| { |
| "best_global_step": 500, |
| "best_metric": 1.32337415, |
| "best_model_checkpoint": "/NHNHOME/data/sanghyeok/qwen-cua/runs/probe2000_idm_code_type_text/v0-20260621-113040/checkpoint-500", |
| "epoch": 0.11531531531531532, |
| "eval_steps": 500, |
| "global_step": 500, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.00023063063063063062, |
| "grad_norm": 1594.41796875, |
| "learning_rate": 1.0000000000000001e-07, |
| "loss": 78.32112121582031, |
| "loss_idm": 4.875, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 1.8660085201263428, |
| "loss_idm_type": 3.003211200237274, |
| "loss_sft": 0.013178251218050718, |
| "step": 1, |
| "token_acc": 0.855064946292334 |
| }, |
| { |
| "epoch": 0.0023063063063063064, |
| "grad_norm": 1612.195068359375, |
| "learning_rate": 1.0000000000000002e-06, |
| "loss": 75.12723795572917, |
| "loss_idm": 4.256944444444445, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 1.771455658806695, |
| "loss_idm_type": 2.897422472635905, |
| "loss_sft": 0.013165258698993258, |
| "step": 10, |
| "token_acc": 0.8557998765011616 |
| }, |
| { |
| "epoch": 0.004612612612612613, |
| "grad_norm": 353.6252136230469, |
| "learning_rate": 2.0000000000000003e-06, |
| "loss": 49.47017822265625, |
| "loss_idm": 2.946875, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 1.6457603931427003, |
| "loss_idm_type": 1.4199507236480713, |
| "loss_sft": 0.01303936094045639, |
| "step": 20, |
| "token_acc": 0.8574650963417557 |
| }, |
| { |
| "epoch": 0.006918918918918919, |
| "grad_norm": 126.47821044921875, |
| "learning_rate": 3e-06, |
| "loss": 43.140435791015626, |
| "loss_idm": 2.525, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 1.6730948686599731, |
| "loss_idm_type": 0.9972076416015625, |
| "loss_sft": 0.013039363175630569, |
| "step": 30, |
| "token_acc": 0.8573986811865872 |
| }, |
| { |
| "epoch": 0.009225225225225226, |
| "grad_norm": 162.51278686523438, |
| "learning_rate": 4.000000000000001e-06, |
| "loss": 39.47804260253906, |
| "loss_idm": 2.31875, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 1.5029781103134154, |
| "loss_idm_type": 0.9386286258697509, |
| "loss_sft": 0.013111228495836258, |
| "step": 40, |
| "token_acc": 0.856925661335508 |
| }, |
| { |
| "epoch": 0.011531531531531532, |
| "grad_norm": 356.9111633300781, |
| "learning_rate": 5e-06, |
| "loss": 36.0298828125, |
| "loss_idm": 2.125, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 1.4550552129745484, |
| "loss_idm_type": 0.7688171148300171, |
| "loss_sft": 0.014152135699987411, |
| "step": 50, |
| "token_acc": 0.8486634001116418 |
| }, |
| { |
| "epoch": 0.013837837837837838, |
| "grad_norm": 323.67437744140625, |
| "learning_rate": 6e-06, |
| "loss": 27.128363037109374, |
| "loss_idm": 1.6203125, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 1.0920132637023925, |
| "loss_idm_type": 0.574056351184845, |
| "loss_sft": 0.01448214016854763, |
| "step": 60, |
| "token_acc": 0.8464492900583126 |
| }, |
| { |
| "epoch": 0.016144144144144144, |
| "grad_norm": 259.5957336425781, |
| "learning_rate": 7e-06, |
| "loss": 28.27000732421875, |
| "loss_idm": 1.675, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 1.0847449064254762, |
| "loss_idm_type": 0.6541952967643738, |
| "loss_sft": 0.014033125713467598, |
| "step": 70, |
| "token_acc": 0.8503391363046473 |
| }, |
| { |
| "epoch": 0.01845045045045045, |
