2000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.8838860988616943, | |
| "epoch": 0.73, | |
| "learning_rate": 4.349837368991688e-05, | |
| "loss": 1.9315, | |
| "step": 2010, | |
| "task_loss": 1.5416061878204346 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.8987743854522705, | |
| "epoch": 0.73, | |
| "learning_rate": 4.371521503433322e-05, | |
| "loss": 1.9744, | |
| "step": 2020, | |
| "task_loss": 1.4941067695617676 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.972581386566162, | |
| "epoch": 0.73, | |
| "learning_rate": 4.393205637874955e-05, | |
| "loss": 1.8702, | |
| "step": 2030, | |
| "task_loss": 1.1873316764831543 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.1103854179382324, | |
| "epoch": 0.74, | |
| "learning_rate": 4.414889772316588e-05, | |
| "loss": 1.7989, | |
| "step": 2040, | |
| "task_loss": 1.9999923706054688 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.1972813606262207, | |
| "epoch": 0.74, | |
| "learning_rate": 4.436573906758222e-05, | |
| "loss": 2.0754, | |
| "step": 2050, | |
| "task_loss": 1.3373115062713623 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.598461627960205, | |
| "epoch": 0.74, | |
| "learning_rate": 4.4582580411998554e-05, | |
| "loss": 1.9482, | |
| "step": 2060, | |
| "task_loss": 0.7865440845489502 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.3904833793640137, | |
| "epoch": 0.75, | |
| "learning_rate": 4.479942175641489e-05, | |
| "loss": 1.9746, | |
| "step": 2070, | |
| "task_loss": 1.9388031959533691 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6508259773254395, | |
| "epoch": 0.75, | |
| "learning_rate": 4.5016263100831226e-05, | |
| "loss": 1.9099, | |
| "step": 2080, | |
| "task_loss": 1.0399653911590576 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.772383451461792, | |
| "epoch": 0.76, | |
| "learning_rate": 4.523310444524756e-05, | |
| "loss": 1.7396, | |
| "step": 2090, | |
| "task_loss": 0.9913707375526428 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4372773170471191, | |
| "epoch": 0.76, | |
| "learning_rate": 4.54499457896639e-05, | |
| "loss": 1.6997, | |
| "step": 2100, | |
| "task_loss": 0.9053939580917358 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.8861255645751953, | |
| "epoch": 0.76, | |
| "learning_rate": 4.566678713408023e-05, | |
| "loss": 1.9588, | |
| "step": 2110, | |
| "task_loss": 1.1382219791412354 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.492857575416565, | |
| "epoch": 0.77, | |
| "learning_rate": 4.588362847849657e-05, | |
| "loss": 2.0162, | |
| "step": 2120, | |
| "task_loss": 1.0419156551361084 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.141202926635742, | |
| "epoch": 0.77, | |
| "learning_rate": 4.6100469822912903e-05, | |
| "loss": 1.9327, | |
| "step": 2130, | |
| "task_loss": 1.0418000221252441 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.920304536819458, | |
| "epoch": 0.77, | |
| "learning_rate": 4.631731116732924e-05, | |
| "loss": 1.8879, | |
| "step": 2140, | |
| "task_loss": 1.1978967189788818 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5354628562927246, | |
| "epoch": 0.78, | |
| "learning_rate": 4.6534152511745575e-05, | |
| "loss": 1.7451, | |
| "step": 2150, | |
| "task_loss": 0.8436750173568726 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4595377445220947, | |
| "epoch": 0.78, | |
| "learning_rate": 4.675099385616191e-05, | |
| "loss": 1.7424, | |
| "step": 2160, | |
| "task_loss": 1.038141131401062 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.34226393699646, | |
| "epoch": 0.78, | |
| "learning_rate": 4.696783520057824e-05, | |
| "loss": 1.7037, | |
| "step": 2170, | |
| "task_loss": 1.9839649200439453 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.8102070093154907, | |
| "epoch": 0.79, | |
| "learning_rate": 4.718467654499458e-05, | |
| "loss": 1.7556, | |
| "step": 2180, | |
| "task_loss": 1.7777656316757202 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5364153385162354, | |
| "epoch": 0.79, | |
| "learning_rate": 4.740151788941091e-05, | |
| "loss": 1.8958, | |
| "step": 2190, | |
| "task_loss": 0.9634038805961609 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.3901851177215576, | |
| "epoch": 0.8, | |
| "learning_rate": 4.761835923382725e-05, | |
| "loss": 1.8381, | |
| "step": 2200, | |
| "task_loss": 1.4805151224136353 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.725096344947815, | |
| "epoch": 0.8, | |
| "learning_rate": 4.783520057824358e-05, | |
| "loss": 1.6794, | |
| "step": 2210, | |
| "task_loss": 1.4682140350341797 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.2090702056884766, | |
| "epoch": 0.8, | |
| "learning_rate": 4.8052041922659924e-05, | |
| "loss": 1.9593, | |
| "step": 2220, | |
| "task_loss": 1.230177402496338 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.041393756866455, | |
| "epoch": 0.81, | |
| "learning_rate": 4.826888326707626e-05, | |
| "loss": 1.809, | |
| "step": 2230, | |
| "task_loss": 1.3010313510894775 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.7629666328430176, | |
| "epoch": 0.81, | |
| "learning_rate": 4.8485724611492595e-05, | |
| "loss": 1.9261, | |
| "step": 2240, | |
| "task_loss": 0.9947839379310608 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.264925003051758, | |
| "epoch": 0.81, | |
| "learning_rate": 4.870256595590893e-05, | |
| "loss": 1.902, | |
| "step": 2250, | |
| "task_loss": 1.4164844751358032 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "eval_exact_match": 74.87228003784296, | |
| "eval_f1": 84.24213227343729, | |
| "step": 2250 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6021579504013062, | |
| "epoch": 0.82, | |
| "learning_rate": 4.8919407300325266e-05, | |
| "loss": 1.7243, | |
| "step": 2260, | |
| "task_loss": 1.630518913269043 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.2044320106506348, | |
| "epoch": 0.82, | |
| "learning_rate": 4.91362486447416e-05, | |
| "loss": 1.7822, | |
| "step": 2270, | |
| "task_loss": 1.8210062980651855 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.79202401638031, | |
| "epoch": 0.82, | |
| "learning_rate": 4.935308998915793e-05, | |
| "loss": 1.7531, | |
| "step": 2280, | |
| "task_loss": 1.189091444015503 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.06649112701416, | |
| "epoch": 0.83, | |
| "learning_rate": 4.956993133357427e-05, | |
| "loss": 1.844, | |
| "step": 2290, | |
| "task_loss": 1.1661924123764038 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.682025671005249, | |
| "epoch": 0.83, | |
| "learning_rate": 4.97867726779906e-05, | |
| "loss": 1.7941, | |
| "step": 2300, | |
| "task_loss": 1.0544310808181763 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3249014616012573, | |
| "epoch": 0.83, | |
| "learning_rate": 5.0003614022406944e-05, | |
| "loss": 1.7689, | |
| "step": 2310, | |
| "task_loss": 0.8618177771568298 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.0047874450683594, | |
| "epoch": 0.84, | |
| "learning_rate": 5.022045536682327e-05, | |
| "loss": 1.6503, | |
| "step": 2320, | |
| "task_loss": 1.8260812759399414 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6946525573730469, | |
| "epoch": 0.84, | |
| "learning_rate": 5.0437296711239615e-05, | |
| "loss": 1.6434, | |
| "step": 2330, | |
| "task_loss": 0.9825450778007507 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3918914794921875, | |
| "epoch": 0.85, | |
| "learning_rate": 5.0654138055655944e-05, | |
| "loss": 1.7978, | |
| "step": 2340, | |
| "task_loss": 1.3643735647201538 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.7238832712173462, | |
| "epoch": 0.85, | |
| "learning_rate": 5.0870979400072286e-05, | |
| "loss": 1.7657, | |
| "step": 2350, | |
| "task_loss": 1.165932536125183 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.7872267961502075, | |
| "epoch": 0.85, | |
| "learning_rate": 5.1087820744488615e-05, | |
| "loss": 1.7084, | |
| "step": 2360, | |
| "task_loss": 1.5586845874786377 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.0670828819274902, | |
| "epoch": 0.86, | |
| "learning_rate": 5.130466208890496e-05, | |
| "loss": 1.7082, | |
| "step": 2370, | |
| "task_loss": 1.581449270248413 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.378608226776123, | |
| "epoch": 0.86, | |
| "learning_rate": 5.1521503433321286e-05, | |
| "loss": 1.755, | |
| "step": 2380, | |
| "task_loss": 0.980684757232666 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.0133185386657715, | |
| "epoch": 0.86, | |
| "learning_rate": 5.173834477773763e-05, | |
| "loss": 1.8557, | |
| "step": 2390, | |
| "task_loss": 1.6304588317871094 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.729528546333313, | |
| "epoch": 0.87, | |
| "learning_rate": 5.195518612215396e-05, | |
| "loss": 1.7122, | |
| "step": 2400, | |
| "task_loss": 1.540935754776001 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3772521018981934, | |
| "epoch": 0.87, | |
| "learning_rate": 5.217202746657029e-05, | |
| "loss": 1.7996, | |
| "step": 2410, | |
| "task_loss": 0.889678955078125 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.8892799615859985, | |
| "epoch": 0.87, | |
| "learning_rate": 5.238886881098663e-05, | |
| "loss": 1.8532, | |
| "step": 2420, | |
| "task_loss": 1.5641158819198608 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5366939306259155, | |
| "epoch": 0.88, | |
| "learning_rate": 5.2605710155402964e-05, | |
| "loss": 1.5529, | |
| "step": 2430, | |
| "task_loss": 0.9511038661003113 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 3.0125789642333984, | |
| "epoch": 0.88, | |
| "learning_rate": 5.28225514998193e-05, | |
| "loss": 1.7891, | |
| "step": 2440, | |
| "task_loss": 2.29156756401062 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.3766469955444336, | |
| "epoch": 0.89, | |
| "learning_rate": 5.3039392844235635e-05, | |
| "loss": 1.6349, | |
| "step": 2450, | |
| "task_loss": 1.3916804790496826 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5166912078857422, | |
| "epoch": 0.89, | |
| "learning_rate": 5.325623418865197e-05, | |
| "loss": 1.7089, | |
| "step": 2460, | |
| "task_loss": 1.1395492553710938 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4228307008743286, | |
| "epoch": 0.89, | |
| "learning_rate": 5.3473075533068306e-05, | |
| "loss": 1.4976, | |
| "step": 2470, | |
| "task_loss": 1.0344961881637573 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.241983413696289, | |
| "epoch": 0.9, | |
| "learning_rate": 5.368991687748464e-05, | |
| "loss": 1.5189, | |
| "step": 2480, | |
| "task_loss": 1.3167787790298462 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.9524130821228027, | |
| "epoch": 0.9, | |
| "learning_rate": 5.390675822190098e-05, | |
| "loss": 1.6949, | |
| "step": 2490, | |
| "task_loss": 1.677308201789856 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.7088284492492676, | |
| "epoch": 0.9, | |
| "learning_rate": 5.412359956631731e-05, | |
| "loss": 1.7772, | |
| "step": 2500, | |
| "task_loss": 1.0601410865783691 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "eval_exact_match": 74.76821192052981, | |
| "eval_f1": 84.28227498844669, | |
| "step": 2500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.7284023761749268, | |
| "epoch": 0.91, | |
| "learning_rate": 5.434044091073365e-05, | |
| "loss": 1.7942, | |
| "step": 2510, | |
| "task_loss": 1.2665657997131348 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.8622101545333862, | |
| "epoch": 0.91, | |
| "learning_rate": 5.455728225514998e-05, | |
| "loss": 1.6878, | |
| "step": 2520, | |
| "task_loss": 1.2940951585769653 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3830608129501343, | |
| "epoch": 0.91, | |
| "learning_rate": 5.477412359956632e-05, | |
| "loss": 1.628, | |
| "step": 2530, | |
| "task_loss": 1.0994198322296143 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.014238119125366, | |
| "epoch": 0.92, | |
| "learning_rate": 5.499096494398265e-05, | |
| "loss": 1.5324, | |
| "step": 2540, | |
| "task_loss": 1.940596103668213 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.274725914001465, | |
| "epoch": 0.92, | |
| "learning_rate": 5.520780628839899e-05, | |
| "loss": 1.6801, | |
| "step": 2550, | |
| "task_loss": 1.9991852045059204 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5515773296356201, | |
| "epoch": 0.93, | |
| "learning_rate": 5.542464763281532e-05, | |
| "loss": 1.649, | |
| "step": 2560, | |
| "task_loss": 1.3755850791931152 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5370211601257324, | |
| "epoch": 0.93, | |
| "learning_rate": 5.564148897723166e-05, | |
| "loss": 1.6046, | |
| "step": 2570, | |
| "task_loss": 1.4268473386764526 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6613398790359497, | |
| "epoch": 0.93, | |
| "learning_rate": 5.585833032164799e-05, | |
| "loss": 1.5996, | |
| "step": 2580, | |
| "task_loss": 0.9327517151832581 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.47848641872406, | |
| "epoch": 0.94, | |
| "learning_rate": 5.607517166606433e-05, | |
| "loss": 1.713, | |
| "step": 2590, | |
| "task_loss": 0.9023648500442505 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.7783399820327759, | |
| "epoch": 0.94, | |
| "learning_rate": 5.629201301048067e-05, | |
| "loss": 1.6175, | |
| "step": 2600, | |
| "task_loss": 1.1293476819992065 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.9461766481399536, | |
| "epoch": 0.94, | |
| "learning_rate": 5.6508854354897004e-05, | |
| "loss": 1.6266, | |
| "step": 2610, | |
| "task_loss": 1.0467326641082764 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.567404866218567, | |
| "epoch": 0.95, | |
| "learning_rate": 5.672569569931334e-05, | |
| "loss": 1.8742, | |
| "step": 2620, | |
| "task_loss": 1.0961357355117798 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.381565809249878, | |
| "epoch": 0.95, | |
| "learning_rate": 5.6942537043729675e-05, | |
| "loss": 1.6772, | |
| "step": 2630, | |
| "task_loss": 1.4862456321716309 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3669390678405762, | |
| "epoch": 0.95, | |
| "learning_rate": 5.715937838814601e-05, | |
| "loss": 1.6426, | |
| "step": 2640, | |
| "task_loss": 1.7144615650177002 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4433438777923584, | |
| "epoch": 0.96, | |
| "learning_rate": 5.737621973256234e-05, | |
| "loss": 1.5835, | |
| "step": 2650, | |
| "task_loss": 1.1417357921600342 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4862744808197021, | |
| "epoch": 0.96, | |
| "learning_rate": 5.759306107697868e-05, | |
| "loss": 1.7158, | |
| "step": 2660, | |
| "task_loss": 0.9948188066482544 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.210496187210083, | |
| "epoch": 0.96, | |
| "learning_rate": 5.780990242139501e-05, | |
| "loss": 1.796, | |
| "step": 2670, | |
| "task_loss": 1.136523723602295 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4428620338439941, | |
| "epoch": 0.97, | |
| "learning_rate": 5.802674376581135e-05, | |
| "loss": 1.632, | |
| "step": 2680, | |
| "task_loss": 0.7866153120994568 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.05281662940979, | |
| "epoch": 0.97, | |
| "learning_rate": 5.824358511022768e-05, | |
| "loss": 1.5928, | |
| "step": 2690, | |
| "task_loss": 0.7293530106544495 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4683610200881958, | |
| "epoch": 0.98, | |
| "learning_rate": 5.8460426454644024e-05, | |
| "loss": 1.4844, | |
| "step": 2700, | |
| "task_loss": 0.6756585836410522 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5305559635162354, | |
| "epoch": 0.98, | |
| "learning_rate": 5.867726779906035e-05, | |
| "loss": 1.5709, | |
| "step": 2710, | |
| "task_loss": 1.3757680654525757 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.54231595993042, | |
| "epoch": 0.98, | |
| "learning_rate": 5.8894109143476696e-05, | |
| "loss": 1.5264, | |
| "step": 2720, | |
| "task_loss": 1.2477138042449951 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.7188405990600586, | |
| "epoch": 0.99, | |
| "learning_rate": 5.9110950487893024e-05, | |
| "loss": 1.6459, | |
| "step": 2730, | |
| "task_loss": 1.166077971458435 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.8873350620269775, | |
| "epoch": 0.99, | |
| "learning_rate": 5.932779183230937e-05, | |
| "loss": 1.7529, | |
| "step": 2740, | |
| "task_loss": 1.92569899559021 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3297995328903198, | |
| "epoch": 0.99, | |
| "learning_rate": 5.9544633176725695e-05, | |
| "loss": 1.3638, | |
| "step": 2750, | |
| "task_loss": 1.0652023553848267 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "eval_exact_match": 76.26300851466415, | |
| "eval_f1": 85.48165328587461, | |
| "step": 2750 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.8641085624694824, | |
| "epoch": 1.0, | |
| "learning_rate": 5.976147452114203e-05, | |
| "loss": 1.628, | |
| "step": 2760, | |
| "task_loss": 1.1830246448516846 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2197985649108887, | |
| "epoch": 1.0, | |
| "learning_rate": 5.9978315865558367e-05, | |
| "loss": 1.4132, | |
| "step": 2770, | |
| "task_loss": 1.025228500366211 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6511722803115845, | |
| "epoch": 1.0, | |
| "learning_rate": 5.999995649340457e-05, | |
| "loss": 1.4993, | |
| "step": 2780, | |
| "task_loss": 1.00981605052948 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4045798778533936, | |
| "epoch": 1.01, | |
| "learning_rate": 5.9999806100397186e-05, | |
| "loss": 1.2535, | |
| "step": 2790, | |
| "task_loss": 0.8313103318214417 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.8121707439422607, | |
| "epoch": 1.01, | |
| "learning_rate": 5.999954828439778e-05, | |
| "loss": 1.4897, | |
| "step": 2800, | |
| "task_loss": 1.642676830291748 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6657578945159912, | |
| "epoch": 1.02, | |
| "learning_rate": 5.999918304632955e-05, | |
| "loss": 1.5951, | |
| "step": 2810, | |
| "task_loss": 1.2371957302093506 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5906226634979248, | |
| "epoch": 1.02, | |
| "learning_rate": 5.999871038750032e-05, | |
| "loss": 1.4819, | |
| "step": 2820, | |
| "task_loss": 1.2596104145050049 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4319016933441162, | |
| "epoch": 1.02, | |
| "learning_rate": 5.999813030960259e-05, | |
| "loss": 1.4997, | |
| "step": 2830, | |
| "task_loss": 1.103452205657959 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.490473985671997, | |
| "epoch": 1.03, | |
| "learning_rate": 5.99974428147135e-05, | |
| "loss": 1.496, | |
| "step": 2840, | |
| "task_loss": 1.228161096572876 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.737858772277832, | |
| "epoch": 1.03, | |
| "learning_rate": 5.999664790529482e-05, | |
| "loss": 1.3197, | |
| "step": 2850, | |
| "task_loss": 1.2385176420211792 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.8279550075531006, | |
| "epoch": 1.03, | |
| "learning_rate": 5.9995745584192954e-05, | |
| "loss": 1.5408, | |
| "step": 2860, | |
| "task_loss": 1.2323086261749268 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.0422871112823486, | |
| "epoch": 1.04, | |
| "learning_rate": 5.9994735854638916e-05, | |
| "loss": 1.5051, | |
| "step": 2870, | |
| "task_loss": 1.5764563083648682 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6473441123962402, | |
| "epoch": 1.04, | |
| "learning_rate": 5.999361872024835e-05, | |
| "loss": 1.431, | |
| "step": 2880, | |
| "task_loss": 1.0629370212554932 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0594135522842407, | |
| "epoch": 1.04, | |
| "learning_rate": 5.9992394185021474e-05, | |
| "loss": 1.3824, | |
| "step": 2890, | |
| "task_loss": 0.8771895170211792 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4817508459091187, | |
| "epoch": 1.05, | |
| "learning_rate": 5.999106225334308e-05, | |
| "loss": 1.3694, | |
| "step": 2900, | |
| "task_loss": 0.8263894319534302 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1033601760864258, | |
| "epoch": 1.05, | |
| "learning_rate": 5.998962292998255e-05, | |
| "loss": 1.3645, | |
| "step": 2910, | |
| "task_loss": 1.4630831480026245 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2878718376159668, | |
| "epoch": 1.06, | |
| "learning_rate": 5.99880762200938e-05, | |
| "loss": 1.4124, | |
| "step": 2920, | |
| "task_loss": 0.641461968421936 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.33697772026062, | |
| "epoch": 1.06, | |
| "learning_rate": 5.9986422129215255e-05, | |
| "loss": 1.5289, | |
| "step": 2930, | |
| "task_loss": 1.9020227193832397 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.129021167755127, | |
| "epoch": 1.06, | |
| "learning_rate": 5.998466066326988e-05, | |
| "loss": 1.4724, | |
| "step": 2940, | |
| "task_loss": 1.0405011177062988 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0331007242202759, | |
| "epoch": 1.07, | |
| "learning_rate": 5.998279182856511e-05, | |
| "loss": 1.3023, | |
| "step": 2950, | |
| "task_loss": 0.7486757040023804 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3960323333740234, | |
| "epoch": 1.07, | |
| "learning_rate": 5.9980815631792844e-05, | |
| "loss": 1.483, | |
| "step": 2960, | |
| "task_loss": 1.3152785301208496 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.7268671989440918, | |
| "epoch": 1.07, | |
| "learning_rate": 5.997873208002943e-05, | |
| "loss": 1.4653, | |
| "step": 2970, | |
| "task_loss": 1.341980218887329 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3513455390930176, | |
| "epoch": 1.08, | |
| "learning_rate": 5.997654118073564e-05, | |
| "loss": 1.4756, | |
| "step": 2980, | |
| "task_loss": 1.1631968021392822 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6115731000900269, | |
| "epoch": 1.08, | |
| "learning_rate": 5.99742429417566e-05, | |
| "loss": 1.4344, | |
| "step": 2990, | |
| "task_loss": 0.9157006740570068 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3850129842758179, | |
| "epoch": 1.08, | |
| "learning_rate": 5.997183737132184e-05, | |
| "loss": 1.553, | |
| "step": 3000, | |
| "task_loss": 1.205183744430542 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "eval_exact_match": 75.96972563859981, | |
| "eval_f1": 85.09594063703909, | |
| "step": 3000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6631824970245361, | |
| "epoch": 1.09, | |
| "learning_rate": 5.99693244780452e-05, | |
| "loss": 1.4326, | |
| "step": 3010, | |
| "task_loss": 1.2684969902038574 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.7456169128417969, | |
| "epoch": 1.09, | |
| "learning_rate": 5.996670427092481e-05, | |
| "loss": 1.4876, | |
| "step": 3020, | |
| "task_loss": 1.0792250633239746 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.8971850872039795, | |
| "epoch": 1.1, | |
| "learning_rate": 5.996397675934309e-05, | |
| "loss": 1.6011, | |
| "step": 3030, | |
| "task_loss": 1.1246172189712524 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.48433518409729, | |
| "epoch": 1.1, | |
| "learning_rate": 5.996114195306668e-05, | |
| "loss": 1.4407, | |
| "step": 3040, | |
| "task_loss": 1.1288623809814453 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5189406871795654, | |
| "epoch": 1.1, | |
| "learning_rate": 5.995819986224643e-05, | |
| "loss": 1.6002, | |
| "step": 3050, | |
| "task_loss": 1.3538494110107422 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.7961808443069458, | |
| "epoch": 1.11, | |
| "learning_rate": 5.995515049741734e-05, | |
| "loss": 1.5678, | |
| "step": 3060, | |
| "task_loss": 1.821069598197937 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3772679567337036, | |
| "epoch": 1.11, | |
| "learning_rate": 5.995199386949855e-05, | |
| "loss": 1.4492, | |
| "step": 3070, | |
| "task_loss": 1.3162784576416016 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.128143548965454, | |
| "epoch": 1.11, | |
| "learning_rate": 5.994872998979327e-05, | |
| "loss": 1.4028, | |
| "step": 3080, | |
| "task_loss": 1.0793228149414062 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6217098236083984, | |
| "epoch": 1.12, | |
| "learning_rate": 5.9945358869988796e-05, | |
| "loss": 1.5213, | |
| "step": 3090, | |
| "task_loss": 1.173330307006836 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3088253736495972, | |
| "epoch": 1.12, | |
| "learning_rate": 5.994188052215636e-05, | |
| "loss": 1.3198, | |
| "step": 3100, | |
| "task_loss": 0.9987668395042419 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6880836486816406, | |
| "epoch": 1.12, | |
| "learning_rate": 5.993829495875121e-05, | |
| "loss": 1.3887, | |
| "step": 3110, | |
| "task_loss": 1.2901586294174194 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.024364948272705, | |
| "epoch": 1.13, | |
| "learning_rate": 5.9934602192612506e-05, | |
| "loss": 1.2978, | |
| "step": 3120, | |
| "task_loss": 0.765283465385437 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4316623210906982, | |
| "epoch": 1.13, | |
| "learning_rate": 5.9930802236963245e-05, | |
| "loss": 1.5247, | |
| "step": 3130, | |
| "task_loss": 0.8690903186798096 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6122153997421265, | |
| "epoch": 1.13, | |
| "learning_rate": 5.992689510541028e-05, | |
| "loss": 1.42, | |
| "step": 3140, | |
| "task_loss": 1.2866846323013306 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.323101282119751, | |
| "epoch": 1.14, | |
| "learning_rate": 5.992288081194423e-05, | |
| "loss": 1.3262, | |
| "step": 3150, | |
| "task_loss": 1.4218419790267944 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2536742687225342, | |
| "epoch": 1.14, | |
| "learning_rate": 5.991875937093943e-05, | |
| "loss": 1.4379, | |
| "step": 3160, | |
| "task_loss": 1.2389776706695557 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0673775672912598, | |
| "epoch": 1.15, | |
| "learning_rate": 5.991453079715389e-05, | |
| "loss": 1.3706, | |
| "step": 3170, | |
| "task_loss": 0.973718523979187 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4817243814468384, | |
| "epoch": 1.15, | |
| "learning_rate": 5.991019510572925e-05, | |
| "loss": 1.3935, | |
| "step": 3180, | |
| "task_loss": 0.9717172384262085 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.444387435913086, | |
| "epoch": 1.15, | |
| "learning_rate": 5.990575231219071e-05, | |
| "loss": 1.3291, | |
| "step": 3190, | |
| "task_loss": 1.6736479997634888 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3083770275115967, | |
| "epoch": 1.16, | |
| "learning_rate": 5.9901202432446966e-05, | |
| "loss": 1.1301, | |
| "step": 3200, | |
| "task_loss": 1.0902491807937622 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2770934104919434, | |
| "epoch": 1.16, | |
| "learning_rate": 5.989654548279019e-05, | |
| "loss": 1.3383, | |
| "step": 3210, | |
| "task_loss": 0.8913994431495667 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9612149000167847, | |
| "epoch": 1.16, | |
| "learning_rate": 5.989178147989594e-05, | |
| "loss": 1.2876, | |
| "step": 3220, | |
| "task_loss": 0.866273045539856 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.750402569770813, | |
| "epoch": 1.17, | |
| "learning_rate": 5.988691044082309e-05, | |
| "loss": 1.3676, | |
| "step": 3230, | |
| "task_loss": 1.519735336303711 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3474881649017334, | |
| "epoch": 1.17, | |
| "learning_rate": 5.988193238301383e-05, | |
| "loss": 1.1273, | |
| "step": 3240, | |
| "task_loss": 0.7416272163391113 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.463036060333252, | |
| "epoch": 1.17, | |
| "learning_rate": 5.987684732429352e-05, | |
| "loss": 1.3382, | |
| "step": 3250, | |
| "task_loss": 1.3737218379974365 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "eval_exact_match": 77.60643330179754, | |
| "eval_f1": 86.28265990982867, | |
| "step": 3250 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9102663397789001, | |
| "epoch": 1.18, | |
| "learning_rate": 5.987165528287069e-05, | |
| "loss": 1.3387, | |
| "step": 3260, | |
| "task_loss": 0.9284595251083374 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6090664863586426, | |
| "epoch": 1.18, | |
| "learning_rate": 5.9866356277336964e-05, | |
| "loss": 1.293, | |
| "step": 3270, | |
| "task_loss": 1.4329043626785278 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5263841152191162, | |
| "epoch": 1.19, | |
| "learning_rate": 5.9860950326666935e-05, | |
| "loss": 1.4872, | |
| "step": 3280, | |
| "task_loss": 1.0852687358856201 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2740449905395508, | |
| "epoch": 1.19, | |
| "learning_rate": 5.985543745021821e-05, | |
| "loss": 1.3863, | |
| "step": 3290, | |
| "task_loss": 1.306997537612915 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2338653802871704, | |
| "epoch": 1.19, | |
| "learning_rate": 5.98498176677312e-05, | |
| "loss": 1.2642, | |
| "step": 3300, | |
| "task_loss": 0.6805293560028076 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0945351123809814, | |
| "epoch": 1.2, | |
| "learning_rate": 5.98440909993292e-05, | |
| "loss": 1.2649, | |
| "step": 3310, | |
| "task_loss": 0.9938749074935913 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.241112232208252, | |
| "epoch": 1.2, | |
| "learning_rate": 5.983825746551817e-05, | |
| "loss": 1.4417, | |
| "step": 3320, | |
| "task_loss": 2.230130910873413 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2390515804290771, | |
| "epoch": 1.2, | |
| "learning_rate": 5.9832317087186795e-05, | |
| "loss": 1.368, | |
| "step": 3330, | |
| "task_loss": 0.8296461701393127 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5527942180633545, | |
| "epoch": 1.21, | |
| "learning_rate": 5.982626988560631e-05, | |
| "loss": 1.4451, | |
| "step": 3340, | |
| "task_loss": 1.1470730304718018 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8781901001930237, | |
| "epoch": 1.21, | |
| "learning_rate": 5.9820115882430476e-05, | |
| "loss": 1.3569, | |
| "step": 3350, | |
| "task_loss": 0.6853946447372437 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.358176350593567, | |
| "epoch": 1.21, | |
| "learning_rate": 5.981385509969547e-05, | |
| "loss": 1.2828, | |
| "step": 3360, | |
| "task_loss": 1.4298112392425537 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2665464878082275, | |
| "epoch": 1.22, | |
| "learning_rate": 5.980748755981984e-05, | |
| "loss": 1.3211, | |
| "step": 3370, | |
| "task_loss": 1.4579181671142578 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.965965211391449, | |
| "epoch": 1.22, | |
| "learning_rate": 5.980101328560442e-05, | |
| "loss": 1.3321, | |
| "step": 3380, | |
| "task_loss": 0.7342075109481812 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5226565599441528, | |
| "epoch": 1.23, | |
| "learning_rate": 5.979443230023221e-05, | |
| "loss": 1.4229, | |
| "step": 3390, | |
| "task_loss": 1.4543688297271729 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.332014799118042, | |
| "epoch": 1.23, | |
| "learning_rate": 5.978774462726834e-05, | |
| "loss": 1.3364, | |
| "step": 3400, | |
| "task_loss": 0.9676311612129211 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0038352012634277, | |
| "epoch": 1.23, | |
| "learning_rate": 5.9780950290659965e-05, | |
| "loss": 1.2919, | |
| "step": 3410, | |
| "task_loss": 0.9453527331352234 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.033329725265503, | |
| "epoch": 1.24, | |
| "learning_rate": 5.977404931473615e-05, | |
| "loss": 1.2794, | |
| "step": 3420, | |
| "task_loss": 0.5825457572937012 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0678160190582275, | |
| "epoch": 1.24, | |
| "learning_rate": 5.976704172420787e-05, | |
| "loss": 1.2897, | |
| "step": 3430, | |
| "task_loss": 1.085493803024292 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.299585223197937, | |
| "epoch": 1.24, | |
| "learning_rate": 5.975992754416782e-05, | |
| "loss": 1.2425, | |
| "step": 3440, | |
| "task_loss": 0.9612342119216919 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2164323329925537, | |
| "epoch": 1.25, | |
| "learning_rate": 5.975270680009036e-05, | |
| "loss": 1.3221, | |
| "step": 3450, | |
| "task_loss": 1.375262975692749 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5335781574249268, | |
| "epoch": 1.25, | |
| "learning_rate": 5.974537951783148e-05, | |
| "loss": 1.4045, | |
| "step": 3460, | |
| "task_loss": 1.7295887470245361 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2506041526794434, | |
| "epoch": 1.25, | |
| "learning_rate": 5.9737945723628635e-05, | |
| "loss": 1.2224, | |
| "step": 3470, | |
| "task_loss": 1.2525482177734375 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.2599070072174072, | |
| "epoch": 1.26, | |
| "learning_rate": 5.973040544410066e-05, | |
| "loss": 1.3498, | |
| "step": 3480, | |
| "task_loss": 1.7536072731018066 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.718400239944458, | |
| "epoch": 1.26, | |
| "learning_rate": 5.972275870624773e-05, | |
| "loss": 1.2841, | |
| "step": 3490, | |
| "task_loss": 1.339259147644043 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5253331661224365, | |
| "epoch": 1.26, | |
| "learning_rate": 5.971500553745119e-05, | |
| "loss": 1.4498, | |
| "step": 3500, | |
| "task_loss": 0.9578819274902344 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "eval_exact_match": 78.38221381267739, | |
| "eval_f1": 86.75896485683346, | |
| "step": 3500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6366232633590698, | |
| "epoch": 1.27, | |
| "learning_rate": 5.9707145965473516e-05, | |
| "loss": 1.4598, | |
| "step": 3510, | |
| "task_loss": 1.8735042810440063 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0762263536453247, | |
| "epoch": 1.27, | |
| "learning_rate": 5.969918001845817e-05, | |
| "loss": 1.2466, | |
| "step": 3520, | |
| "task_loss": 0.8542821407318115 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3200879096984863, | |
| "epoch": 1.28, | |
| "learning_rate": 5.969110772492954e-05, | |
| "loss": 1.269, | |
| "step": 3530, | |
| "task_loss": 1.0112990140914917 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1145355701446533, | |
| "epoch": 1.28, | |
| "learning_rate": 5.968292911379281e-05, | |
| "loss": 1.4295, | |
| "step": 3540, | |
| "task_loss": 0.8630874752998352 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.371903657913208, | |
| "epoch": 1.28, | |
| "learning_rate": 5.967464421433385e-05, | |
| "loss": 1.3867, | |
| "step": 3550, | |
| "task_loss": 1.1606991291046143 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.486452341079712, | |
| "epoch": 1.29, | |
| "learning_rate": 5.966625305621916e-05, | |
| "loss": 1.2152, | |
| "step": 3560, | |
| "task_loss": 1.3023685216903687 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.478879690170288, | |
| "epoch": 1.29, | |
| "learning_rate": 5.965775566949571e-05, | |
| "loss": 1.238, | |
| "step": 3570, | |
| "task_loss": 1.165961503982544 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.483465552330017, | |
| "epoch": 1.29, | |
| "learning_rate": 5.964915208459085e-05, | |
| "loss": 1.3574, | |
| "step": 3580, | |
| "task_loss": 1.0961742401123047 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.294953465461731, | |
| "epoch": 1.3, | |
| "learning_rate": 5.9640442332312195e-05, | |
| "loss": 1.2928, | |
| "step": 3590, | |
| "task_loss": 0.8126479387283325 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0170869827270508, | |
| "epoch": 1.3, | |
| "learning_rate": 5.963162644384755e-05, | |
| "loss": 1.1543, | |
| "step": 3600, | |
| "task_loss": 0.8312715291976929 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5762726068496704, | |
| "epoch": 1.3, | |
| "learning_rate": 5.9622704450764756e-05, | |
| "loss": 1.2988, | |
| "step": 3610, | |
| "task_loss": 0.9494091272354126 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0313997268676758, | |
| "epoch": 1.31, | |
| "learning_rate": 5.9613676385011585e-05, | |
| "loss": 1.1966, | |
| "step": 3620, | |
| "task_loss": 0.8011406660079956 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.594609260559082, | |
| "epoch": 1.31, | |
| "learning_rate": 5.960454227891564e-05, | |
| "loss": 1.4918, | |
| "step": 3630, | |
| "task_loss": 1.3429150581359863 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.721564531326294, | |
| "epoch": 1.32, | |
| "learning_rate": 5.9595302165184246e-05, | |
| "loss": 1.423, | |
| "step": 3640, | |
| "task_loss": 1.5580472946166992 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9121531248092651, | |
| "epoch": 1.32, | |
| "learning_rate": 5.9585956076904284e-05, | |
| "loss": 1.3897, | |
| "step": 3650, | |
| "task_loss": 1.109339952468872 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6109825372695923, | |
| "epoch": 1.32, | |
| "learning_rate": 5.9576504047542156e-05, | |
| "loss": 1.41, | |
| "step": 3660, | |
| "task_loss": 0.9962501525878906 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4419324398040771, | |
| "epoch": 1.33, | |
| "learning_rate": 5.956694611094356e-05, | |
| "loss": 1.2896, | |
| "step": 3670, | |
| "task_loss": 0.6805934906005859 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3609169721603394, | |
| "epoch": 1.33, | |
| "learning_rate": 5.955728230133347e-05, | |
| "loss": 1.2873, | |
| "step": 3680, | |
| "task_loss": 0.8458003997802734 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1782557964324951, | |
| "epoch": 1.33, | |
| "learning_rate": 5.954751265331597e-05, | |
| "loss": 1.3991, | |
| "step": 3690, | |
| "task_loss": 1.114344835281372 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5857070684432983, | |
| "epoch": 1.34, | |
| "learning_rate": 5.9537637201874086e-05, | |
| "loss": 1.2406, | |
| "step": 3700, | |
| "task_loss": 1.243434190750122 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4138120412826538, | |
| "epoch": 1.34, | |
| "learning_rate": 5.952765598236975e-05, | |
| "loss": 1.3789, | |
| "step": 3710, | |
| "task_loss": 0.6327986717224121 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.529003620147705, | |
| "epoch": 1.34, | |
| "learning_rate": 5.95175690305436e-05, | |
| "loss": 1.3643, | |
| "step": 3720, | |
| "task_loss": 1.2512716054916382 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.472851037979126, | |
| "epoch": 1.35, | |
| "learning_rate": 5.950737638251488e-05, | |
| "loss": 1.311, | |
| "step": 3730, | |
| "task_loss": 1.6475411653518677 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.9043711423873901, | |
| "epoch": 1.35, | |
| "learning_rate": 5.9497078074781344e-05, | |
| "loss": 1.3004, | |
| "step": 3740, | |
| "task_loss": 1.6194053888320923 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0727344751358032, | |
| "epoch": 1.36, | |
| "learning_rate": 5.948667414421904e-05, | |
| "loss": 1.2777, | |
| "step": 3750, | |
| "task_loss": 1.061992883682251 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "eval_exact_match": 78.76064333017976, | |
| "eval_f1": 87.07184243155801, | |
| "step": 3750 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.449849247932434, | |
| "epoch": 1.36, | |
| "learning_rate": 5.947616462808226e-05, | |
| "loss": 1.4524, | |
| "step": 3760, | |
| "task_loss": 0.9752452373504639 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.321598768234253, | |
| "epoch": 1.36, | |
| "learning_rate": 5.946554956400337e-05, | |
| "loss": 1.3379, | |
| "step": 3770, | |
| "task_loss": 1.1972038745880127 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2405821084976196, | |
| "epoch": 1.37, | |
| "learning_rate": 5.945482898999269e-05, | |
| "loss": 1.3715, | |
| "step": 3780, | |
| "task_loss": 1.1690441370010376 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1473963260650635, | |
| "epoch": 1.37, | |
| "learning_rate": 5.9444002944438315e-05, | |
| "loss": 1.2458, | |
| "step": 3790, | |
| "task_loss": 0.898491382598877 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.310509443283081, | |
| "epoch": 1.37, | |
| "learning_rate": 5.943307146610606e-05, | |
| "loss": 1.353, | |
| "step": 3800, | |
| "task_loss": 1.1752980947494507 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2511919736862183, | |
| "epoch": 1.38, | |
| "learning_rate": 5.942203459413925e-05, | |
| "loss": 1.1887, | |
| "step": 3810, | |
| "task_loss": 1.156741738319397 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2756710052490234, | |
| "epoch": 1.38, | |
| "learning_rate": 5.941089236805858e-05, | |
| "loss": 1.2036, | |
| "step": 3820, | |
| "task_loss": 0.9336721301078796 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.8774323463439941, | |
| "epoch": 1.38, | |
| "learning_rate": 5.9399644827762026e-05, | |
| "loss": 1.2673, | |
| "step": 3830, | |
| "task_loss": 1.782167673110962 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3417766094207764, | |
| "epoch": 1.39, | |
| "learning_rate": 5.938829201352467e-05, | |
| "loss": 1.2705, | |
| "step": 3840, | |
| "task_loss": 1.7044801712036133 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4353686571121216, | |
| "epoch": 1.39, | |
| "learning_rate": 5.937683396599854e-05, | |
| "loss": 1.1901, | |
| "step": 3850, | |
| "task_loss": 1.3189786672592163 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0781793594360352, | |
| "epoch": 1.4, | |
| "learning_rate": 5.9365270726212497e-05, | |
| "loss": 1.2764, | |
| "step": 3860, | |
| "task_loss": 0.7336215376853943 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6718968152999878, | |
| "epoch": 1.4, | |
| "learning_rate": 5.935360233557207e-05, | |
| "loss": 1.3173, | |
| "step": 3870, | |
| "task_loss": 1.3569270372390747 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3217222690582275, | |
| "epoch": 1.4, | |
| "learning_rate": 5.934182883585932e-05, | |
| "loss": 1.3197, | |
| "step": 3880, | |
| "task_loss": 1.3436063528060913 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8721194267272949, | |
| "epoch": 1.41, | |
| "learning_rate": 5.9329950269232654e-05, | |
| "loss": 1.2029, | |
| "step": 3890, | |
| "task_loss": 0.7747288942337036 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2571134567260742, | |
| "epoch": 1.41, | |
| "learning_rate": 5.9317966678226725e-05, | |
| "loss": 1.4305, | |
| "step": 3900, | |
| "task_loss": 1.112687110900879 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2334290742874146, | |
| "epoch": 1.41, | |
| "learning_rate": 5.930587810575225e-05, | |
| "loss": 1.3896, | |
| "step": 3910, | |
| "task_loss": 0.8130580186843872 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6533408164978027, | |
| "epoch": 1.42, | |
| "learning_rate": 5.929368459509586e-05, | |
| "loss": 1.1132, | |
| "step": 3920, | |
| "task_loss": 1.5629518032073975 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.402529001235962, | |
| "epoch": 1.42, | |
| "learning_rate": 5.9281386189919965e-05, | |
| "loss": 1.1492, | |
| "step": 3930, | |
| "task_loss": 1.2752315998077393 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1418523788452148, | |
| "epoch": 1.42, | |
| "learning_rate": 5.926898293426255e-05, | |
| "loss": 1.2519, | |
| "step": 3940, | |
| "task_loss": 1.0082168579101562 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.589359164237976, | |
| "epoch": 1.43, | |
| "learning_rate": 5.925647487253707e-05, | |
| "loss": 1.2353, | |
| "step": 3950, | |
| "task_loss": 1.2787137031555176 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0001301765441895, | |
| "epoch": 1.43, | |
| "learning_rate": 5.9243862049532264e-05, | |
| "loss": 1.3728, | |
| "step": 3960, | |
| "task_loss": 0.9775519967079163 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.7581260204315186, | |
| "epoch": 1.43, | |
| "learning_rate": 5.9231144510411994e-05, | |
| "loss": 1.3053, | |
| "step": 3970, | |
| "task_loss": 1.2524868249893188 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1798157691955566, | |
| "epoch": 1.44, | |
| "learning_rate": 5.921832230071508e-05, | |
| "loss": 1.3237, | |
| "step": 3980, | |
| "task_loss": 1.0549383163452148 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2880291938781738, | |
| "epoch": 1.44, | |
| "learning_rate": 5.9205395466355186e-05, | |
| "loss": 1.2282, | |
| "step": 3990, | |
| "task_loss": 0.7487226128578186 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.314383625984192, | |
| "epoch": 1.45, | |
| "learning_rate": 5.9192364053620554e-05, | |
| "loss": 1.3069, | |
| "step": 4000, | |
| "task_loss": 1.3157868385314941 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "eval_exact_match": 78.62819299905392, | |
| "eval_f1": 87.15414215858682, | |
| "step": 4000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.793086051940918, | |
| "epoch": 1.45, | |
| "learning_rate": 5.917922810917394e-05, | |
| "loss": 1.175, | |
| "step": 4010, | |
| "task_loss": 0.5701212286949158 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.295424461364746, | |
| "epoch": 1.45, | |
| "learning_rate": 5.91659876800524e-05, | |
| "loss": 1.2627, | |
| "step": 4020, | |
| "task_loss": 0.9144801497459412 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0756990909576416, | |
| "epoch": 1.46, | |
| "learning_rate": 5.9152642813667135e-05, | |
| "loss": 1.2696, | |
| "step": 4030, | |
| "task_loss": 0.8201438188552856 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3234689235687256, | |
| "epoch": 1.46, | |
| "learning_rate": 5.913919355780329e-05, | |
| "loss": 1.3347, | |
| "step": 4040, | |
| "task_loss": 1.1706106662750244 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0304203033447266, | |
| "epoch": 1.46, | |
| "learning_rate": 5.912563996061981e-05, | |
| "loss": 1.1509, | |
| "step": 4050, | |
| "task_loss": 1.0333361625671387 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8914575576782227, | |
| "epoch": 1.47, | |
| "learning_rate": 5.9111982070649294e-05, | |
| "loss": 1.2647, | |
| "step": 4060, | |
| "task_loss": 1.0233019590377808 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5193976163864136, | |
| "epoch": 1.47, | |
| "learning_rate": 5.909821993679775e-05, | |
| "loss": 1.336, | |
| "step": 4070, | |
| "task_loss": 1.1142679452896118 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.550922155380249, | |
| "epoch": 1.47, | |
| "learning_rate": 5.908435360834451e-05, | |
| "loss": 1.2458, | |
| "step": 4080, | |
| "task_loss": 1.194461464881897 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4292857646942139, | |
| "epoch": 1.48, | |
| "learning_rate": 5.9070383134941953e-05, | |
| "loss": 1.3929, | |
| "step": 4090, | |
| "task_loss": 1.3362990617752075 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2451667785644531, | |
| "epoch": 1.48, | |
| "learning_rate": 5.9056308566615434e-05, | |
| "loss": 1.1498, | |
| "step": 4100, | |
| "task_loss": 0.7560177445411682 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4586573839187622, | |
| "epoch": 1.49, | |
| "learning_rate": 5.904212995376298e-05, | |
| "loss": 1.1929, | |
| "step": 4110, | |
| "task_loss": 1.1912312507629395 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1983109712600708, | |
| "epoch": 1.49, | |
| "learning_rate": 5.9027847347155253e-05, | |
| "loss": 1.1934, | |
| "step": 4120, | |
| "task_loss": 1.048568844795227 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.7856746912002563, | |
| "epoch": 1.49, | |
| "learning_rate": 5.901346079793525e-05, | |
| "loss": 1.25, | |
| "step": 4130, | |
| "task_loss": 1.6442539691925049 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.616405963897705, | |
| "epoch": 1.5, | |
| "learning_rate": 5.899897035761817e-05, | |
| "loss": 1.2341, | |
| "step": 4140, | |
| "task_loss": 1.356785774230957 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2377030849456787, | |
| "epoch": 1.5, | |
| "learning_rate": 5.898437607809124e-05, | |
| "loss": 1.1177, | |
| "step": 4150, | |
| "task_loss": 1.0740208625793457 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0152101516723633, | |
| "epoch": 1.5, | |
| "learning_rate": 5.896967801161349e-05, | |
| "loss": 1.3281, | |
| "step": 4160, | |
| "task_loss": 1.027454137802124 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5116093158721924, | |
| "epoch": 1.51, | |
| "learning_rate": 5.895487621081562e-05, | |
| "loss": 1.2888, | |
| "step": 4170, | |
| "task_loss": 1.2181144952774048 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1471505165100098, | |
| "epoch": 1.51, | |
| "learning_rate": 5.893997072869975e-05, | |
| "loss": 1.1994, | |
| "step": 4180, | |
| "task_loss": 0.8226631879806519 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0915307998657227, | |
| "epoch": 1.51, | |
| "learning_rate": 5.892496161863928e-05, | |
| "loss": 1.2187, | |
| "step": 4190, | |
| "task_loss": 0.5205323100090027 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2426114082336426, | |
| "epoch": 1.52, | |
| "learning_rate": 5.8909848934378674e-05, | |
| "loss": 1.1759, | |
| "step": 4200, | |
| "task_loss": 1.330606460571289 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.944818913936615, | |
| "epoch": 1.52, | |
| "learning_rate": 5.889463273003328e-05, | |
| "loss": 1.2345, | |
| "step": 4210, | |
| "task_loss": 0.6433451175689697 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1503543853759766, | |
| "epoch": 1.53, | |
| "learning_rate": 5.887931306008911e-05, | |
| "loss": 1.241, | |
| "step": 4220, | |
| "task_loss": 1.847876787185669 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0494859218597412, | |
| "epoch": 1.53, | |
| "learning_rate": 5.8863889979402696e-05, | |
| "loss": 1.2631, | |
| "step": 4230, | |
| "task_loss": 0.9382550716400146 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.224144458770752, | |
| "epoch": 1.53, | |
| "learning_rate": 5.8848363543200816e-05, | |
| "loss": 1.2842, | |
| "step": 4240, | |
| "task_loss": 1.0627460479736328 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.124106764793396, | |
| "epoch": 1.54, | |
| "learning_rate": 5.88327338070804e-05, | |
| "loss": 1.3314, | |
| "step": 4250, | |
| "task_loss": 1.1247596740722656 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "eval_exact_match": 78.56196783349101, | |
| "eval_f1": 87.1287016330811, | |
| "step": 4250 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6598260402679443, | |
| "epoch": 1.54, | |
| "learning_rate": 5.8817000827008224e-05, | |
| "loss": 1.1831, | |
| "step": 4260, | |
| "task_loss": 1.4295686483383179 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6338021755218506, | |
| "epoch": 1.54, | |
| "learning_rate": 5.88011646593208e-05, | |
| "loss": 1.3631, | |
| "step": 4270, | |
| "task_loss": 1.3802179098129272 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0921931266784668, | |
| "epoch": 1.55, | |
| "learning_rate": 5.878522536072409e-05, | |
| "loss": 1.0914, | |
| "step": 4280, | |
| "task_loss": 1.2787001132965088 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1799564361572266, | |
| "epoch": 1.55, | |
| "learning_rate": 5.876918298829337e-05, | |
| "loss": 1.1864, | |
| "step": 4290, | |
| "task_loss": 1.129196047782898 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.8012198209762573, | |
| "epoch": 1.55, | |
| "learning_rate": 5.875303759947301e-05, | |
| "loss": 1.1393, | |
| "step": 4300, | |
| "task_loss": 1.3329541683197021 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3091764450073242, | |
| "epoch": 1.56, | |
| "learning_rate": 5.873678925207624e-05, | |
| "loss": 1.3255, | |
| "step": 4310, | |
| "task_loss": 1.0994157791137695 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.598827600479126, | |
| "epoch": 1.56, | |
| "learning_rate": 5.872043800428498e-05, | |
| "loss": 1.1574, | |
| "step": 4320, | |
| "task_loss": 1.1182217597961426 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.304560899734497, | |
| "epoch": 1.56, | |
| "learning_rate": 5.870398391464961e-05, | |
| "loss": 1.153, | |
| "step": 4330, | |
| "task_loss": 1.2183809280395508 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5170609951019287, | |
| "epoch": 1.57, | |
| "learning_rate": 5.868742704208875e-05, | |
| "loss": 1.3349, | |
| "step": 4340, | |
| "task_loss": 1.345301628112793 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2407028675079346, | |
| "epoch": 1.57, | |
| "learning_rate": 5.867076744588908e-05, | |
| "loss": 1.2039, | |
| "step": 4350, | |
| "task_loss": 1.1238722801208496 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2521941661834717, | |
| "epoch": 1.58, | |
| "learning_rate": 5.8654005185705114e-05, | |
| "loss": 1.1578, | |
| "step": 4360, | |
| "task_loss": 0.9843517541885376 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4626522064208984, | |
| "epoch": 1.58, | |
| "learning_rate": 5.863714032155897e-05, | |
| "loss": 1.2691, | |
| "step": 4370, | |
| "task_loss": 0.9148292541503906 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6399775743484497, | |
| "epoch": 1.58, | |
| "learning_rate": 5.8620172913840186e-05, | |
| "loss": 1.2639, | |
| "step": 4380, | |
| "task_loss": 1.2807812690734863 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.63152277469635, | |
| "epoch": 1.59, | |
| "learning_rate": 5.860310302330548e-05, | |
| "loss": 1.3362, | |
| "step": 4390, | |
| "task_loss": 1.241642713546753 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.8616843223571777, | |
| "epoch": 1.59, | |
| "learning_rate": 5.8585930711078514e-05, | |
| "loss": 1.2129, | |
| "step": 4400, | |
| "task_loss": 1.7467364072799683 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.506812334060669, | |
| "epoch": 1.59, | |
| "learning_rate": 5.856865603864975e-05, | |
| "loss": 1.4104, | |
| "step": 4410, | |
| "task_loss": 1.2961748838424683 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5448143482208252, | |
| "epoch": 1.6, | |
| "learning_rate": 5.855127906787615e-05, | |
| "loss": 1.2444, | |
| "step": 4420, | |
| "task_loss": 0.8679540157318115 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8945345878601074, | |
| "epoch": 1.6, | |
| "learning_rate": 5.853379986098098e-05, | |
| "loss": 1.2738, | |
| "step": 4430, | |
| "task_loss": 1.0072073936462402 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8297909498214722, | |
| "epoch": 1.6, | |
| "learning_rate": 5.85162184805536e-05, | |
| "loss": 1.3369, | |
| "step": 4440, | |
| "task_loss": 0.6960026025772095 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.204261064529419, | |
| "epoch": 1.61, | |
| "learning_rate": 5.849853498954926e-05, | |
| "loss": 1.2531, | |
| "step": 4450, | |
| "task_loss": 0.7097033262252808 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0963704586029053, | |
| "epoch": 1.61, | |
| "learning_rate": 5.848074945128877e-05, | |
| "loss": 1.1261, | |
| "step": 4460, | |
| "task_loss": 1.0379323959350586 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0468240976333618, | |
| "epoch": 1.62, | |
| "learning_rate": 5.846286192945845e-05, | |
| "loss": 1.1079, | |
| "step": 4470, | |
| "task_loss": 0.7108166813850403 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8962523341178894, | |
| "epoch": 1.62, | |
| "learning_rate": 5.844487248810972e-05, | |
| "loss": 1.1465, | |
| "step": 4480, | |
| "task_loss": 0.902863621711731 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2622666358947754, | |
| "epoch": 1.62, | |
| "learning_rate": 5.8426781191659e-05, | |
| "loss": 1.1282, | |
| "step": 4490, | |
| "task_loss": 1.0960261821746826 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3841495513916016, | |
| "epoch": 1.63, | |
| "learning_rate": 5.840858810488741e-05, | |
| "loss": 1.1508, | |
| "step": 4500, | |
| "task_loss": 1.3540650606155396 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "eval_exact_match": 79.4228949858089, | |
| "eval_f1": 87.49554502449215, | |
| "step": 4500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1759483814239502, | |
| "epoch": 1.63, | |
| "learning_rate": 5.8390293292940554e-05, | |
| "loss": 1.2905, | |
| "step": 4510, | |
| "task_loss": 1.0437816381454468 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.183558702468872, | |
| "epoch": 1.63, | |
| "learning_rate": 5.837189682132831e-05, | |
| "loss": 1.1933, | |
| "step": 4520, | |
| "task_loss": 1.1283735036849976 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1170258522033691, | |
| "epoch": 1.64, | |
| "learning_rate": 5.8353398755924576e-05, | |
| "loss": 1.2729, | |
| "step": 4530, | |
| "task_loss": 1.0031014680862427 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4635521173477173, | |
| "epoch": 1.64, | |
| "learning_rate": 5.833479916296704e-05, | |
| "loss": 1.2865, | |
| "step": 4540, | |
| "task_loss": 1.0605723857879639 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9159049391746521, | |
| "epoch": 1.64, | |
| "learning_rate": 5.8316098109056905e-05, | |
| "loss": 1.119, | |
| "step": 4550, | |
| "task_loss": 0.7583830952644348 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0998435020446777, | |
| "epoch": 1.65, | |
| "learning_rate": 5.829729566115874e-05, | |
| "loss": 1.0501, | |
| "step": 4560, | |
| "task_loss": 0.7868894338607788 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 2.326512575149536, | |
| "epoch": 1.65, | |
| "learning_rate": 5.827839188660012e-05, | |
| "loss": 1.5112, | |
| "step": 4570, | |
| "task_loss": 2.1364526748657227 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4276812076568604, | |
| "epoch": 1.66, | |
| "learning_rate": 5.825938685307151e-05, | |
| "loss": 1.185, | |
| "step": 4580, | |
| "task_loss": 1.419746994972229 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2273736000061035, | |
| "epoch": 1.66, | |
| "learning_rate": 5.824028062862592e-05, | |
| "loss": 1.0967, | |
| "step": 4590, | |
| "task_loss": 0.9421610832214355 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2494888305664062, | |
| "epoch": 1.66, | |
| "learning_rate": 5.822107328167873e-05, | |
| "loss": 1.2568, | |
| "step": 4600, | |
| "task_loss": 1.162670612335205 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.387427568435669, | |
| "epoch": 1.67, | |
| "learning_rate": 5.8201764881007395e-05, | |
| "loss": 1.2342, | |
| "step": 4610, | |
| "task_loss": 1.1487003564834595 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7944004535675049, | |
| "epoch": 1.67, | |
| "learning_rate": 5.8182355495751244e-05, | |
| "loss": 1.0527, | |
| "step": 4620, | |
| "task_loss": 0.8616865873336792 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0871223211288452, | |
| "epoch": 1.67, | |
| "learning_rate": 5.8162845195411193e-05, | |
| "loss": 1.0867, | |
| "step": 4630, | |
| "task_loss": 1.0131196975708008 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1689753532409668, | |
| "epoch": 1.68, | |
| "learning_rate": 5.814323404984954e-05, | |
| "loss": 1.239, | |
| "step": 4640, | |
| "task_loss": 0.9322003126144409 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7448208928108215, | |
| "epoch": 1.68, | |
| "learning_rate": 5.8123522129289646e-05, | |
| "loss": 1.0301, | |
| "step": 4650, | |
| "task_loss": 0.4830577075481415 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9654221534729004, | |
| "epoch": 1.68, | |
| "learning_rate": 5.8103709504315755e-05, | |
| "loss": 1.1417, | |
| "step": 4660, | |
| "task_loss": 1.0902562141418457 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.245806097984314, | |
| "epoch": 1.69, | |
| "learning_rate": 5.808379624587272e-05, | |
| "loss": 1.1587, | |
| "step": 4670, | |
| "task_loss": 1.5773122310638428 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8751307129859924, | |
| "epoch": 1.69, | |
| "learning_rate": 5.806378242526572e-05, | |
| "loss": 1.2707, | |
| "step": 4680, | |
| "task_loss": 0.7810333967208862 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.968869686126709, | |
| "epoch": 1.69, | |
| "learning_rate": 5.804366811416004e-05, | |
| "loss": 1.0442, | |
| "step": 4690, | |
| "task_loss": 0.7142177224159241 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0504038333892822, | |
| "epoch": 1.7, | |
| "learning_rate": 5.8023453384580784e-05, | |
| "loss": 1.2519, | |
| "step": 4700, | |
| "task_loss": 0.9022510051727295 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1235523223876953, | |
| "epoch": 1.7, | |
| "learning_rate": 5.800313830891265e-05, | |
| "loss": 1.3013, | |
| "step": 4710, | |
| "task_loss": 0.8178448677062988 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7393571138381958, | |
| "epoch": 1.71, | |
| "learning_rate": 5.798272295989965e-05, | |
| "loss": 1.28, | |
| "step": 4720, | |
| "task_loss": 0.8073184490203857 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.7382944822311401, | |
| "epoch": 1.71, | |
| "learning_rate": 5.796220741064486e-05, | |
| "loss": 1.4239, | |
| "step": 4730, | |
| "task_loss": 1.6766483783721924 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2721589803695679, | |
| "epoch": 1.71, | |
| "learning_rate": 5.794159173461013e-05, | |
| "loss": 1.0153, | |
| "step": 4740, | |
| "task_loss": 1.1552908420562744 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.029242753982544, | |
| "epoch": 1.72, | |
| "learning_rate": 5.7920876005615866e-05, | |
| "loss": 1.2561, | |
| "step": 4750, | |
| "task_loss": 1.1739318370819092 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "eval_exact_match": 79.47019867549669, | |
| "eval_f1": 87.64686274053376, | |
| "step": 4750 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.001348614692688, | |
| "epoch": 1.72, | |
| "learning_rate": 5.790006029784072e-05, | |
| "loss": 1.0992, | |
| "step": 4760, | |
| "task_loss": 0.8562111258506775 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5794899463653564, | |
| "epoch": 1.72, | |
| "learning_rate": 5.787914468582138e-05, | |
| "loss": 1.1878, | |
| "step": 4770, | |
| "task_loss": 1.2599658966064453 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5452587604522705, | |
| "epoch": 1.73, | |
| "learning_rate": 5.7858129244452245e-05, | |
| "loss": 1.1883, | |
| "step": 4780, | |
| "task_loss": 1.063377857208252 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1784594058990479, | |
| "epoch": 1.73, | |
| "learning_rate": 5.783701404898518e-05, | |
| "loss": 1.1611, | |
| "step": 4790, | |
| "task_loss": 0.8253315091133118 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.212900996208191, | |
| "epoch": 1.73, | |
| "learning_rate": 5.781579917502926e-05, | |
| "loss": 1.2314, | |
| "step": 4800, | |
| "task_loss": 0.9298550486564636 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.448960542678833, | |
| "epoch": 1.74, | |
| "learning_rate": 5.7794484698550484e-05, | |
| "loss": 1.1093, | |
| "step": 4810, | |
| "task_loss": 1.1630914211273193 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6462247371673584, | |
| "epoch": 1.74, | |
| "learning_rate": 5.777307069587152e-05, | |
| "loss": 1.1482, | |
| "step": 4820, | |
| "task_loss": 1.3052637577056885 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3470487594604492, | |
| "epoch": 1.75, | |
| "learning_rate": 5.775155724367138e-05, | |
| "loss": 1.0766, | |
| "step": 4830, | |
| "task_loss": 1.4758813381195068 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2199983596801758, | |
| "epoch": 1.75, | |
| "learning_rate": 5.7729944418985225e-05, | |
| "loss": 1.1171, | |
| "step": 4840, | |
| "task_loss": 0.9905753135681152 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9857596755027771, | |
| "epoch": 1.75, | |
| "learning_rate": 5.770823229920403e-05, | |
| "loss": 1.1234, | |
| "step": 4850, | |
| "task_loss": 1.0089142322540283 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3423244953155518, | |
| "epoch": 1.76, | |
| "learning_rate": 5.7686420962074325e-05, | |
| "loss": 1.0579, | |
| "step": 4860, | |
| "task_loss": 0.8072549104690552 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5363335609436035, | |
| "epoch": 1.76, | |
| "learning_rate": 5.766451048569792e-05, | |
| "loss": 1.2695, | |
| "step": 4870, | |
| "task_loss": 1.405518889427185 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8228256702423096, | |
| "epoch": 1.76, | |
| "learning_rate": 5.7642500948531614e-05, | |
| "loss": 1.1095, | |
| "step": 4880, | |
| "task_loss": 0.8263888955116272 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6186543703079224, | |
| "epoch": 1.77, | |
| "learning_rate": 5.762039242938693e-05, | |
| "loss": 1.2104, | |
| "step": 4890, | |
| "task_loss": 1.1083335876464844 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.022383451461792, | |
| "epoch": 1.77, | |
| "learning_rate": 5.759818500742981e-05, | |
| "loss": 1.139, | |
| "step": 4900, | |
| "task_loss": 0.8983309268951416 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3909627199172974, | |
| "epoch": 1.77, | |
| "learning_rate": 5.757587876218039e-05, | |
| "loss": 1.1174, | |
| "step": 4910, | |
| "task_loss": 1.272632360458374 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5401933193206787, | |
| "epoch": 1.78, | |
| "learning_rate": 5.755347377351262e-05, | |
| "loss": 1.1882, | |
| "step": 4920, | |
| "task_loss": 1.1316332817077637 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4937539100646973, | |
| "epoch": 1.78, | |
| "learning_rate": 5.753097012165404e-05, | |
| "loss": 1.1599, | |
| "step": 4930, | |
| "task_loss": 1.5117436647415161 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2138440608978271, | |
| "epoch": 1.79, | |
| "learning_rate": 5.750836788718551e-05, | |
| "loss": 1.0335, | |
| "step": 4940, | |
| "task_loss": 1.3014496564865112 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9365776777267456, | |
| "epoch": 1.79, | |
| "learning_rate": 5.748566715104086e-05, | |
| "loss": 1.1542, | |
| "step": 4950, | |
| "task_loss": 0.8453970551490784 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6637686491012573, | |
| "epoch": 1.79, | |
| "learning_rate": 5.746286799450667e-05, | |
| "loss": 1.14, | |
| "step": 4960, | |
| "task_loss": 1.3181530237197876 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5403568744659424, | |
| "epoch": 1.8, | |
| "learning_rate": 5.743997049922189e-05, | |
| "loss": 1.3815, | |
| "step": 4970, | |
| "task_loss": 1.4503953456878662 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2465837001800537, | |
| "epoch": 1.8, | |
| "learning_rate": 5.741697474717765e-05, | |
| "loss": 1.0925, | |
| "step": 4980, | |
| "task_loss": 0.894507646560669 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8164676427841187, | |
| "epoch": 1.8, | |
| "learning_rate": 5.73938808207169e-05, | |
| "loss": 1.178, | |
| "step": 4990, | |
| "task_loss": 0.6441599130630493 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4209399223327637, | |
| "epoch": 1.81, | |
| "learning_rate": 5.737068880253413e-05, | |
| "loss": 1.1209, | |
| "step": 5000, | |
| "task_loss": 1.1921095848083496 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "eval_exact_match": 79.59318826868495, | |
| "eval_f1": 87.62692030696158, | |
| "step": 5000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5741052627563477, | |
| "epoch": 1.81, | |
| "learning_rate": 5.7347398775675064e-05, | |
| "loss": 1.1256, | |
| "step": 5010, | |
| "task_loss": 0.9303328990936279 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1673991680145264, | |
| "epoch": 1.81, | |
| "learning_rate": 5.7324010823536405e-05, | |
| "loss": 1.2911, | |
| "step": 5020, | |
| "task_loss": 1.5069319009780884 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5451232194900513, | |
| "epoch": 1.82, | |
| "learning_rate": 5.730052502986547e-05, | |
| "loss": 1.2214, | |
| "step": 5030, | |
| "task_loss": 1.338415503501892 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8016589879989624, | |
| "epoch": 1.82, | |
| "learning_rate": 5.727694147875996e-05, | |
| "loss": 1.2434, | |
| "step": 5040, | |
| "task_loss": 0.6144353151321411 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4259705543518066, | |
| "epoch": 1.83, | |
| "learning_rate": 5.725326025466759e-05, | |
| "loss": 1.231, | |
| "step": 5050, | |
| "task_loss": 1.2222867012023926 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8470112085342407, | |
| "epoch": 1.83, | |
| "learning_rate": 5.722948144238586e-05, | |
| "loss": 1.0969, | |
| "step": 5060, | |
| "task_loss": 1.0450043678283691 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1960935592651367, | |
| "epoch": 1.83, | |
| "learning_rate": 5.7205605127061685e-05, | |
| "loss": 1.2254, | |
| "step": 5070, | |
| "task_loss": 0.9947346448898315 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4478334188461304, | |
| "epoch": 1.84, | |
| "learning_rate": 5.718163139419111e-05, | |
| "loss": 1.2485, | |
| "step": 5080, | |
| "task_loss": 1.0019992589950562 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9755243062973022, | |
| "epoch": 1.84, | |
| "learning_rate": 5.7157560329619036e-05, | |
| "loss": 1.0562, | |
| "step": 5090, | |
| "task_loss": 1.1016393899917603 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9732208251953125, | |
| "epoch": 1.84, | |
| "learning_rate": 5.7133392019538904e-05, | |
| "loss": 1.0489, | |
| "step": 5100, | |
| "task_loss": 1.0136487483978271 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.704603910446167, | |
| "epoch": 1.85, | |
| "learning_rate": 5.7109126550492306e-05, | |
| "loss": 1.3115, | |
| "step": 5110, | |
| "task_loss": 1.5213062763214111 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.960157573223114, | |
| "epoch": 1.85, | |
| "learning_rate": 5.70847640093688e-05, | |
| "loss": 0.987, | |
| "step": 5120, | |
| "task_loss": 0.6706898212432861 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2005268335342407, | |
| "epoch": 1.85, | |
| "learning_rate": 5.706030448340552e-05, | |
| "loss": 1.1784, | |
| "step": 5130, | |
| "task_loss": 1.5834624767303467 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.155214548110962, | |
| "epoch": 1.86, | |
| "learning_rate": 5.7035748060186886e-05, | |
| "loss": 1.2087, | |
| "step": 5140, | |
| "task_loss": 1.0313799381256104 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6764129996299744, | |
| "epoch": 1.86, | |
| "learning_rate": 5.701109482764426e-05, | |
| "loss": 1.2403, | |
| "step": 5150, | |
| "task_loss": 1.207790493965149 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2213506698608398, | |
| "epoch": 1.86, | |
| "learning_rate": 5.69863448740557e-05, | |
| "loss": 1.1657, | |
| "step": 5160, | |
| "task_loss": 0.9657225608825684 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.703071117401123, | |
| "epoch": 1.87, | |
| "learning_rate": 5.6961498288045576e-05, | |
| "loss": 1.2662, | |
| "step": 5170, | |
| "task_loss": 1.522236704826355 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8341003656387329, | |
| "epoch": 1.87, | |
| "learning_rate": 5.6936555158584276e-05, | |
| "loss": 1.121, | |
| "step": 5180, | |
| "task_loss": 1.0218309164047241 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0469675064086914, | |
| "epoch": 1.88, | |
| "learning_rate": 5.6911515574987906e-05, | |
| "loss": 1.2055, | |
| "step": 5190, | |
| "task_loss": 0.9404164552688599 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4735569953918457, | |
| "epoch": 1.88, | |
| "learning_rate": 5.688637962691794e-05, | |
| "loss": 1.0771, | |
| "step": 5200, | |
| "task_loss": 1.57379150390625 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.191293478012085, | |
| "epoch": 1.88, | |
| "learning_rate": 5.6861147404380914e-05, | |
| "loss": 1.0159, | |
| "step": 5210, | |
| "task_loss": 0.8541813492774963 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.772650957107544, | |
| "epoch": 1.89, | |
| "learning_rate": 5.6835818997728116e-05, | |
| "loss": 1.1434, | |
| "step": 5220, | |
| "task_loss": 0.8448182344436646 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.254989743232727, | |
| "epoch": 1.89, | |
| "learning_rate": 5.6810394497655246e-05, | |
| "loss": 1.0368, | |
| "step": 5230, | |
| "task_loss": 1.1865497827529907 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.007494568824768, | |
| "epoch": 1.89, | |
| "learning_rate": 5.678487399520206e-05, | |
| "loss": 1.0729, | |
| "step": 5240, | |
| "task_loss": 0.6971535086631775 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3780134916305542, | |
| "epoch": 1.9, | |
| "learning_rate": 5.6759257581752135e-05, | |
| "loss": 1.1506, | |
| "step": 5250, | |
| "task_loss": 0.8896455764770508 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "eval_exact_match": 80.1608325449385, | |
| "eval_f1": 87.85411242609273, | |
| "step": 5250 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5140453577041626, | |
| "epoch": 1.9, | |
| "learning_rate": 5.673354534903244e-05, | |
| "loss": 1.2356, | |
| "step": 5260, | |
| "task_loss": 1.2429227828979492 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8260661959648132, | |
| "epoch": 1.9, | |
| "learning_rate": 5.670773738911308e-05, | |
| "loss": 1.1357, | |
| "step": 5270, | |
| "task_loss": 0.6863052845001221 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.963780164718628, | |
| "epoch": 1.91, | |
| "learning_rate": 5.668183379440692e-05, | |
| "loss": 1.334, | |
| "step": 5280, | |
| "task_loss": 1.6351208686828613 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0707981586456299, | |
| "epoch": 1.91, | |
| "learning_rate": 5.665583465766929e-05, | |
| "loss": 1.1792, | |
| "step": 5290, | |
| "task_loss": 1.136970043182373 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.8824594020843506, | |
| "epoch": 1.92, | |
| "learning_rate": 5.662974007199761e-05, | |
| "loss": 1.1516, | |
| "step": 5300, | |
| "task_loss": 1.401637077331543 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5174405574798584, | |
| "epoch": 1.92, | |
| "learning_rate": 5.660355013083112e-05, | |
| "loss": 1.2431, | |
| "step": 5310, | |
| "task_loss": 1.5356242656707764 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1089816093444824, | |
| "epoch": 1.92, | |
| "learning_rate": 5.657726492795047e-05, | |
| "loss": 1.2988, | |
| "step": 5320, | |
| "task_loss": 0.8448818922042847 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6607980728149414, | |
| "epoch": 1.93, | |
| "learning_rate": 5.655088455747745e-05, | |
| "loss": 1.1833, | |
| "step": 5330, | |
| "task_loss": 1.336568832397461 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1899293661117554, | |
| "epoch": 1.93, | |
| "learning_rate": 5.6524409113874617e-05, | |
| "loss": 1.0574, | |
| "step": 5340, | |
| "task_loss": 0.9695686101913452 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5410118103027344, | |
| "epoch": 1.93, | |
| "learning_rate": 5.649783869194495e-05, | |
| "loss": 1.176, | |
| "step": 5350, | |
| "task_loss": 1.28522527217865 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2499977350234985, | |
| "epoch": 1.94, | |
| "learning_rate": 5.647117338683155e-05, | |
| "loss": 1.1411, | |
| "step": 5360, | |
| "task_loss": 1.0919628143310547 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4456626176834106, | |
| "epoch": 1.94, | |
| "learning_rate": 5.6444413294017266e-05, | |
| "loss": 1.204, | |
| "step": 5370, | |
| "task_loss": 0.9402436017990112 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0688556432724, | |
| "epoch": 1.94, | |
| "learning_rate": 5.641755850932434e-05, | |
| "loss": 1.0496, | |
| "step": 5380, | |
| "task_loss": 0.4863385856151581 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1923037767410278, | |
| "epoch": 1.95, | |
| "learning_rate": 5.639060912891412e-05, | |
| "loss": 1.3074, | |
| "step": 5390, | |
| "task_loss": 1.196105718612671 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5121135711669922, | |
| "epoch": 1.95, | |
| "learning_rate": 5.636356524928666e-05, | |
| "loss": 1.2095, | |
| "step": 5400, | |
| "task_loss": 1.1289417743682861 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0193203687667847, | |
| "epoch": 1.96, | |
| "learning_rate": 5.6336426967280403e-05, | |
| "loss": 1.0156, | |
| "step": 5410, | |
| "task_loss": 0.6196513175964355 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1136207580566406, | |
| "epoch": 1.96, | |
| "learning_rate": 5.6309194380071825e-05, | |
| "loss": 1.0743, | |
| "step": 5420, | |
| "task_loss": 0.920132577419281 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0087127685546875, | |
| "epoch": 1.96, | |
| "learning_rate": 5.6281867585175094e-05, | |
| "loss": 1.1059, | |
| "step": 5430, | |
| "task_loss": 0.8849406242370605 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9823505282402039, | |
| "epoch": 1.97, | |
| "learning_rate": 5.62544466804417e-05, | |
| "loss": 1.1327, | |
| "step": 5440, | |
| "task_loss": 1.1544190645217896 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.929497480392456, | |
| "epoch": 1.97, | |
| "learning_rate": 5.622693176406015e-05, | |
| "loss": 1.2371, | |
| "step": 5450, | |
| "task_loss": 0.6615207195281982 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8319283723831177, | |
| "epoch": 1.97, | |
| "learning_rate": 5.619932293455555e-05, | |
| "loss": 1.1588, | |
| "step": 5460, | |
| "task_loss": 0.6978183388710022 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0234538316726685, | |
| "epoch": 1.98, | |
| "learning_rate": 5.617162029078931e-05, | |
| "loss": 1.2865, | |
| "step": 5470, | |
| "task_loss": 1.0632944107055664 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2524006366729736, | |
| "epoch": 1.98, | |
| "learning_rate": 5.614382393195878e-05, | |
| "loss": 1.1138, | |
| "step": 5480, | |
| "task_loss": 0.9473165273666382 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9574975967407227, | |
| "epoch": 1.98, | |
| "learning_rate": 5.611593395759687e-05, | |
| "loss": 1.1326, | |
| "step": 5490, | |
| "task_loss": 0.6084072589874268 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1083191633224487, | |
| "epoch": 1.99, | |
| "learning_rate": 5.6087950467571686e-05, | |
| "loss": 1.0398, | |
| "step": 5500, | |
| "task_loss": 1.5072331428527832 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "eval_exact_match": 79.93377483443709, | |
| "eval_f1": 87.66760484249721, | |
| "step": 5500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1833401918411255, | |
| "epoch": 1.99, | |
| "learning_rate": 5.605987356208624e-05, | |
| "loss": 1.1425, | |
| "step": 5510, | |
| "task_loss": 0.9314687252044678 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9431557059288025, | |
| "epoch": 1.99, | |
| "learning_rate": 5.603170334167802e-05, | |
| "loss": 1.1334, | |
| "step": 5520, | |
| "task_loss": 0.8197354078292847 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1592495441436768, | |
| "epoch": 2.0, | |
| "learning_rate": 5.6003439907218656e-05, | |
| "loss": 1.2719, | |
| "step": 5530, | |
| "task_loss": 1.1633968353271484 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9738726615905762, | |
| "epoch": 2.0, | |
| "learning_rate": 5.597508335991354e-05, | |
| "loss": 1.0092, | |
| "step": 5540, | |
| "task_loss": 0.6517045497894287 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9099946022033691, | |
| "epoch": 2.01, | |
| "learning_rate": 5.594663380130153e-05, | |
| "loss": 0.9207, | |
| "step": 5550, | |
| "task_loss": 0.8611887693405151 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8894250392913818, | |
| "epoch": 2.01, | |
| "learning_rate": 5.591809133325448e-05, | |
| "loss": 0.8773, | |
| "step": 5560, | |
| "task_loss": 0.7208254933357239 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1827757358551025, | |
| "epoch": 2.01, | |
| "learning_rate": 5.588945605797698e-05, | |
| "loss": 0.8743, | |
| "step": 5570, | |
| "task_loss": 1.253017783164978 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9436140060424805, | |
| "epoch": 2.02, | |
| "learning_rate": 5.5860728078005916e-05, | |
| "loss": 0.983, | |
| "step": 5580, | |
| "task_loss": 0.6787883639335632 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3159337043762207, | |
| "epoch": 2.02, | |
| "learning_rate": 5.583190749621014e-05, | |
| "loss": 0.9276, | |
| "step": 5590, | |
| "task_loss": 1.879063367843628 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7113268375396729, | |
| "epoch": 2.02, | |
| "learning_rate": 5.580299441579008e-05, | |
| "loss": 0.8873, | |
| "step": 5600, | |
| "task_loss": 1.2208278179168701 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8449651002883911, | |
| "epoch": 2.03, | |
| "learning_rate": 5.5773988940277416e-05, | |
| "loss": 0.8183, | |
| "step": 5610, | |
| "task_loss": 0.7712351083755493 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.184571385383606, | |
| "epoch": 2.03, | |
| "learning_rate": 5.574489117353463e-05, | |
| "loss": 0.8997, | |
| "step": 5620, | |
| "task_loss": 1.2251086235046387 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.027880311012268, | |
| "epoch": 2.03, | |
| "learning_rate": 5.571570121975472e-05, | |
| "loss": 0.8811, | |
| "step": 5630, | |
| "task_loss": 1.2804369926452637 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5541472434997559, | |
| "epoch": 2.04, | |
| "learning_rate": 5.568641918346074e-05, | |
| "loss": 1.0087, | |
| "step": 5640, | |
| "task_loss": 1.7716798782348633 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9887833595275879, | |
| "epoch": 2.04, | |
| "learning_rate": 5.565704516950552e-05, | |
| "loss": 0.9056, | |
| "step": 5650, | |
| "task_loss": 1.108952283859253 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8718394041061401, | |
| "epoch": 2.05, | |
| "learning_rate": 5.562757928307121e-05, | |
| "loss": 0.8578, | |
| "step": 5660, | |
| "task_loss": 0.7642139196395874 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.6347434520721436, | |
| "epoch": 2.05, | |
| "learning_rate": 5.559802162966897e-05, | |
| "loss": 0.9785, | |
| "step": 5670, | |
| "task_loss": 1.5460071563720703 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.013885736465454, | |
| "epoch": 2.05, | |
| "learning_rate": 5.556837231513852e-05, | |
| "loss": 0.9215, | |
| "step": 5680, | |
| "task_loss": 1.1903676986694336 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8532087802886963, | |
| "epoch": 2.06, | |
| "learning_rate": 5.553863144564781e-05, | |
| "loss": 0.9352, | |
| "step": 5690, | |
| "task_loss": 1.0140926837921143 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6742889881134033, | |
| "epoch": 2.06, | |
| "learning_rate": 5.550879912769264e-05, | |
| "loss": 0.9399, | |
| "step": 5700, | |
| "task_loss": 0.7566931843757629 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.184865117073059, | |
| "epoch": 2.06, | |
| "learning_rate": 5.5478875468096265e-05, | |
| "loss": 0.9145, | |
| "step": 5710, | |
| "task_loss": 1.0496134757995605 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8306396007537842, | |
| "epoch": 2.07, | |
| "learning_rate": 5.5448860574009015e-05, | |
| "loss": 0.8817, | |
| "step": 5720, | |
| "task_loss": 0.7623034715652466 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2012348175048828, | |
| "epoch": 2.07, | |
| "learning_rate": 5.5418754552907905e-05, | |
| "loss": 0.9289, | |
| "step": 5730, | |
| "task_loss": 1.1866190433502197 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.015893578529358, | |
| "epoch": 2.07, | |
| "learning_rate": 5.5388557512596255e-05, | |
| "loss": 0.8678, | |
| "step": 5740, | |
| "task_loss": 0.9989817142486572 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8217979669570923, | |
| "epoch": 2.08, | |
| "learning_rate": 5.535826956120332e-05, | |
| "loss": 0.9298, | |
| "step": 5750, | |
| "task_loss": 1.2064638137817383 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "eval_exact_match": 79.97161778618732, | |
| "eval_f1": 88.04204326637895, | |
| "step": 5750 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9215903878211975, | |
| "epoch": 2.08, | |
| "learning_rate": 5.532789080718388e-05, | |
| "loss": 0.9191, | |
| "step": 5760, | |
| "task_loss": 1.0276141166687012 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.038353681564331, | |
| "epoch": 2.09, | |
| "learning_rate": 5.5297421359317855e-05, | |
| "loss": 0.8936, | |
| "step": 5770, | |
| "task_loss": 1.220682144165039 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6983292102813721, | |
| "epoch": 2.09, | |
| "learning_rate": 5.5266861326709916e-05, | |
| "loss": 0.93, | |
| "step": 5780, | |
| "task_loss": 1.0011334419250488 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.005040168762207, | |
| "epoch": 2.09, | |
| "learning_rate": 5.5236210818789134e-05, | |
| "loss": 0.9048, | |
| "step": 5790, | |
| "task_loss": 0.8836946487426758 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.034933090209961, | |
| "epoch": 2.1, | |
| "learning_rate": 5.52054699453085e-05, | |
| "loss": 0.8101, | |
| "step": 5800, | |
| "task_loss": 0.8946719765663147 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0742028951644897, | |
| "epoch": 2.1, | |
| "learning_rate": 5.5174638816344634e-05, | |
| "loss": 1.0352, | |
| "step": 5810, | |
| "task_loss": 0.7151690721511841 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9790550470352173, | |
| "epoch": 2.1, | |
| "learning_rate": 5.514371754229731e-05, | |
| "loss": 0.8449, | |
| "step": 5820, | |
| "task_loss": 0.990552544593811 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0028564929962158, | |
| "epoch": 2.11, | |
| "learning_rate": 5.5112706233889095e-05, | |
| "loss": 0.8105, | |
| "step": 5830, | |
| "task_loss": 0.9065827131271362 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9125075340270996, | |
| "epoch": 2.11, | |
| "learning_rate": 5.508160500216497e-05, | |
| "loss": 0.7907, | |
| "step": 5840, | |
| "task_loss": 0.7284402251243591 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.029099464416504, | |
| "epoch": 2.11, | |
| "learning_rate": 5.50504139584919e-05, | |
| "loss": 1.1059, | |
| "step": 5850, | |
| "task_loss": 0.6841808557510376 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.064493179321289, | |
| "epoch": 2.12, | |
| "learning_rate": 5.5019133214558446e-05, | |
| "loss": 1.0304, | |
| "step": 5860, | |
| "task_loss": 1.0819746255874634 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7983106374740601, | |
| "epoch": 2.12, | |
| "learning_rate": 5.4987762882374365e-05, | |
| "loss": 0.8406, | |
| "step": 5870, | |
| "task_loss": 0.940750002861023 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7291630506515503, | |
| "epoch": 2.13, | |
| "learning_rate": 5.495630307427021e-05, | |
| "loss": 0.8193, | |
| "step": 5880, | |
| "task_loss": 0.7505497932434082 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1587085723876953, | |
| "epoch": 2.13, | |
| "learning_rate": 5.492475390289695e-05, | |
| "loss": 0.9177, | |
| "step": 5890, | |
| "task_loss": 1.006648302078247 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9014360904693604, | |
| "epoch": 2.13, | |
| "learning_rate": 5.489311548122551e-05, | |
| "loss": 0.9625, | |
| "step": 5900, | |
| "task_loss": 0.8569836616516113 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8141177892684937, | |
| "epoch": 2.14, | |
| "learning_rate": 5.486138792254644e-05, | |
| "loss": 0.9307, | |
| "step": 5910, | |
| "task_loss": 0.8875386714935303 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6669865250587463, | |
| "epoch": 2.14, | |
| "learning_rate": 5.482957134046943e-05, | |
| "loss": 0.9236, | |
| "step": 5920, | |
| "task_loss": 1.001587152481079 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8241477012634277, | |
| "epoch": 2.14, | |
| "learning_rate": 5.479766584892297e-05, | |
| "loss": 0.8674, | |
| "step": 5930, | |
| "task_loss": 1.012243628501892 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.938452959060669, | |
| "epoch": 2.15, | |
| "learning_rate": 5.476567156215392e-05, | |
| "loss": 0.9174, | |
| "step": 5940, | |
| "task_loss": 0.5267655253410339 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8367009162902832, | |
| "epoch": 2.15, | |
| "learning_rate": 5.4733588594727075e-05, | |
| "loss": 0.8657, | |
| "step": 5950, | |
| "task_loss": 0.5014982223510742 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9218536615371704, | |
| "epoch": 2.15, | |
| "learning_rate": 5.470141706152479e-05, | |
| "loss": 0.9756, | |
| "step": 5960, | |
| "task_loss": 1.0196517705917358 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.5319288372993469, | |
| "epoch": 2.16, | |
| "learning_rate": 5.466915707774656e-05, | |
| "loss": 0.8654, | |
| "step": 5970, | |
| "task_loss": 0.6182997226715088 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7957490682601929, | |
| "epoch": 2.16, | |
| "learning_rate": 5.463680875890861e-05, | |
| "loss": 0.8933, | |
| "step": 5980, | |
| "task_loss": 0.7909960746765137 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.5595934391021729, | |
| "epoch": 2.16, | |
| "learning_rate": 5.460437222084344e-05, | |
| "loss": 0.9095, | |
| "step": 5990, | |
| "task_loss": 0.6065921783447266 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3127511739730835, | |
| "epoch": 2.17, | |
| "learning_rate": 5.4571847579699476e-05, | |
| "loss": 0.9389, | |
| "step": 6000, | |
| "task_loss": 1.0244468450546265 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "eval_exact_match": 80.4162724692526, | |
| "eval_f1": 88.26245305245004, | |
| "step": 6000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9304770231246948, | |
| "epoch": 2.17, | |
| "learning_rate": 5.453923495194062e-05, | |
| "loss": 0.8205, | |
| "step": 6010, | |
| "task_loss": 0.7821296453475952 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4573397636413574, | |
| "epoch": 2.18, | |
| "learning_rate": 5.4506534454345814e-05, | |
| "loss": 0.9258, | |
| "step": 6020, | |
| "task_loss": 1.1528065204620361 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7485623359680176, | |
| "epoch": 2.18, | |
| "learning_rate": 5.4473746204008664e-05, | |
| "loss": 1.008, | |
| "step": 6030, | |
| "task_loss": 0.8168891668319702 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9035183191299438, | |
| "epoch": 2.18, | |
| "learning_rate": 5.444087031833699e-05, | |
| "loss": 0.8543, | |
| "step": 6040, | |
| "task_loss": 1.0606868267059326 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1698881387710571, | |
| "epoch": 2.19, | |
| "learning_rate": 5.4407906915052425e-05, | |
| "loss": 0.8646, | |
| "step": 6050, | |
| "task_loss": 1.4439537525177002 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8415951132774353, | |
| "epoch": 2.19, | |
| "learning_rate": 5.437485611218999e-05, | |
| "loss": 0.8619, | |
| "step": 6060, | |
| "task_loss": 0.7370737791061401 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1627850532531738, | |
| "epoch": 2.19, | |
| "learning_rate": 5.434171802809764e-05, | |
| "loss": 0.9036, | |
| "step": 6070, | |
| "task_loss": 0.8056766390800476 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7362239956855774, | |
| "epoch": 2.2, | |
| "learning_rate": 5.430849278143587e-05, | |
| "loss": 0.8756, | |
| "step": 6080, | |
| "task_loss": 1.0063245296478271 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8030550479888916, | |
| "epoch": 2.2, | |
| "learning_rate": 5.427518049117732e-05, | |
| "loss": 0.9322, | |
| "step": 6090, | |
| "task_loss": 0.8871850967407227 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0586618185043335, | |
| "epoch": 2.2, | |
| "learning_rate": 5.4241781276606274e-05, | |
| "loss": 0.8834, | |
| "step": 6100, | |
| "task_loss": 1.0324198007583618 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6400578022003174, | |
| "epoch": 2.21, | |
| "learning_rate": 5.4208295257318296e-05, | |
| "loss": 0.8739, | |
| "step": 6110, | |
| "task_loss": 0.48888489603996277 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6222153306007385, | |
| "epoch": 2.21, | |
| "learning_rate": 5.417472255321977e-05, | |
| "loss": 0.8171, | |
| "step": 6120, | |
| "task_loss": 0.6630902290344238 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6763343811035156, | |
| "epoch": 2.22, | |
| "learning_rate": 5.414106328452749e-05, | |
| "loss": 0.9668, | |
| "step": 6130, | |
| "task_loss": 0.6360142827033997 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9001832008361816, | |
| "epoch": 2.22, | |
| "learning_rate": 5.4107317571768215e-05, | |
| "loss": 0.9618, | |
| "step": 6140, | |
| "task_loss": 0.7632064819335938 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.07004714012146, | |
| "epoch": 2.22, | |
| "learning_rate": 5.407348553577823e-05, | |
| "loss": 0.8878, | |
| "step": 6150, | |
| "task_loss": 1.4211241006851196 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8392785787582397, | |
| "epoch": 2.23, | |
| "learning_rate": 5.403956729770294e-05, | |
| "loss": 0.9421, | |
| "step": 6160, | |
| "task_loss": 0.6807509660720825 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8789479732513428, | |
| "epoch": 2.23, | |
| "learning_rate": 5.4005562978996424e-05, | |
| "loss": 0.9041, | |
| "step": 6170, | |
| "task_loss": 0.6013485193252563 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9375628232955933, | |
| "epoch": 2.23, | |
| "learning_rate": 5.3971472701420984e-05, | |
| "loss": 1.005, | |
| "step": 6180, | |
| "task_loss": 0.9426907300949097 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8983448147773743, | |
| "epoch": 2.24, | |
| "learning_rate": 5.393729658704673e-05, | |
| "loss": 0.8028, | |
| "step": 6190, | |
| "task_loss": 0.7798272371292114 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8339853882789612, | |
| "epoch": 2.24, | |
| "learning_rate": 5.3903034758251136e-05, | |
| "loss": 0.8309, | |
| "step": 6200, | |
| "task_loss": 0.4220605194568634 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8184782266616821, | |
| "epoch": 2.24, | |
| "learning_rate": 5.386868733771861e-05, | |
| "loss": 0.8682, | |
| "step": 6210, | |
| "task_loss": 0.509628415107727 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.024518609046936, | |
| "epoch": 2.25, | |
| "learning_rate": 5.383425444844001e-05, | |
| "loss": 0.8848, | |
| "step": 6220, | |
| "task_loss": 1.1755342483520508 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6047682166099548, | |
| "epoch": 2.25, | |
| "learning_rate": 5.379973621371229e-05, | |
| "loss": 0.8786, | |
| "step": 6230, | |
| "task_loss": 0.5626362562179565 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2282063961029053, | |
| "epoch": 2.26, | |
| "learning_rate": 5.3765132757137976e-05, | |
| "loss": 0.9194, | |
| "step": 6240, | |
| "task_loss": 0.9038418531417847 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0795953273773193, | |
| "epoch": 2.26, | |
| "learning_rate": 5.373044420262477e-05, | |
| "loss": 0.8973, | |
| "step": 6250, | |
| "task_loss": 1.1270602941513062 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "eval_exact_match": 80.32166508987702, | |
| "eval_f1": 88.11121062075824, | |
| "step": 6250 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9112297296524048, | |
| "epoch": 2.26, | |
| "learning_rate": 5.369567067438508e-05, | |
| "loss": 0.8557, | |
| "step": 6260, | |
| "task_loss": 1.6011927127838135 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8365263938903809, | |
| "epoch": 2.27, | |
| "learning_rate": 5.3660812296935584e-05, | |
| "loss": 0.8108, | |
| "step": 6270, | |
| "task_loss": 0.6560771465301514 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1342790126800537, | |
| "epoch": 2.27, | |
| "learning_rate": 5.362586919509681e-05, | |
| "loss": 0.9093, | |
| "step": 6280, | |
| "task_loss": 1.2042913436889648 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9722185134887695, | |
| "epoch": 2.27, | |
| "learning_rate": 5.359084149399264e-05, | |
| "loss": 0.8678, | |
| "step": 6290, | |
| "task_loss": 0.9090611934661865 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0596710443496704, | |
| "epoch": 2.28, | |
| "learning_rate": 5.355572931904991e-05, | |
| "loss": 0.9807, | |
| "step": 6300, | |
| "task_loss": 1.3353559970855713 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.045316457748413, | |
| "epoch": 2.28, | |
| "learning_rate": 5.3520532795997916e-05, | |
| "loss": 0.8632, | |
| "step": 6310, | |
| "task_loss": 0.6300460696220398 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8498081564903259, | |
| "epoch": 2.28, | |
| "learning_rate": 5.348525205086801e-05, | |
| "loss": 0.9675, | |
| "step": 6320, | |
| "task_loss": 0.6726857423782349 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9325256943702698, | |
| "epoch": 2.29, | |
| "learning_rate": 5.344988720999312e-05, | |
| "loss": 0.9498, | |
| "step": 6330, | |
| "task_loss": 0.6439615488052368 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8513062000274658, | |
| "epoch": 2.29, | |
| "learning_rate": 5.3414438400007276e-05, | |
| "loss": 0.8753, | |
| "step": 6340, | |
| "task_loss": 0.7384096384048462 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8684785962104797, | |
| "epoch": 2.29, | |
| "learning_rate": 5.337890574784523e-05, | |
| "loss": 0.8852, | |
| "step": 6350, | |
| "task_loss": 0.7421172261238098 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8197358846664429, | |
| "epoch": 2.3, | |
| "learning_rate": 5.334328938074193e-05, | |
| "loss": 0.8828, | |
| "step": 6360, | |
| "task_loss": 0.8697389364242554 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9179954528808594, | |
| "epoch": 2.3, | |
| "learning_rate": 5.3307589426232076e-05, | |
| "loss": 0.8377, | |
| "step": 6370, | |
| "task_loss": 1.053146481513977 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9556145071983337, | |
| "epoch": 2.31, | |
| "learning_rate": 5.327180601214971e-05, | |
| "loss": 0.8407, | |
| "step": 6380, | |
| "task_loss": 0.9011473655700684 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7499949932098389, | |
| "epoch": 2.31, | |
| "learning_rate": 5.323593926662768e-05, | |
| "loss": 0.9696, | |
| "step": 6390, | |
| "task_loss": 0.6749441623687744 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.013110876083374, | |
| "epoch": 2.31, | |
| "learning_rate": 5.319998931809728e-05, | |
| "loss": 0.9384, | |
| "step": 6400, | |
| "task_loss": 0.854491651058197 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0834851264953613, | |
| "epoch": 2.32, | |
| "learning_rate": 5.316395629528771e-05, | |
| "loss": 0.7752, | |
| "step": 6410, | |
| "task_loss": 1.330308198928833 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.5485963821411133, | |
| "epoch": 2.32, | |
| "learning_rate": 5.312784032722561e-05, | |
| "loss": 0.9658, | |
| "step": 6420, | |
| "task_loss": 1.17985200881958 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8666309118270874, | |
| "epoch": 2.32, | |
| "learning_rate": 5.309164154323469e-05, | |
| "loss": 0.9012, | |
| "step": 6430, | |
| "task_loss": 0.88444983959198 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6017141938209534, | |
| "epoch": 2.33, | |
| "learning_rate": 5.305536007293517e-05, | |
| "loss": 0.9155, | |
| "step": 6440, | |
| "task_loss": 0.6628368496894836 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7634067535400391, | |
| "epoch": 2.33, | |
| "learning_rate": 5.301899604624336e-05, | |
| "loss": 0.8433, | |
| "step": 6450, | |
| "task_loss": 0.7978712320327759 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9220677614212036, | |
| "epoch": 2.33, | |
| "learning_rate": 5.298254959337119e-05, | |
| "loss": 0.8675, | |
| "step": 6460, | |
| "task_loss": 1.0927445888519287 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6314671039581299, | |
| "epoch": 2.34, | |
| "learning_rate": 5.294602084482573e-05, | |
| "loss": 0.8681, | |
| "step": 6470, | |
| "task_loss": 0.41532278060913086 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7226696014404297, | |
| "epoch": 2.34, | |
| "learning_rate": 5.2909409931408735e-05, | |
| "loss": 0.9045, | |
| "step": 6480, | |
| "task_loss": 0.8908495306968689 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8101774454116821, | |
| "epoch": 2.35, | |
| "learning_rate": 5.28727169842162e-05, | |
| "loss": 0.9288, | |
| "step": 6490, | |
| "task_loss": 0.5963428020477295 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7302408814430237, | |
| "epoch": 2.35, | |
| "learning_rate": 5.283594213463783e-05, | |
| "loss": 0.7634, | |
| "step": 6500, | |
| "task_loss": 0.6015718579292297 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "eval_exact_match": 80.53926206244087, | |
| "eval_f1": 88.24102075665736, | |
| "step": 6500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8358856439590454, | |
| "epoch": 2.35, | |
| "learning_rate": 5.279908551435662e-05, | |
| "loss": 0.8888, | |
| "step": 6510, | |
| "task_loss": 0.5410736799240112 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7577072381973267, | |
| "epoch": 2.36, | |
| "learning_rate": 5.276214725534839e-05, | |
| "loss": 0.9435, | |
| "step": 6520, | |
| "task_loss": 0.9352724552154541 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6489107012748718, | |
| "epoch": 2.36, | |
| "learning_rate": 5.272512748988126e-05, | |
| "loss": 0.8791, | |
| "step": 6530, | |
| "task_loss": 0.7899311780929565 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6359513998031616, | |
| "epoch": 2.36, | |
| "learning_rate": 5.268802635051522e-05, | |
| "loss": 0.7681, | |
| "step": 6540, | |
| "task_loss": 0.3422088027000427 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0363128185272217, | |
| "epoch": 2.37, | |
| "learning_rate": 5.265084397010164e-05, | |
| "loss": 0.9111, | |
| "step": 6550, | |
| "task_loss": 1.0027375221252441 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0372557640075684, | |
| "epoch": 2.37, | |
| "learning_rate": 5.26135804817828e-05, | |
| "loss": 0.8456, | |
| "step": 6560, | |
| "task_loss": 1.0056697130203247 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9141188263893127, | |
| "epoch": 2.37, | |
| "learning_rate": 5.257623601899141e-05, | |
| "loss": 0.8281, | |
| "step": 6570, | |
| "task_loss": 1.3693073987960815 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7791051864624023, | |
| "epoch": 2.38, | |
| "learning_rate": 5.253881071545013e-05, | |
| "loss": 0.8698, | |
| "step": 6580, | |
| "task_loss": 0.7371783256530762 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7637522220611572, | |
| "epoch": 2.38, | |
| "learning_rate": 5.25013047051711e-05, | |
| "loss": 0.8717, | |
| "step": 6590, | |
| "task_loss": 0.41452422738075256 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6556802988052368, | |
| "epoch": 2.39, | |
| "learning_rate": 5.2463718122455444e-05, | |
| "loss": 0.8335, | |
| "step": 6600, | |
| "task_loss": 0.6950621604919434 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9156330823898315, | |
| "epoch": 2.39, | |
| "learning_rate": 5.242605110189281e-05, | |
| "loss": 0.9741, | |
| "step": 6610, | |
| "task_loss": 0.9320003986358643 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6790286302566528, | |
| "epoch": 2.39, | |
| "learning_rate": 5.2388303778360865e-05, | |
| "loss": 0.8083, | |
| "step": 6620, | |
| "task_loss": 0.5760271549224854 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9703154563903809, | |
| "epoch": 2.4, | |
| "learning_rate": 5.235047628702483e-05, | |
| "loss": 0.9415, | |
| "step": 6630, | |
| "task_loss": 0.8003017902374268 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7981125712394714, | |
| "epoch": 2.4, | |
| "learning_rate": 5.2312568763336995e-05, | |
| "loss": 1.0161, | |
| "step": 6640, | |
| "task_loss": 0.6487212181091309 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8091559410095215, | |
| "epoch": 2.4, | |
| "learning_rate": 5.227458134303623e-05, | |
| "loss": 0.8289, | |
| "step": 6650, | |
| "task_loss": 1.0017086267471313 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7591532468795776, | |
| "epoch": 2.41, | |
| "learning_rate": 5.223651416214749e-05, | |
| "loss": 0.8141, | |
| "step": 6660, | |
| "task_loss": 0.4736844599246979 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.658312201499939, | |
| "epoch": 2.41, | |
| "learning_rate": 5.2198367356981325e-05, | |
| "loss": 0.8889, | |
| "step": 6670, | |
| "task_loss": 0.6780068278312683 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9467959403991699, | |
| "epoch": 2.41, | |
| "learning_rate": 5.2160141064133426e-05, | |
| "loss": 0.8872, | |
| "step": 6680, | |
| "task_loss": 1.043292760848999 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7465643882751465, | |
| "epoch": 2.42, | |
| "learning_rate": 5.2125669551728134e-05, | |
| "loss": 0.7908, | |
| "step": 6690, | |
| "task_loss": 0.5276561975479126 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8351773023605347, | |
| "epoch": 2.42, | |
| "learning_rate": 5.208729260962506e-05, | |
| "loss": 0.9061, | |
| "step": 6700, | |
| "task_loss": 0.9505938291549683 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.628624439239502, | |
| "epoch": 2.43, | |
| "learning_rate": 5.204883657757553e-05, | |
| "loss": 0.879, | |
| "step": 6710, | |
| "task_loss": 0.8255483508110046 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6216802000999451, | |
| "epoch": 2.43, | |
| "learning_rate": 5.201030159328252e-05, | |
| "loss": 0.8161, | |
| "step": 6720, | |
| "task_loss": 0.7303711175918579 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.187577247619629, | |
| "epoch": 2.43, | |
| "learning_rate": 5.1971687794731685e-05, | |
| "loss": 0.8769, | |
| "step": 6730, | |
| "task_loss": 0.9617449045181274 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.022446870803833, | |
| "epoch": 2.44, | |
| "learning_rate": 5.193299532019094e-05, | |
| "loss": 0.9059, | |
| "step": 6740, | |
| "task_loss": 1.0978424549102783 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7576553821563721, | |
| "epoch": 2.44, | |
| "learning_rate": 5.189422430820986e-05, | |
| "loss": 0.9146, | |
| "step": 6750, | |
| "task_loss": 0.9070881605148315 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "eval_exact_match": 80.6244087038789, | |
| "eval_f1": 88.24355008399391, | |
| "step": 6750 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3398449420928955, | |
| "epoch": 2.44, | |
| "learning_rate": 5.185537489761931e-05, | |
| "loss": 1.0719, | |
| "step": 6760, | |
| "task_loss": 1.251268744468689 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8856955766677856, | |
| "epoch": 2.45, | |
| "learning_rate": 5.181644722753083e-05, | |
| "loss": 0.8418, | |
| "step": 6770, | |
| "task_loss": 0.7368022799491882 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0002448558807373, | |
| "epoch": 2.45, | |
| "learning_rate": 5.177744143733622e-05, | |
| "loss": 0.8679, | |
| "step": 6780, | |
| "task_loss": 1.0230883359909058 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0123604536056519, | |
| "epoch": 2.45, | |
| "learning_rate": 5.173835766670701e-05, | |
| "loss": 0.9899, | |
| "step": 6790, | |
| "task_loss": 1.1285487413406372 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8969544172286987, | |
| "epoch": 2.46, | |
| "learning_rate": 5.1699196055593954e-05, | |
| "loss": 0.9321, | |
| "step": 6800, | |
| "task_loss": 0.9808671474456787 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.399101734161377, | |
| "epoch": 2.46, | |
| "learning_rate": 5.165995674422654e-05, | |
| "loss": 1.0175, | |
| "step": 6810, | |
| "task_loss": 0.9696444272994995 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7293881177902222, | |
| "epoch": 2.46, | |
| "learning_rate": 5.162063987311249e-05, | |
| "loss": 0.9052, | |
| "step": 6820, | |
| "task_loss": 0.8145143389701843 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7391337156295776, | |
| "epoch": 2.47, | |
| "learning_rate": 5.158124558303723e-05, | |
| "loss": 0.7945, | |
| "step": 6830, | |
| "task_loss": 0.7841509580612183 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6791837215423584, | |
| "epoch": 2.47, | |
| "learning_rate": 5.1541774015063435e-05, | |
| "loss": 0.8292, | |
| "step": 6840, | |
| "task_loss": 0.6428923606872559 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.803602397441864, | |
| "epoch": 2.48, | |
| "learning_rate": 5.150222531053048e-05, | |
| "loss": 0.8626, | |
| "step": 6850, | |
| "task_loss": 0.8221435546875 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7116631269454956, | |
| "epoch": 2.48, | |
| "learning_rate": 5.146259961105396e-05, | |
| "loss": 0.9257, | |
| "step": 6860, | |
| "task_loss": 0.5078368186950684 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9278137683868408, | |
| "epoch": 2.48, | |
| "learning_rate": 5.142289705852514e-05, | |
| "loss": 0.9753, | |
| "step": 6870, | |
| "task_loss": 1.4697847366333008 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.93769770860672, | |
| "epoch": 2.49, | |
| "learning_rate": 5.138311779511054e-05, | |
| "loss": 1.012, | |
| "step": 6880, | |
| "task_loss": 1.0208895206451416 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.5001537799835205, | |
| "epoch": 2.49, | |
| "learning_rate": 5.134326196325131e-05, | |
| "loss": 0.8302, | |
| "step": 6890, | |
| "task_loss": 0.9387691020965576 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.733636736869812, | |
| "epoch": 2.49, | |
| "learning_rate": 5.130332970566278e-05, | |
| "loss": 0.822, | |
| "step": 6900, | |
| "task_loss": 0.7175207734107971 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.924873948097229, | |
| "epoch": 2.5, | |
| "learning_rate": 5.1263321165334e-05, | |
| "loss": 1.0059, | |
| "step": 6910, | |
| "task_loss": 0.8183651566505432 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9852616190910339, | |
| "epoch": 2.5, | |
| "learning_rate": 5.122323648552711e-05, | |
| "loss": 0.9519, | |
| "step": 6920, | |
| "task_loss": 1.253139853477478 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8422321081161499, | |
| "epoch": 2.5, | |
| "learning_rate": 5.11830758097769e-05, | |
| "loss": 0.771, | |
| "step": 6930, | |
| "task_loss": 0.6653087139129639 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.179046392440796, | |
| "epoch": 2.51, | |
| "learning_rate": 5.114283928189032e-05, | |
| "loss": 1.021, | |
| "step": 6940, | |
| "task_loss": 0.617653489112854 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.5847468376159668, | |
| "epoch": 2.51, | |
| "learning_rate": 5.110252704594591e-05, | |
| "loss": 0.8193, | |
| "step": 6950, | |
| "task_loss": 0.5160527229309082 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0966300964355469, | |
| "epoch": 2.52, | |
| "learning_rate": 5.106213924629328e-05, | |
| "loss": 1.0079, | |
| "step": 6960, | |
| "task_loss": 0.8059948682785034 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7070465087890625, | |
| "epoch": 2.52, | |
| "learning_rate": 5.102167602755267e-05, | |
| "loss": 0.9058, | |
| "step": 6970, | |
| "task_loss": 0.5223201513290405 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9722675681114197, | |
| "epoch": 2.52, | |
| "learning_rate": 5.0981137534614325e-05, | |
| "loss": 0.8457, | |
| "step": 6980, | |
| "task_loss": 0.7618287801742554 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8757837414741516, | |
| "epoch": 2.53, | |
| "learning_rate": 5.094052391263807e-05, | |
| "loss": 0.8939, | |
| "step": 6990, | |
| "task_loss": 0.8423348665237427 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3165154457092285, | |
| "epoch": 2.53, | |
| "learning_rate": 5.089983530705272e-05, | |
| "loss": 0.8859, | |
| "step": 7000, | |
| "task_loss": 0.809664785861969 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "eval_exact_match": 80.70955534531693, | |
| "eval_f1": 88.51734461953797, | |
| "step": 7000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7622497081756592, | |
| "epoch": 2.53, | |
| "learning_rate": 5.085907186355564e-05, | |
| "loss": 0.8179, | |
| "step": 7010, | |
| "task_loss": 0.8336362242698669 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6575897932052612, | |
| "epoch": 2.54, | |
| "learning_rate": 5.081823372811212e-05, | |
| "loss": 0.894, | |
| "step": 7020, | |
| "task_loss": 0.7306101322174072 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7808577418327332, | |
| "epoch": 2.54, | |
| "learning_rate": 5.0777321046954936e-05, | |
| "loss": 0.8247, | |
| "step": 7030, | |
| "task_loss": 0.9445192217826843 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7908771634101868, | |
| "epoch": 2.54, | |
| "learning_rate": 5.073633396658378e-05, | |
| "loss": 0.8793, | |
| "step": 7040, | |
| "task_loss": 1.0309138298034668 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.759590208530426, | |
| "epoch": 2.55, | |
| "learning_rate": 5.069527263376478e-05, | |
| "loss": 0.8812, | |
| "step": 7050, | |
| "task_loss": 0.7510941028594971 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9497214555740356, | |
| "epoch": 2.55, | |
| "learning_rate": 5.06541371955299e-05, | |
| "loss": 0.9211, | |
| "step": 7060, | |
| "task_loss": 1.1385433673858643 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8788970708847046, | |
| "epoch": 2.56, | |
| "learning_rate": 5.061292779917651e-05, | |
| "loss": 0.9177, | |
| "step": 7070, | |
| "task_loss": 1.2690625190734863 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.5018882751464844, | |
| "epoch": 2.56, | |
| "learning_rate": 5.0571644592266784e-05, | |
| "loss": 0.898, | |
| "step": 7080, | |
| "task_loss": 0.3722462058067322 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0037760734558105, | |
| "epoch": 2.56, | |
| "learning_rate": 5.053028772262718e-05, | |
| "loss": 0.9261, | |
| "step": 7090, | |
| "task_loss": 0.9891495704650879 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2450032234191895, | |
| "epoch": 2.57, | |
| "learning_rate": 5.048885733834797e-05, | |
| "loss": 0.8834, | |
| "step": 7100, | |
| "task_loss": 1.3762762546539307 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.4217417240142822, | |
| "epoch": 2.57, | |
| "learning_rate": 5.044735358778261e-05, | |
| "loss": 0.8716, | |
| "step": 7110, | |
| "task_loss": 1.2648154497146606 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0889158248901367, | |
| "epoch": 2.57, | |
| "learning_rate": 5.040577661954731e-05, | |
| "loss": 0.9575, | |
| "step": 7120, | |
| "task_loss": 0.7835713624954224 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.707399845123291, | |
| "epoch": 2.58, | |
| "learning_rate": 5.0364126582520454e-05, | |
| "loss": 0.9368, | |
| "step": 7130, | |
| "task_loss": 0.6754993200302124 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8570347428321838, | |
| "epoch": 2.58, | |
| "learning_rate": 5.0322403625842056e-05, | |
| "loss": 0.8977, | |
| "step": 7140, | |
| "task_loss": 1.0160921812057495 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1405144929885864, | |
| "epoch": 2.58, | |
| "learning_rate": 5.028060789891323e-05, | |
| "loss": 0.9357, | |
| "step": 7150, | |
| "task_loss": 1.246722936630249 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8279078006744385, | |
| "epoch": 2.59, | |
| "learning_rate": 5.0238739551395684e-05, | |
| "loss": 0.8706, | |
| "step": 7160, | |
| "task_loss": 0.8700082302093506 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7681478261947632, | |
| "epoch": 2.59, | |
| "learning_rate": 5.019679873321117e-05, | |
| "loss": 0.874, | |
| "step": 7170, | |
| "task_loss": 0.9047025442123413 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7837231159210205, | |
| "epoch": 2.59, | |
| "learning_rate": 5.0154785594540905e-05, | |
| "loss": 0.7157, | |
| "step": 7180, | |
| "task_loss": 0.44350481033325195 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7090331315994263, | |
| "epoch": 2.6, | |
| "learning_rate": 5.0112700285825134e-05, | |
| "loss": 0.897, | |
| "step": 7190, | |
| "task_loss": 0.5929510593414307 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7562023401260376, | |
| "epoch": 2.6, | |
| "learning_rate": 5.007054295776246e-05, | |
| "loss": 0.8891, | |
| "step": 7200, | |
| "task_loss": 0.5599946975708008 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1266264915466309, | |
| "epoch": 2.61, | |
| "learning_rate": 5.002831376130942e-05, | |
| "loss": 1.002, | |
| "step": 7210, | |
| "task_loss": 1.0086379051208496 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6372247338294983, | |
| "epoch": 2.61, | |
| "learning_rate": 4.9986012847679855e-05, | |
| "loss": 0.8134, | |
| "step": 7220, | |
| "task_loss": 0.7226533889770508 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8293584585189819, | |
| "epoch": 2.61, | |
| "learning_rate": 4.9943640368344464e-05, | |
| "loss": 0.9293, | |
| "step": 7230, | |
| "task_loss": 1.0355051755905151 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0152957439422607, | |
| "epoch": 2.62, | |
| "learning_rate": 4.990119647503016e-05, | |
| "loss": 0.8569, | |
| "step": 7240, | |
| "task_loss": 1.2577180862426758 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9773650169372559, | |
| "epoch": 2.62, | |
| "learning_rate": 4.98586813197196e-05, | |
| "loss": 0.8754, | |
| "step": 7250, | |
| "task_loss": 1.1356110572814941 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "eval_exact_match": 81.0879848628193, | |
| "eval_f1": 88.62796907187332, | |
| "step": 7250 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7708771824836731, | |
| "epoch": 2.62, | |
| "learning_rate": 4.9816095054650605e-05, | |
| "loss": 0.8444, | |
| "step": 7260, | |
| "task_loss": 0.6651926636695862 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1072356700897217, | |
| "epoch": 2.63, | |
| "learning_rate": 4.977343783231563e-05, | |
| "loss": 0.9954, | |
| "step": 7270, | |
| "task_loss": 1.019878625869751 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0349875688552856, | |
| "epoch": 2.63, | |
| "learning_rate": 4.9730709805461207e-05, | |
| "loss": 0.8893, | |
| "step": 7280, | |
| "task_loss": 0.8438126444816589 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8099746704101562, | |
| "epoch": 2.63, | |
| "learning_rate": 4.96879111270874e-05, | |
| "loss": 0.8308, | |
| "step": 7290, | |
| "task_loss": 0.702311635017395 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2747955322265625, | |
| "epoch": 2.64, | |
| "learning_rate": 4.964504195044729e-05, | |
| "loss": 0.9728, | |
| "step": 7300, | |
| "task_loss": 1.0143351554870605 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1044409275054932, | |
| "epoch": 2.64, | |
| "learning_rate": 4.960210242904637e-05, | |
| "loss": 0.9161, | |
| "step": 7310, | |
| "task_loss": 1.0179443359375 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9350137710571289, | |
| "epoch": 2.65, | |
| "learning_rate": 4.955909271664201e-05, | |
| "loss": 0.7646, | |
| "step": 7320, | |
| "task_loss": 0.8446428775787354 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.1761360168457031, | |
| "epoch": 2.65, | |
| "learning_rate": 4.951601296724296e-05, | |
| "loss": 0.9383, | |
| "step": 7330, | |
| "task_loss": 1.650334119796753 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2493460178375244, | |
| "epoch": 2.65, | |
| "learning_rate": 4.947286333510872e-05, | |
| "loss": 0.8523, | |
| "step": 7340, | |
| "task_loss": 1.3208677768707275 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.021695852279663, | |
| "epoch": 2.66, | |
| "learning_rate": 4.942964397474906e-05, | |
| "loss": 0.8172, | |
| "step": 7350, | |
| "task_loss": 0.7468795776367188 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8202105760574341, | |
| "epoch": 2.66, | |
| "learning_rate": 4.9386355040923396e-05, | |
| "loss": 0.7931, | |
| "step": 7360, | |
| "task_loss": 0.6085814833641052 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8396191000938416, | |
| "epoch": 2.66, | |
| "learning_rate": 4.934299668864031e-05, | |
| "loss": 0.9374, | |
| "step": 7370, | |
| "task_loss": 0.8447757959365845 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8239990472793579, | |
| "epoch": 2.67, | |
| "learning_rate": 4.929956907315692e-05, | |
| "loss": 0.9837, | |
| "step": 7380, | |
| "task_loss": 0.9020982980728149 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9127852916717529, | |
| "epoch": 2.67, | |
| "learning_rate": 4.9256072349978375e-05, | |
| "loss": 0.8494, | |
| "step": 7390, | |
| "task_loss": 0.6982762813568115 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0156875848770142, | |
| "epoch": 2.67, | |
| "learning_rate": 4.92125066748573e-05, | |
| "loss": 0.8612, | |
| "step": 7400, | |
| "task_loss": 0.7961791753768921 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.091951847076416, | |
| "epoch": 2.68, | |
| "learning_rate": 4.916887220379319e-05, | |
| "loss": 0.8624, | |
| "step": 7410, | |
| "task_loss": 0.8009432554244995 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8252087831497192, | |
| "epoch": 2.68, | |
| "learning_rate": 4.912516909303193e-05, | |
| "loss": 0.8018, | |
| "step": 7420, | |
| "task_loss": 0.7994389533996582 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0799195766448975, | |
| "epoch": 2.69, | |
| "learning_rate": 4.9081397499065144e-05, | |
| "loss": 0.8456, | |
| "step": 7430, | |
| "task_loss": 0.98964923620224 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9133539199829102, | |
| "epoch": 2.69, | |
| "learning_rate": 4.9037557578629696e-05, | |
| "loss": 0.7018, | |
| "step": 7440, | |
| "task_loss": 1.0649144649505615 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7813083529472351, | |
| "epoch": 2.69, | |
| "learning_rate": 4.8993649488707116e-05, | |
| "loss": 0.7919, | |
| "step": 7450, | |
| "task_loss": 0.7633917331695557 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.905107021331787, | |
| "epoch": 2.7, | |
| "learning_rate": 4.894967338652305e-05, | |
| "loss": 0.9709, | |
| "step": 7460, | |
| "task_loss": 1.3844743967056274 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8429162502288818, | |
| "epoch": 2.7, | |
| "learning_rate": 4.890562942954664e-05, | |
| "loss": 0.8252, | |
| "step": 7470, | |
| "task_loss": 0.8414809703826904 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7957346439361572, | |
| "epoch": 2.7, | |
| "learning_rate": 4.886151777549004e-05, | |
| "loss": 0.7663, | |
| "step": 7480, | |
| "task_loss": 0.5705951452255249 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7642992734909058, | |
| "epoch": 2.71, | |
| "learning_rate": 4.8817338582307804e-05, | |
| "loss": 0.9121, | |
| "step": 7490, | |
| "task_loss": 0.5456397533416748 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.651875376701355, | |
| "epoch": 2.71, | |
| "learning_rate": 4.877309200819631e-05, | |
| "loss": 0.8589, | |
| "step": 7500, | |
| "task_loss": 1.0916081666946411 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "eval_exact_match": 81.42857142857143, | |
| "eval_f1": 88.55823715684251, | |
| "step": 7500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.2253670692443848, | |
| "epoch": 2.71, | |
| "learning_rate": 4.872877821159325e-05, | |
| "loss": 0.8689, | |
| "step": 7510, | |
| "task_loss": 0.8776388764381409 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7492198944091797, | |
| "epoch": 2.72, | |
| "learning_rate": 4.868439735117698e-05, | |
| "loss": 0.907, | |
| "step": 7520, | |
| "task_loss": 0.6415897607803345 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6262823343276978, | |
| "epoch": 2.72, | |
| "learning_rate": 4.863994958586604e-05, | |
| "loss": 0.8402, | |
| "step": 7530, | |
| "task_loss": 0.598420262336731 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.678829550743103, | |
| "epoch": 2.72, | |
| "learning_rate": 4.85954350748185e-05, | |
| "loss": 0.8539, | |
| "step": 7540, | |
| "task_loss": 0.8512382507324219 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.43244946002960205, | |
| "epoch": 2.73, | |
| "learning_rate": 4.855085397743149e-05, | |
| "loss": 0.913, | |
| "step": 7550, | |
| "task_loss": 0.4512510895729065 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9347466230392456, | |
| "epoch": 2.73, | |
| "learning_rate": 4.850620645334051e-05, | |
| "loss": 1.0302, | |
| "step": 7560, | |
| "task_loss": 1.0103384256362915 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0060513019561768, | |
| "epoch": 2.74, | |
| "learning_rate": 4.8461492662418953e-05, | |
| "loss": 1.0036, | |
| "step": 7570, | |
| "task_loss": 1.0222327709197998 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8404515981674194, | |
| "epoch": 2.74, | |
| "learning_rate": 4.8416712764777496e-05, | |
| "loss": 0.9034, | |
| "step": 7580, | |
| "task_loss": 1.377656102180481 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6688359975814819, | |
| "epoch": 2.74, | |
| "learning_rate": 4.837186692076353e-05, | |
| "loss": 0.7879, | |
| "step": 7590, | |
| "task_loss": 0.7730237245559692 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0158637762069702, | |
| "epoch": 2.75, | |
| "learning_rate": 4.832695529096059e-05, | |
| "loss": 0.8966, | |
| "step": 7600, | |
| "task_loss": 0.9950785636901855 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8134995698928833, | |
| "epoch": 2.75, | |
| "learning_rate": 4.8281978036187764e-05, | |
| "loss": 0.9182, | |
| "step": 7610, | |
| "task_loss": 0.6545308828353882 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.5976283550262451, | |
| "epoch": 2.75, | |
| "learning_rate": 4.823693531749912e-05, | |
| "loss": 0.8383, | |
| "step": 7620, | |
| "task_loss": 0.8224766254425049 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8086435794830322, | |
| "epoch": 2.76, | |
| "learning_rate": 4.8191827296183185e-05, | |
| "loss": 0.9204, | |
| "step": 7630, | |
| "task_loss": 0.6123383641242981 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.7645212411880493, | |
| "epoch": 2.76, | |
| "learning_rate": 4.814665413376226e-05, | |
| "loss": 0.783, | |
| "step": 7640, | |
| "task_loss": 0.8700419068336487 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.0566661357879639, | |
| "epoch": 2.76, | |
| "learning_rate": 4.8101415991991965e-05, | |
| "loss": 0.8458, | |
| "step": 7650, | |
| "task_loss": 1.4529619216918945 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.050924301147461, | |
| "epoch": 2.77, | |
| "learning_rate": 4.805611303286053e-05, | |
| "loss": 0.8807, | |
| "step": 7660, | |
| "task_loss": 1.085532784461975 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.5178258419036865, | |
| "epoch": 2.77, | |
| "learning_rate": 4.801074541858835e-05, | |
| "loss": 0.803, | |
| "step": 7670, | |
| "task_loss": 0.5137467384338379 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.5854527354240417, | |
| "epoch": 2.78, | |
| "learning_rate": 4.7965313311627286e-05, | |
| "loss": 0.8646, | |
| "step": 7680, | |
| "task_loss": 0.5222367644309998 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6712926030158997, | |
| "epoch": 2.78, | |
| "learning_rate": 4.7919816874660145e-05, | |
| "loss": 0.8015, | |
| "step": 7690, | |
| "task_loss": 0.6108640432357788 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.8817957639694214, | |
| "epoch": 2.78, | |
| "learning_rate": 4.787425627060011e-05, | |
| "loss": 0.8394, | |
| "step": 7700, | |
| "task_loss": 1.0321855545043945 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 1.3753621578216553, | |
| "epoch": 2.79, | |
| "learning_rate": 4.78286316625901e-05, | |
| "loss": 0.9047, | |
| "step": 7710, | |
| "task_loss": 1.332418441772461 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6856953501701355, | |
| "epoch": 2.79, | |
| "learning_rate": 4.778294321400225e-05, | |
| "loss": 0.8821, | |
| "step": 7720, | |
| "task_loss": 1.043306589126587 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.5687346458435059, | |
| "epoch": 2.79, | |
| "learning_rate": 4.773719108843727e-05, | |
| "loss": 0.8, | |
| "step": 7730, | |
| "task_loss": 0.7144980430603027 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.6648460626602173, | |
| "epoch": 2.8, | |
| "learning_rate": 4.769137544972389e-05, | |
| "loss": 0.7907, | |
| "step": 7740, | |
| "task_loss": 1.0056627988815308 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "distillation_loss": 0.9545556306838989, | |
| "epoch": 2.8, | |
| "learning_rate": 4.764549646191827e-05, | |
| "loss": 0.8531, | |
| "step": 7750, | |
| "task_loss": 1.079555869102478 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "eval_exact_match": 81.42857142857143, | |
| "eval_f1": 88.77005789642749, | |
| "step": 7750 | |
| } | |
| ], | |
| "max_steps": 55340, | |
| "num_train_epochs": 20, | |
| "total_flos": 5914832349173760.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |