Invalid JSON:Unexpected token 'I', ..."ad_norm": Infinity,
"... is not valid JSON
| { | |
| "best_global_step": 548, | |
| "best_metric": 0.9628349178910977, | |
| "best_model_checkpoint": "/tmp/finetune/l1/checkpoint-548", | |
| "epoch": 7.0, | |
| "eval_steps": 500, | |
| "global_step": 959, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.072992700729927, | |
| "grad_norm": Infinity, | |
| "learning_rate": 8.737864077669904e-07, | |
| "loss": 1.1059, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.145985401459854, | |
| "grad_norm": 44.42300796508789, | |
| "learning_rate": 1.8446601941747574e-06, | |
| "loss": 1.154, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.21897810218978103, | |
| "grad_norm": 25.56485939025879, | |
| "learning_rate": 2.8155339805825245e-06, | |
| "loss": 1.2043, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.291970802919708, | |
| "grad_norm": 21.406190872192383, | |
| "learning_rate": 3.7864077669902915e-06, | |
| "loss": 1.0319, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.36496350364963503, | |
| "grad_norm": 33.165164947509766, | |
| "learning_rate": 4.7572815533980585e-06, | |
| "loss": 0.8044, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.43795620437956206, | |
| "grad_norm": 20.365318298339844, | |
| "learning_rate": 5.728155339805825e-06, | |
| "loss": 0.7708, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.5109489051094891, | |
| "grad_norm": 22.465822219848633, | |
| "learning_rate": 6.6990291262135935e-06, | |
| "loss": 0.6323, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.583941605839416, | |
| "grad_norm": 16.91489601135254, | |
| "learning_rate": 7.66990291262136e-06, | |
| "loss": 0.4324, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.656934306569343, | |
| "grad_norm": 21.687286376953125, | |
| "learning_rate": 8.640776699029127e-06, | |
| "loss": 0.4886, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.7299270072992701, | |
| "grad_norm": 31.576948165893555, | |
| "learning_rate": 9.611650485436894e-06, | |
| "loss": 0.5687, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.8029197080291971, | |
| "grad_norm": 1.9548248052597046, | |
| "learning_rate": 1.0582524271844662e-05, | |
| "loss": 0.2169, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.8759124087591241, | |
| "grad_norm": 22.017236709594727, | |
| "learning_rate": 1.1553398058252427e-05, | |
| "loss": 0.2982, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.948905109489051, | |
| "grad_norm": 13.639375686645508, | |
| "learning_rate": 1.2524271844660197e-05, | |
| "loss": 0.2452, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_f1_macro": 0.904341174899243, | |
| "eval_f1_micro": 0.9032800672834315, | |
| "eval_f1_weighted": 0.9042588234821014, | |
| "eval_loss": 0.14684204757213593, | |
| "eval_runtime": 1.6696, | |
| "eval_samples_per_second": 327.624, | |
| "eval_steps_per_second": 20.963, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.0218978102189782, | |
| "grad_norm": 4.722266674041748, | |
| "learning_rate": 1.3495145631067962e-05, | |
| "loss": 0.1568, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.094890510948905, | |
| "grad_norm": 25.024124145507812, | |
| "learning_rate": 1.446601941747573e-05, | |
| "loss": 0.1066, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.167883211678832, | |
| "grad_norm": 5.781569957733154, | |
| "learning_rate": 1.5436893203883496e-05, | |
| "loss": 0.1277, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.2408759124087592, | |
| "grad_norm": 17.08236312866211, | |
| "learning_rate": 1.6407766990291263e-05, | |
| "loss": 0.0935, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.313868613138686, | |
| "grad_norm": 10.940235137939453, | |
| "learning_rate": 1.737864077669903e-05, | |
| "loss": 0.1571, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.3868613138686132, | |
| "grad_norm": 4.13682222366333, | |
| "learning_rate": 1.8349514563106798e-05, | |
| "loss": 0.1316, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.4598540145985401, | |
| "grad_norm": 3.4570860862731934, | |
| "learning_rate": 1.9320388349514565e-05, | |
| "loss": 0.1334, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.5328467153284673, | |
| "grad_norm": 12.481249809265137, | |
| "learning_rate": 1.9967550027041647e-05, | |
| "loss": 0.1491, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.6058394160583942, | |
| "grad_norm": 3.959305763244629, | |
| "learning_rate": 1.9859383450513794e-05, | |
| "loss": 0.1444, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.6788321167883211, | |
| "grad_norm": 2.6921675205230713, | |
| "learning_rate": 1.9751216873985937e-05, | |
| "loss": 0.0971, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.7518248175182483, | |
| "grad_norm": 3.2789011001586914, | |
| "learning_rate": 1.9643050297458088e-05, | |
| "loss": 0.0657, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.8248175182481752, | |
| "grad_norm": 47.328712463378906, | |
| "learning_rate": 1.9534883720930235e-05, | |
| "loss": 0.1203, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.897810218978102, | |
| "grad_norm": 3.395271062850952, | |
| "learning_rate": 1.942671714440238e-05, | |
| "loss": 0.1544, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.9708029197080292, | |
| "grad_norm": 12.733712196350098, | |
| "learning_rate": 1.931855056787453e-05, | |
| "loss": 0.1459, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_f1_macro": 0.9477708161509741, | |
| "eval_f1_micro": 0.9472789115646258, | |
| "eval_f1_weighted": 0.9480129474475043, | |
| "eval_loss": 0.12245870381593704, | |
| "eval_runtime": 1.7206, | |
| "eval_samples_per_second": 317.907, | |
| "eval_steps_per_second": 20.341, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 2.0437956204379564, | |
| "grad_norm": 10.530322074890137, | |
| "learning_rate": 1.9210383991346675e-05, | |
| "loss": 0.1021, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.116788321167883, | |
| "grad_norm": 2.0978612899780273, | |
| "learning_rate": 1.9102217414818822e-05, | |
| "loss": 0.0758, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 2.18978102189781, | |
| "grad_norm": 9.21833610534668, | |
| "learning_rate": 1.899405083829097e-05, | |
| "loss": 0.0923, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.2627737226277373, | |
| "grad_norm": 5.938577651977539, | |
| "learning_rate": 1.8885884261763116e-05, | |
| "loss": 0.0452, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 2.335766423357664, | |
| "grad_norm": 3.3875350952148438, | |
| "learning_rate": 1.8777717685235263e-05, | |
| "loss": 0.0975, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 2.408759124087591, | |
| "grad_norm": 14.354475021362305, | |
| "learning_rate": 1.866955110870741e-05, | |
| "loss": 0.0924, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 2.4817518248175183, | |
| "grad_norm": 1.4160842895507812, | |
| "learning_rate": 1.8561384532179557e-05, | |
| "loss": 0.0611, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 2.554744525547445, | |
| "grad_norm": 4.338066577911377, | |
| "learning_rate": 1.8453217955651707e-05, | |
| "loss": 0.0775, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.627737226277372, | |
| "grad_norm": 2.572096347808838, | |
| "learning_rate": 1.8345051379123854e-05, | |
| "loss": 0.0453, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 2.7007299270072993, | |
| "grad_norm": 1.1255489587783813, | |
| "learning_rate": 1.8236884802596e-05, | |
| "loss": 0.0324, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 2.7737226277372264, | |
| "grad_norm": 15.42068099975586, | |
| "learning_rate": 1.8128718226068148e-05, | |
| "loss": 0.0729, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 2.846715328467153, | |
| "grad_norm": 1.9290357828140259, | |
| "learning_rate": 1.8020551649540295e-05, | |
| "loss": 0.1018, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 2.9197080291970803, | |
| "grad_norm": 1.0439828634262085, | |
| "learning_rate": 1.791238507301244e-05, | |
| "loss": 0.0856, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.9927007299270074, | |
| "grad_norm": 10.982110977172852, | |
| "learning_rate": 1.7804218496484588e-05, | |
| "loss": 0.077, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_f1_macro": 0.9476025336319455, | |
| "eval_f1_micro": 0.9478260869565217, | |
| "eval_f1_weighted": 0.947698687741686, | |
| "eval_loss": 0.14635968208312988, | |
| "eval_runtime": 1.6325, | |
| "eval_samples_per_second": 335.077, | |
| "eval_steps_per_second": 21.44, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 3.065693430656934, | |
| "grad_norm": 1.0763516426086426, | |
| "learning_rate": 1.7696051919956735e-05, | |
| "loss": 0.0499, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 3.1386861313868613, | |
| "grad_norm": 4.268288612365723, | |
| "learning_rate": 1.7587885343428882e-05, | |
| "loss": 0.0415, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 3.2116788321167884, | |
| "grad_norm": 1.0834965705871582, | |
| "learning_rate": 1.747971876690103e-05, | |
| "loss": 0.0505, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 3.2846715328467155, | |
| "grad_norm": 1.1049467325210571, | |
| "learning_rate": 1.7371552190373176e-05, | |
| "loss": 0.0495, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 3.3576642335766422, | |
| "grad_norm": 4.047983646392822, | |
| "learning_rate": 1.7263385613845323e-05, | |
| "loss": 0.0615, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 3.4306569343065694, | |
| "grad_norm": 2.435213327407837, | |
| "learning_rate": 1.715521903731747e-05, | |
| "loss": 0.043, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 3.5036496350364965, | |
| "grad_norm": 2.0357930660247803, | |
| "learning_rate": 1.7047052460789617e-05, | |
| "loss": 0.1102, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 3.576642335766423, | |
| "grad_norm": 30.20710563659668, | |