| "grad_norm": 144.2830810546875, |
| "learning_rate": 8.000000000000001e-06, |
| "loss": 24.161322021484374, |
| "loss_idm": 1.415625, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.8819304466247558, |
| "loss_idm_type": 0.600779139995575, |
| "loss_sft": 0.013908731937408447, |
| "step": 80, |
| "token_acc": 0.8504072708357134 |
| }, |
| { |
| "epoch": 0.020756756756756756, |
| "grad_norm": 273.4071350097656, |
| "learning_rate": 9e-06, |
| "loss": 23.594862365722655, |
| "loss_idm": 1.390625, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.8310215711593628, |
| "loss_idm_type": 0.6153789043426514, |
| "loss_sft": 0.013951633498072625, |
| "step": 90, |
| "token_acc": 0.8492400086679256 |
| }, |
| { |
| "epoch": 0.023063063063063063, |
| "grad_norm": 251.73202514648438, |
| "learning_rate": 1e-05, |
| "loss": 25.476950073242186, |
| "loss_idm": 1.49375, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.9382863879203797, |
| "loss_idm_type": 0.6257401466369629, |
| "loss_sft": 0.014301823824644089, |
| "step": 100, |
| "token_acc": 0.8458359752497978 |
| }, |
| { |
| "epoch": 0.025369369369369368, |
| "grad_norm": 286.3401184082031, |
| "learning_rate": 9.999316524962347e-06, |
| "loss": 24.417092895507814, |
| "loss_idm": 1.428125, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.904452109336853, |
| "loss_idm_type": 0.593038022518158, |
| "loss_sft": 0.014249381050467491, |
| "step": 110, |
| "token_acc": 0.8465379122197904 |
| }, |
| { |
| "epoch": 0.027675675675675675, |
| "grad_norm": 222.7222442626953, |
| "learning_rate": 9.99726628670463e-06, |
| "loss": 26.060919189453124, |
| "loss_idm": 1.5484375, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.973819124698639, |
| "loss_idm_type": 0.6266206264495849, |
| "loss_sft": 0.014172710105776786, |
| "step": 120, |
| "token_acc": 0.847854247216184 |
| }, |
| { |
| "epoch": 0.029981981981981983, |
| "grad_norm": 275.38079833984375, |
| "learning_rate": 9.993849845741525e-06, |
| "loss": 25.397323608398438, |
| "loss_idm": 1.484375, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.9372030019760131, |
| "loss_idm_type": 0.6214419364929199, |
| "loss_sft": 0.014417377859354019, |
| "step": 130, |
| "token_acc": 0.8455794621991264 |
| }, |
| { |
| "epoch": 0.03228828828828829, |
| "grad_norm": 193.24574279785156, |
| "learning_rate": 9.989068136093873e-06, |
| "loss": 23.078701782226563, |
| "loss_idm": 1.3359375, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.7943235158920288, |
| "loss_idm_type": 0.6200331211090088, |
| "loss_sft": 0.014127673953771592, |
| "step": 140, |
| "token_acc": 0.8480571648523011 |
| }, |
| { |
| "epoch": 0.034594594594594595, |
| "grad_norm": 252.88157653808594, |
| "learning_rate": 9.98292246503335e-06, |
| "loss": 25.396533203125, |
| "loss_idm": 1.50625, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.9365179538726807, |
| "loss_idm_type": 0.6222358584403992, |
| "loss_sft": 0.014179562777280807, |
| "step": 150, |
| "token_acc": 0.846938125590989 |
| }, |
| { |
| "epoch": 0.0369009009009009, |
| "grad_norm": 132.69137573242188, |
| "learning_rate": 9.975414512725058e-06, |
| "loss": 21.090823364257812, |
| "loss_idm": 1.2234375, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.7004317998886108, |
| "loss_idm_type": 0.5901555061340332, |
| "loss_sft": 0.013755682110786437, |
| "step": 160, |
| "token_acc": 0.8513433717153823 |
| }, |
| { |
| "epoch": 0.03920720720720721, |
| "grad_norm": 180.37062072753906, |
| "learning_rate": 9.966546331768192e-06, |
| "loss": 21.250894165039064, |
| "loss_idm": 1.25, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6869279980659485, |
| "loss_idm_type": 0.6129321694374085, |
| "loss_sft": 0.014121741428971291, |
| "step": 170, |
| "token_acc": 0.8468387733683489 |
| }, |
| { |
| "epoch": 0.04151351351351351, |
| "grad_norm": 185.7732696533203, |
| "learning_rate": 9.956320346634877e-06, |
| "loss": 22.12333984375, |
| "loss_idm": 1.2890625, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.7296592950820923, |
| "loss_idm_type": 0.6256211042404175, |
| "loss_sft": 0.013716945424675941, |
| "step": 180, |
| "token_acc": 0.8514388900899068 |
| }, |
| { |
| "epoch": 0.04381981981981982, |
| "grad_norm": 192.45997619628906, |
| "learning_rate": 9.944739353007344e-06, |
| "loss": 20.918809509277345, |
| "loss_idm": 1.2453125, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6708066940307618, |
| "loss_idm_type": 0.6089411020278931, |
| "loss_sft": 0.01396913044154644, |
| "step": 190, |
| "token_acc": 0.8494491495904698 |
| }, |
| { |
| "epoch": 0.04612612612612613, |
| "grad_norm": 138.6709442138672, |
| "learning_rate": 9.931806517013612e-06, |
| "loss": 21.51440887451172, |
| "loss_idm": 1.2484375, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6464977025985718, |
| "loss_idm_type": 0.6701341271400452, |
| "loss_sft": 0.01424136757850647, |
| "step": 200, |
| "token_acc": 0.8475712431229523 |
| }, |
| { |
| "epoch": 0.048432432432432435, |
| "grad_norm": 156.5552520751953, |
| "learning_rate": 9.917525374361913e-06, |
| "loss": 21.93793182373047, |
| "loss_idm": 1.2875, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.7565288722515107, |
| "loss_idm_type": 0.5858924746513366, |
| "loss_sft": 0.014323525503277779, |
| "step": 210, |
| "token_acc": 0.8463558121278266 |
| }, |
| { |
| "epoch": 0.050738738738738735, |
| "grad_norm": 189.6932830810547, |
| "learning_rate": 9.901899829374048e-06, |
| "loss": 20.054583740234374, |
| "loss_idm": 1.159375, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6576752543449402, |
| "loss_idm_type": 0.567785918712616, |
| "loss_sft": 0.014022772386670112, |
| "step": 220, |
| "token_acc": 0.8487998119313549 |
| }, |
| { |
| "epoch": 0.05304504504504504, |
| "grad_norm": 140.5413360595703, |
| "learning_rate": 9.884934153917998e-06, |
| "loss": 22.073431396484374, |
| "loss_idm": 1.3, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.7526954412460327, |
| "loss_idm_type": 0.5985090017318726, |
| "loss_sft": 0.014082278311252593, |
| "step": 230, |
| "token_acc": 0.8485526712222985 |
| }, |
| { |
| "epoch": 0.05535135135135135, |
| "grad_norm": 133.97390747070312, |
| "learning_rate": 9.86663298624003e-06, |
| "loss": 21.144790649414062, |
| "loss_idm": 1.2375, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.7176072120666503, |
| "loss_idm_type": 0.5748069643974304, |
| "loss_sft": 0.014453838020563126, |
| "step": 240, |
| "token_acc": 0.8452394567722186 |
| }, |
| { |
| "epoch": 0.05765765765765766, |
| "grad_norm": 142.247314453125, |
| "learning_rate": 9.847001329696653e-06, |
| "loss": 20.1635498046875, |
| "loss_idm": 1.178125, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6437483191490173, |
| "loss_idm_type": 0.5870923161506653, |
| "loss_sft": 0.014495516940951348, |
| "step": 250, |
| "token_acc": 0.8446375732200728 |
| }, |
| { |
| "epoch": 0.059963963963963966, |
| "grad_norm": 202.28245544433594, |
| "learning_rate": 9.826044551386743e-06, |
| "loss": 21.010342407226563, |
| "loss_idm": 1.2265625, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6738935589790345, |
| "loss_idm_type": 0.6098474025726318, |
| "loss_sft": 0.014558565244078637, |
| "step": 260, |
| "token_acc": 0.8439074729940742 |
| }, |
| { |
| "epoch": 0.06227027027027027, |
| "grad_norm": 180.11004638671875, |
| "learning_rate": 9.803768380684242e-06, |
| "loss": 20.15380096435547, |
| "loss_idm": 1.178125, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6457672238349914, |
| "loss_idm_type": 0.5851505637168884, |
| "loss_sft": 0.01434638984501362, |
| "step": 270, |
| "token_acc": 0.8461026195507937 |
| }, |
| { |
| "epoch": 0.06457657657657657, |
| "grad_norm": 108.87934875488281, |
| "learning_rate": 9.780178907671788e-06, |
| "loss": 20.92729797363281, |
| "loss_idm": 1.2203125, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6868312954902649, |
| "loss_idm_type": 0.5920520305633545, |
| "loss_sft": 0.014423593878746033, |
| "step": 280, |
| "token_acc": 0.8456991679616043 |
| }, |
| { |
| "epoch": 0.06688288288288288, |
| "grad_norm": 149.56564331054688, |
| "learning_rate": 9.755282581475769e-06, |
| "loss": 18.62049102783203, |
| "loss_idm": 1.09375, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.58313547372818, |
| "loss_idm_type": 0.5516607403755188, |
| "loss_sft": 0.014439313858747482, |
| "step": 290, |
| "token_acc": 0.8454596672959642 |
| }, |
| { |
| "epoch": 0.06918918918918919, |
| "grad_norm": 153.041015625, |
| "learning_rate": 9.729086208503174e-06, |
| "loss": 20.677386474609374, |
| "loss_idm": 1.215625, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.7001219868659974, |
| "loss_idm_type": 0.5628572702407837, |
| "loss_sft": 0.014919180795550347, |
| "step": 300, |
| "token_acc": 0.84088942131384 |
| }, |
| { |
| "epoch": 0.0714954954954955, |
| "grad_norm": 177.866943359375, |
| "learning_rate": 9.701596950580807e-06, |
| "loss": 19.88404235839844, |
| "loss_idm": 1.1453125, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6421271324157715, |
| "loss_idm_type": 0.5717909574508667, |
| "loss_sft": 0.014424886554479599, |
| "step": 310, |
| "token_acc": 0.846033168982519 |
| }, |
| { |
| "epoch": 0.0738018018018018, |
| "grad_norm": 129.16363525390625, |
| "learning_rate": 9.672822322997305e-06, |
| "loss": 17.68323974609375, |
| "loss_idm": 1.0296875, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.5472238302230835, |
| "loss_idm_type": 0.5287622213363647, |
| "loss_sft": 0.014649244770407677, |
| "step": 320, |
| "token_acc": 0.8431312749319898 |
| }, |
| { |
| "epoch": 0.07610810810810811, |
| "grad_norm": 156.39739990234375, |
| "learning_rate": 9.642770192448537e-06, |
| "loss": 21.36103515625, |
| "loss_idm": 1.253125, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.7405102610588074, |
| "loss_idm_type": 0.5659091114997864, |
| "loss_sft": 0.014557573571801185, |
| "step": 330, |
| "token_acc": 0.8441445527139338 |
| }, |
| { |
| "epoch": 0.07841441441441442, |
| "grad_norm": 129.96697998046875, |
| "learning_rate": 9.611448774886925e-06, |
| "loss": 21.416160583496094, |
| "loss_idm": 1.234375, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.7018738508224487, |
| "loss_idm_type": 0.607703685760498, |
| "loss_sft": 0.014446442201733588, |
| "step": 340, |
| "token_acc": 0.8443760364686431 |
| }, |
| { |
| "epoch": 0.08072072072072072, |
| "grad_norm": 180.76304626464844, |
| "learning_rate": 9.578866633275289e-06, |
| "loss": 20.02213134765625, |
| "loss_idm": 1.1671875, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6722601294517517, |
| "loss_idm_type": 0.5498591780662536, |
| "loss_sft": 0.014624126255512238, |
| "step": 350, |
| "token_acc": 0.8432171463399417 |
| }, |
| { |
| "epoch": 0.08302702702702702, |
| "grad_norm": 118.01971435546875, |
| "learning_rate": 9.545032675245814e-06, |
| "loss": 18.858502197265626, |
| "loss_idm": 1.109375, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6102901577949524, |
| "loss_idm_type": 0.5399273037910461, |
| "loss_sft": 0.01428530141711235, |
| "step": 360, |
| "token_acc": 0.846329725225022 |
| }, |
| { |
| "epoch": 0.08533333333333333, |
| "grad_norm": 125.42524719238281, |
| "learning_rate": 9.509956150664796e-06, |
| "loss": 18.46220703125, |
| "loss_idm": 1.0765625, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6047607123851776, |
| "loss_idm_type": 0.5198914885520936, |
| "loss_sft": 0.014666610956192016, |
| "step": 370, |
| "token_acc": 0.8430305689629954 |
| }, |
| { |
| "epoch": 0.08763963963963964, |
| "grad_norm": 180.96270751953125, |
| "learning_rate": 9.473646649103819e-06, |
| "loss": 20.59235534667969, |
| "loss_idm": 1.1921875, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6622613787651062, |
| "loss_idm_type": 0.59524484872818, |
| "loss_sft": 0.014829005673527717, |
| "step": 380, |
| "token_acc": 0.8417713702410807 |
| }, |
| { |
| "epoch": 0.08994594594594595, |
| "grad_norm": 144.25331115722656, |
| "learning_rate": 9.43611409721806e-06, |
| "loss": 18.71649627685547, |
| "loss_idm": 1.0890625, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6037861943244934, |
| "loss_idm_type": 0.5369527459144592, |
| "loss_sft": 0.014459025114774704, |
| "step": 390, |
| "token_acc": 0.8447289389919099 |
| }, |
| { |
| "epoch": 0.09225225225225225, |
| "grad_norm": 184.02232360839844, |
| "learning_rate": 9.397368756032445e-06, |
| "loss": 18.94854278564453, |
| "loss_idm": 1.1, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6332925677299499, |
| "loss_idm_type": 0.5221684217453003, |
| "loss_sft": 0.014429151266813278, |
| "step": 400, |
| "token_acc": 0.8452291937983728 |
| }, |
| { |
| "epoch": 0.09455855855855856, |
| "grad_norm": 173.45257568359375, |
| "learning_rate": 9.357421218136387e-06, |
| "loss": 18.99466552734375, |
| "loss_idm": 1.096875, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6210426688194275, |
| "loss_idm_type": 0.5382487654685975, |
| "loss_sft": 0.014015758782625199, |
| "step": 410, |
| "token_acc": 0.8495117177798309 |
| }, |
| { |
| "epoch": 0.09686486486486487, |
| "grad_norm": 165.1482391357422, |
| "learning_rate": 9.31628240478787e-06, |
| "loss": 19.396931457519532, |
| "loss_idm": 1.1328125, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6425803899765015, |
| "loss_idm_type": 0.5403215169906617, |
| "loss_sft": 0.01475544199347496, |
| "step": 420, |
| "token_acc": 0.8424114958179343 |
| }, |
| { |
| "epoch": 0.09917117117117118, |
| "grad_norm": 179.56707763671875, |
| "learning_rate": 9.273963562927695e-06, |
| "loss": 19.1131591796875, |
| "loss_idm": 1.121875, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.5948790431022644, |
| "loss_idm_type": 0.570659625530243, |
| "loss_sft": 0.014644772931933402, |
| "step": 430, |
| "token_acc": 0.8437164287906691 |
| }, |
| { |
| "epoch": 0.10147747747747747, |
| "grad_norm": 141.6302032470703, |
| "learning_rate": 9.230476262104678e-06, |
| "loss": 20.165437316894533, |
| "loss_idm": 1.175, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6892072916030884, |
| "loss_idm_type": 0.5419347524642945, |
| "loss_sft": 0.014781763404607772, |
| "step": 440, |
| "token_acc": 0.8412593497029246 |
| }, |
| { |
| "epoch": 0.10378378378378378, |
| "grad_norm": 84.81742095947266, |
| "learning_rate": 9.185832391312644e-06, |
| "loss": 19.339076232910156, |
| "loss_idm": 1.13125, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.7115584373474121, |
| "loss_idm_type": 0.4681084334850311, |
| "loss_sft": 0.014591620862483978, |
| "step": 450, |
| "token_acc": 0.8433540866857198 |
| }, |
| { |
| "epoch": 0.10609009009009009, |
| "grad_norm": 164.00550842285156, |
| "learning_rate": 9.140044155740102e-06, |
| "loss": 18.864373779296876, |
| "loss_idm": 1.09375, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6536977291107178, |
| "loss_idm_type": 0.4957777142524719, |
| "loss_sft": 0.014747689291834832, |
| "step": 460, |
| "token_acc": 0.841443148203543 |
| }, |
| { |
| "epoch": 0.1083963963963964, |
| "grad_norm": 188.7095184326172, |
| "learning_rate": 9.093124073433464e-06, |
| "loss": 20.679414367675783, |
| "loss_idm": 1.2171875, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.7044301986694336, |
| "loss_idm_type": 0.5580654740333557, |
| "loss_sft": 0.014930276200175285, |
| "step": 470, |
| "token_acc": 0.8409072937658298 |
| }, |
| { |
| "epoch": 0.1107027027027027, |
| "grad_norm": 124.54925537109375, |
| "learning_rate": 9.045084971874738e-06, |
| "loss": 18.53998260498047, |
| "loss_idm": 1.0703125, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6069076418876648, |
| "loss_idm_type": 0.521586561203003, |
| "loss_sft": 0.015061390027403832, |
| "step": 480, |
| "token_acc": 0.8401639957878136 |
| }, |
| { |
| "epoch": 0.11300900900900901, |
| "grad_norm": 117.37905883789062, |
| "learning_rate": 8.995939984474624e-06, |
| "loss": 19.989923095703126, |
| "loss_idm": 1.1703125, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6740325570106507, |
| "loss_idm_type": 0.5453832864761352, |
| "loss_sft": 0.014970069751143456, |
| "step": 490, |
| "token_acc": 0.8400487229410242 |
| }, |
| { |
| "epoch": 0.11531531531531532, |
| "grad_norm": 132.42762756347656, |
| "learning_rate": 8.94570254698197e-06, |
| "loss": 19.590615844726564, |
| "loss_idm": 1.15625, |
| "loss_idm_anl": 0.0, |
| "loss_idm_coord": 0.0, |
| "loss_idm_text": 0.6635529398918152, |
| "loss_idm_type": 0.5308433651924134, |
| "loss_sft": 0.014947501569986343, |
| "step": 500, |
| "token_acc": 0.84046714392237 |
| }, |
| { |
| "epoch": 0.11531531531531532, |
| "eval_loss": 1.3233741521835327, |
| "eval_runtime": 170.0148, |
| "eval_samples_per_second": 10.264, |
| "eval_steps_per_second": 2.57, |
| "eval_token_acc": 0.828006241506015, |
| "step": 500 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 2000, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 1, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 6.251724037081268e+18, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|