| "learning_rate": 1.6938885884261763e-05, | |
| "loss": 0.0407, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 3.6496350364963503, | |
| "grad_norm": 1.8748276233673096, | |
| "learning_rate": 1.683071930773391e-05, | |
| "loss": 0.0748, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.7226277372262775, | |
| "grad_norm": 2.229926347732544, | |
| "learning_rate": 1.6722552731206057e-05, | |
| "loss": 0.0328, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 3.795620437956204, | |
| "grad_norm": 0.9702017903327942, | |
| "learning_rate": 1.6614386154678208e-05, | |
| "loss": 0.0491, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 3.8686131386861313, | |
| "grad_norm": 0.7791664600372314, | |
| "learning_rate": 1.6506219578150354e-05, | |
| "loss": 0.0218, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 3.9416058394160585, | |
| "grad_norm": 1.9696235656738281, | |
| "learning_rate": 1.63980530016225e-05, | |
| "loss": 0.0641, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_f1_macro": 0.9628456567083955, | |
| "eval_f1_micro": 0.9628349178910977, | |
| "eval_f1_weighted": 0.9628281230678446, | |
| "eval_loss": 0.12824897468090057, | |
| "eval_runtime": 1.8072, | |
| "eval_samples_per_second": 302.683, | |
| "eval_steps_per_second": 19.367, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 4.014598540145985, | |
| "grad_norm": 0.8119364976882935, | |
| "learning_rate": 1.6289886425094648e-05, | |
| "loss": 0.0824, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 4.087591240875913, | |
| "grad_norm": 4.590900421142578, | |
| "learning_rate": 1.6181719848566795e-05, | |
| "loss": 0.0301, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 4.160583941605839, | |
| "grad_norm": 0.6132882237434387, | |
| "learning_rate": 1.6073553272038942e-05, | |
| "loss": 0.025, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 4.233576642335766, | |
| "grad_norm": 0.6952378153800964, | |
| "learning_rate": 1.596538669551109e-05, | |
| "loss": 0.0384, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 4.306569343065694, | |
| "grad_norm": 0.3461630940437317, | |
| "learning_rate": 1.5857220118983236e-05, | |
| "loss": 0.0213, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 4.37956204379562, | |
| "grad_norm": 1.4611117839813232, | |
| "learning_rate": 1.5749053542455383e-05, | |
| "loss": 0.0389, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 4.452554744525547, | |
| "grad_norm": 0.4132407009601593, | |
| "learning_rate": 1.564088696592753e-05, | |
| "loss": 0.0263, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 4.525547445255475, | |
| "grad_norm": 0.4994758069515228, | |
| "learning_rate": 1.5532720389399676e-05, | |
| "loss": 0.0191, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 4.598540145985401, | |
| "grad_norm": 4.1316447257995605, | |
| "learning_rate": 1.5424553812871823e-05, | |
| "loss": 0.0593, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 4.671532846715328, | |
| "grad_norm": 1.5681731700897217, | |
| "learning_rate": 1.531638723634397e-05, | |
| "loss": 0.0199, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 4.744525547445256, | |
| "grad_norm": 1.527457594871521, | |
| "learning_rate": 1.5208220659816117e-05, | |
| "loss": 0.0423, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 4.817518248175182, | |
| "grad_norm": 0.8105906248092651, | |
| "learning_rate": 1.5100054083288264e-05, | |
| "loss": 0.0468, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 4.89051094890511, | |
| "grad_norm": 1.1720935106277466, | |
| "learning_rate": 1.4991887506760413e-05, | |
| "loss": 0.0455, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 4.963503649635037, | |
| "grad_norm": 1.5297259092330933, | |
| "learning_rate": 1.488372093023256e-05, | |
| "loss": 0.0225, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_f1_macro": 0.9531422304959202, | |
| "eval_f1_micro": 0.9528706083976007, | |
| "eval_f1_weighted": 0.9531787151639762, | |
| "eval_loss": 0.1541963517665863, | |
| "eval_runtime": 1.637, | |
| "eval_samples_per_second": 334.141, | |
| "eval_steps_per_second": 21.38, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 5.036496350364963, | |
| "grad_norm": 1.7948929071426392, | |
| "learning_rate": 1.4775554353704706e-05, | |
| "loss": 0.0405, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 5.109489051094891, | |
| "grad_norm": 0.8266754150390625, | |
| "learning_rate": 1.4667387777176853e-05, | |
| "loss": 0.0334, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 5.182481751824818, | |
| "grad_norm": 1.0670099258422852, | |
| "learning_rate": 1.4559221200649e-05, | |
| "loss": 0.018, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 5.255474452554744, | |
| "grad_norm": 5.6616291999816895, | |
| "learning_rate": 1.4451054624121149e-05, | |
| "loss": 0.025, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 5.328467153284672, | |
| "grad_norm": 16.192676544189453, | |
| "learning_rate": 1.4342888047593296e-05, | |
| "loss": 0.0223, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 5.401459854014599, | |
| "grad_norm": 0.4029748737812042, | |
| "learning_rate": 1.4234721471065443e-05, | |
| "loss": 0.0366, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 5.474452554744525, | |
| "grad_norm": 11.750133514404297, | |
| "learning_rate": 1.412655489453759e-05, | |
| "loss": 0.0177, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 5.547445255474453, | |
| "grad_norm": 12.109711647033691, | |
| "learning_rate": 1.4018388318009736e-05, | |
| "loss": 0.0731, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 5.62043795620438, | |
| "grad_norm": 98.53327941894531, | |
| "learning_rate": 1.3910221741481885e-05, | |
| "loss": 0.0499, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 5.693430656934306, | |
| "grad_norm": 0.7491104602813721, | |
| "learning_rate": 1.3802055164954032e-05, | |
| "loss": 0.0199, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 5.766423357664234, | |
| "grad_norm": 0.5023486614227295, | |
| "learning_rate": 1.3693888588426179e-05, | |
| "loss": 0.0122, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 5.839416058394161, | |
| "grad_norm": 0.5226187109947205, | |
| "learning_rate": 1.3585722011898324e-05, | |
| "loss": 0.0304, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 5.912408759124087, | |
| "grad_norm": 0.4328407347202301, | |
| "learning_rate": 1.347755543537047e-05, | |
| "loss": 0.0125, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 5.985401459854015, | |
| "grad_norm": 0.6504618525505066, | |
| "learning_rate": 1.3369388858842618e-05, | |
| "loss": 0.0446, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_f1_macro": 0.9569069708243072, | |
| "eval_f1_micro": 0.9568965517241379, | |
| "eval_f1_weighted": 0.9569308909711778, | |
| "eval_loss": 0.15012916922569275, | |
| "eval_runtime": 1.6536, | |
| "eval_samples_per_second": 330.786, | |
| "eval_steps_per_second": 21.165, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 6.0583941605839415, | |
| "grad_norm": 7.894331455230713, | |
| "learning_rate": 1.3261222282314765e-05, | |
| "loss": 0.0259, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 6.131386861313868, | |
| "grad_norm": 1.5092934370040894, | |
| "learning_rate": 1.3153055705786913e-05, | |
| "loss": 0.0303, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 6.204379562043796, | |
| "grad_norm": 7.92203950881958, | |
| "learning_rate": 1.304488912925906e-05, | |
| "loss": 0.0292, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 6.2773722627737225, | |
| "grad_norm": 0.5470165610313416, | |
| "learning_rate": 1.2936722552731207e-05, | |
| "loss": 0.0386, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 6.350364963503649, | |
| "grad_norm": 0.37207308411598206, | |
| "learning_rate": 1.2828555976203354e-05, | |
| "loss": 0.0121, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 6.423357664233577, | |
| "grad_norm": 0.2992991805076599, | |
| "learning_rate": 1.27203893996755e-05, | |
| "loss": 0.0234, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 6.4963503649635035, | |
| "grad_norm": 2.771857261657715, | |
| "learning_rate": 1.261222282314765e-05, | |
| "loss": 0.0132, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 6.569343065693431, | |
| "grad_norm": 0.9063857793807983, | |
| "learning_rate": 1.2504056246619796e-05, | |
| "loss": 0.0474, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 6.642335766423358, | |
| "grad_norm": 1.0955673456192017, | |
| "learning_rate": 1.2395889670091943e-05, | |
| "loss": 0.0387, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 6.7153284671532845, | |
| "grad_norm": 2.7606654167175293, | |
| "learning_rate": 1.228772309356409e-05, | |
| "loss": 0.0229, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 6.788321167883212, | |
| "grad_norm": 1.1703284978866577, | |
| "learning_rate": 1.2179556517036237e-05, | |
| "loss": 0.0501, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 6.861313868613139, | |
| "grad_norm": 0.3500863015651703, | |
| "learning_rate": 1.2071389940508384e-05, | |
| "loss": 0.0107, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 6.934306569343065, | |
| "grad_norm": 0.4822583794593811, | |
| "learning_rate": 1.1963223363980532e-05, | |
| "loss": 0.0224, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_f1_macro": 0.9524315396877875, | |
| "eval_f1_micro": 0.9522983521248916, | |
| "eval_f1_weighted": 0.9524254856520143, | |
| "eval_loss": 0.1707778424024582, | |
| "eval_runtime": 1.7307, | |
| "eval_samples_per_second": 316.065, | |
| "eval_steps_per_second": 20.224, | |
| "step": 959 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2055, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 15, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 3, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 3 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 201055651468560.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |