{ "best_global_step": 908, "best_metric": 0.5586269497871399, "best_model_checkpoint": "/outputs/work/smollm3-3b_structured/checkpoints/checkpoint-908", "epoch": 2.0, "eval_steps": 100, "global_step": 908, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.02203856749311295, "grad_norm": 0.9534511566162109, "learning_rate": 3.3333333333333335e-05, "loss": 1.2359769821166993, "step": 10 }, { "epoch": 0.0440771349862259, "grad_norm": 0.6362833976745605, "learning_rate": 7.037037037037038e-05, "loss": 1.0208109855651855, "step": 20 }, { "epoch": 0.06611570247933884, "grad_norm": 0.6407135128974915, "learning_rate": 9.999872841297521e-05, "loss": 0.8830883979797364, "step": 30 }, { "epoch": 0.0881542699724518, "grad_norm": 0.8838831186294556, "learning_rate": 9.995422965786025e-05, "loss": 0.8082539558410644, "step": 40 }, { "epoch": 0.11019283746556474, "grad_norm": 0.5342389345169067, "learning_rate": 9.984621621406273e-05, "loss": 0.7579926490783692, "step": 50 }, { "epoch": 0.1322314049586777, "grad_norm": 0.8452595472335815, "learning_rate": 9.96748254161048e-05, "loss": 0.769981575012207, "step": 60 }, { "epoch": 0.15426997245179064, "grad_norm": 0.5976874232292175, "learning_rate": 9.944027518013187e-05, "loss": 0.6899113655090332, "step": 70 }, { "epoch": 0.1763085399449036, "grad_norm": 0.6876652836799622, "learning_rate": 9.914286372684153e-05, "loss": 0.6597393989562989, "step": 80 }, { "epoch": 0.19834710743801653, "grad_norm": 0.6256213784217834, "learning_rate": 9.878296920230869e-05, "loss": 0.6918736934661865, "step": 90 }, { "epoch": 0.22038567493112948, "grad_norm": 0.6878316402435303, "learning_rate": 9.83610491971886e-05, "loss": 0.7168015956878662, "step": 100 }, { "epoch": 0.22038567493112948, "eval_loss": 0.6784517168998718, "eval_runtime": 11.9227, "eval_samples_per_second": 16.104, "eval_steps_per_second": 16.104, "step": 100 }, { "epoch": 0.24242424242424243, "grad_norm": 0.6689047813415527, "learning_rate": 9.787764016490992e-05, "loss": 0.6953184127807617, "step": 110 }, { "epoch": 0.2644628099173554, "grad_norm": 0.47161027789115906, "learning_rate": 9.733335673959671e-05, "loss": 0.6750634193420411, "step": 120 }, { "epoch": 0.2865013774104683, "grad_norm": 0.7510103583335876, "learning_rate": 9.672889095458734e-05, "loss": 0.6649476528167725, "step": 130 }, { "epoch": 0.3085399449035813, "grad_norm": 0.6093885898590088, "learning_rate": 9.606501136254327e-05, "loss": 0.6267831802368165, "step": 140 }, { "epoch": 0.3305785123966942, "grad_norm": 0.6549955606460571, "learning_rate": 9.534256205826684e-05, "loss": 0.6659124374389649, "step": 150 }, { "epoch": 0.3526170798898072, "grad_norm": 0.5473728775978088, "learning_rate": 9.456246160547057e-05, "loss": 0.6131270408630372, "step": 160 }, { "epoch": 0.3746556473829201, "grad_norm": 0.6127820014953613, "learning_rate": 9.372570186886225e-05, "loss": 0.609403657913208, "step": 170 }, { "epoch": 0.39669421487603307, "grad_norm": 0.6445918083190918, "learning_rate": 9.283334675303084e-05, "loss": 0.6382901668548584, "step": 180 }, { "epoch": 0.418732782369146, "grad_norm": 0.7743238210678101, "learning_rate": 9.188653084973692e-05, "loss": 0.6500725746154785, "step": 190 }, { "epoch": 0.44077134986225897, "grad_norm": 0.7458897829055786, "learning_rate": 9.088645799532729e-05, "loss": 0.6170906543731689, "step": 200 }, { "epoch": 0.44077134986225897, "eval_loss": 0.6293469071388245, "eval_runtime": 11.9203, "eval_samples_per_second": 16.107, "eval_steps_per_second": 16.107, "step": 200 }, { "epoch": 0.4628099173553719, "grad_norm": 0.5193160772323608, "learning_rate": 8.983439974010794e-05, "loss": 0.6225500583648682, "step": 210 }, { "epoch": 0.48484848484848486, "grad_norm": 0.5633059740066528, "learning_rate": 8.873169373162174e-05, "loss": 0.6362756729125977, "step": 220 }, { "epoch": 0.5068870523415978, "grad_norm": 0.5231749415397644, "learning_rate": 8.757974201388605e-05, "loss": 0.6010201454162598, "step": 230 }, { "epoch": 0.5289256198347108, "grad_norm": 0.6744521260261536, "learning_rate": 8.638000924475336e-05, "loss": 0.6185548305511475, "step": 240 }, { "epoch": 0.5509641873278237, "grad_norm": 0.6033552885055542, "learning_rate": 8.513402083366054e-05, "loss": 0.5994134426116944, "step": 250 }, { "epoch": 0.5730027548209367, "grad_norm": 0.5614070296287537, "learning_rate": 8.384336100213546e-05, "loss": 0.5692955493927002, "step": 260 }, { "epoch": 0.5950413223140496, "grad_norm": 0.6417648196220398, "learning_rate": 8.250967076952635e-05, "loss": 0.6048506736755371, "step": 270 }, { "epoch": 0.6170798898071626, "grad_norm": 0.6965922713279724, "learning_rate": 8.113464586651516e-05, "loss": 0.6014030456542969, "step": 280 }, { "epoch": 0.6391184573002755, "grad_norm": 0.6192167401313782, "learning_rate": 7.972003457906773e-05, "loss": 0.634767723083496, "step": 290 }, { "epoch": 0.6611570247933884, "grad_norm": 0.720318615436554, "learning_rate": 7.826763552556222e-05, "loss": 0.5867681503295898, "step": 300 }, { "epoch": 0.6611570247933884, "eval_loss": 0.6011102795600891, "eval_runtime": 12.0266, "eval_samples_per_second": 15.965, "eval_steps_per_second": 15.965, "step": 300 }, { "epoch": 0.6831955922865014, "grad_norm": 0.7235103249549866, "learning_rate": 7.677929536992194e-05, "loss": 0.5783907413482666, "step": 310 }, { "epoch": 0.7052341597796143, "grad_norm": 0.7653256058692932, "learning_rate": 7.525690647366032e-05, "loss": 0.6062646865844726, "step": 320 }, { "epoch": 0.7272727272727273, "grad_norm": 0.5779842734336853, "learning_rate": 7.370240448982344e-05, "loss": 0.6198268413543702, "step": 330 }, { "epoch": 0.7493112947658402, "grad_norm": 0.709947943687439, "learning_rate": 7.211776590188898e-05, "loss": 0.6091408729553223, "step": 340 }, { "epoch": 0.7713498622589532, "grad_norm": 0.595930278301239, "learning_rate": 7.050500551075121e-05, "loss": 0.6241033554077149, "step": 350 }, { "epoch": 0.7933884297520661, "grad_norm": 0.5700712203979492, "learning_rate": 6.886617387298689e-05, "loss": 0.6038753509521484, "step": 360 }, { "epoch": 0.8154269972451791, "grad_norm": 0.6845744252204895, "learning_rate": 6.720335469365943e-05, "loss": 0.5801003932952881, "step": 370 }, { "epoch": 0.837465564738292, "grad_norm": 0.5995373725891113, "learning_rate": 6.551866217697602e-05, "loss": 0.5841936588287353, "step": 380 }, { "epoch": 0.859504132231405, "grad_norm": 0.702873170375824, "learning_rate": 6.381423833816653e-05, "loss": 0.5742908477783203, "step": 390 }, { "epoch": 0.8815426997245179, "grad_norm": 0.5957848429679871, "learning_rate": 6.209225028000173e-05, "loss": 0.6035511493682861, "step": 400 }, { "epoch": 0.8815426997245179, "eval_loss": 0.5868632197380066, "eval_runtime": 11.9519, "eval_samples_per_second": 16.064, "eval_steps_per_second": 16.064, "step": 400 }, { "epoch": 0.9035812672176309, "grad_norm": 0.6939573884010315, "learning_rate": 6.0354887437413965e-05, "loss": 0.5780395030975342, "step": 410 }, { "epoch": 0.9256198347107438, "grad_norm": 0.6161796450614929, "learning_rate": 5.860435879372319e-05, "loss": 0.5949501037597656, "step": 420 }, { "epoch": 0.9476584022038568, "grad_norm": 0.6215364336967468, "learning_rate": 5.684289007200819e-05, "loss": 0.5814109802246094, "step": 430 }, { "epoch": 0.9696969696969697, "grad_norm": 0.523289680480957, "learning_rate": 5.5072720905193984e-05, "loss": 0.5734921932220459, "step": 440 }, { "epoch": 0.9917355371900827, "grad_norm": 0.6269870400428772, "learning_rate": 5.329610198845322e-05, "loss": 0.5888432502746582, "step": 450 }, { "epoch": 1.0132231404958678, "grad_norm": 0.6019392013549805, "learning_rate": 5.151529221754262e-05, "loss": 0.5435478687286377, "step": 460 }, { "epoch": 1.0352617079889808, "grad_norm": 0.6337491869926453, "learning_rate": 4.97325558167126e-05, "loss": 0.5148682594299316, "step": 470 }, { "epoch": 1.0573002754820937, "grad_norm": 0.6463139653205872, "learning_rate": 4.795015945984198e-05, "loss": 0.4946620941162109, "step": 480 }, { "epoch": 1.0793388429752067, "grad_norm": 0.667147159576416, "learning_rate": 4.617036938845806e-05, "loss": 0.5063100337982178, "step": 490 }, { "epoch": 1.1013774104683196, "grad_norm": 0.5854341983795166, "learning_rate": 4.43954485303065e-05, "loss": 0.5141865730285644, "step": 500 }, { "epoch": 1.1013774104683196, "eval_loss": 0.5751394629478455, "eval_runtime": 11.933, "eval_samples_per_second": 16.09, "eval_steps_per_second": 16.09, "step": 500 }, { "epoch": 1.1234159779614326, "grad_norm": 0.7505147457122803, "learning_rate": 4.2627653622134346e-05, "loss": 0.5038840770721436, "step": 510 }, { "epoch": 1.1454545454545455, "grad_norm": 0.6712458729743958, "learning_rate": 4.086923234034493e-05, "loss": 0.49640936851501466, "step": 520 }, { "epoch": 1.1674931129476585, "grad_norm": 0.6768338084220886, "learning_rate": 3.912242044317233e-05, "loss": 0.5040760993957519, "step": 530 }, { "epoch": 1.1895316804407714, "grad_norm": 0.6140052080154419, "learning_rate": 3.738943892800951e-05, "loss": 0.4928180694580078, "step": 540 }, { "epoch": 1.2115702479338843, "grad_norm": 0.7098944187164307, "learning_rate": 3.5672491207504e-05, "loss": 0.49433393478393556, "step": 550 }, { "epoch": 1.2336088154269973, "grad_norm": 0.603570282459259, "learning_rate": 3.397376030801215e-05, "loss": 0.5312057495117187, "step": 560 }, { "epoch": 1.2556473829201102, "grad_norm": 0.8194121718406677, "learning_rate": 3.229540609397334e-05, "loss": 0.5198305130004883, "step": 570 }, { "epoch": 1.2776859504132232, "grad_norm": 0.7442294955253601, "learning_rate": 3.0639562521733935e-05, "loss": 0.4971327304840088, "step": 580 }, { "epoch": 1.2997245179063361, "grad_norm": 0.664591372013092, "learning_rate": 2.900833492631191e-05, "loss": 0.4868955612182617, "step": 590 }, { "epoch": 1.321763085399449, "grad_norm": 0.7322269678115845, "learning_rate": 2.7403797344552696e-05, "loss": 0.5075324058532715, "step": 600 }, { "epoch": 1.321763085399449, "eval_loss": 0.5670042634010315, "eval_runtime": 11.8846, "eval_samples_per_second": 16.155, "eval_steps_per_second": 16.155, "step": 600 }, { "epoch": 1.343801652892562, "grad_norm": 0.7925372123718262, "learning_rate": 2.5827989878078972e-05, "loss": 0.5191412925720215, "step": 610 }, { "epoch": 1.365840220385675, "grad_norm": 0.7049777507781982, "learning_rate": 2.428291609938786e-05, "loss": 0.4850438117980957, "step": 620 }, { "epoch": 1.387878787878788, "grad_norm": 0.5821607112884521, "learning_rate": 2.2770540504393224e-05, "loss": 0.4845143795013428, "step": 630 }, { "epoch": 1.4099173553719009, "grad_norm": 0.8042203187942505, "learning_rate": 2.1292786014652398e-05, "loss": 0.5041770935058594, "step": 640 }, { "epoch": 1.4319559228650138, "grad_norm": 0.6269323229789734, "learning_rate": 1.9851531532452665e-05, "loss": 0.4969661712646484, "step": 650 }, { "epoch": 1.4539944903581268, "grad_norm": 0.6438784599304199, "learning_rate": 1.8448609551866647e-05, "loss": 0.49288215637207033, "step": 660 }, { "epoch": 1.4760330578512397, "grad_norm": 0.6186577081680298, "learning_rate": 1.7085803828813546e-05, "loss": 0.5123368263244629, "step": 670 }, { "epoch": 1.4980716253443527, "grad_norm": 0.8740533590316772, "learning_rate": 1.5764847113089094e-05, "loss": 0.47509055137634276, "step": 680 }, { "epoch": 1.5201101928374654, "grad_norm": 0.7086982727050781, "learning_rate": 1.4487418945247438e-05, "loss": 0.49436202049255373, "step": 690 }, { "epoch": 1.5421487603305786, "grad_norm": 0.6316142082214355, "learning_rate": 1.3255143521136498e-05, "loss": 0.5204075336456299, "step": 700 }, { "epoch": 1.5421487603305786, "eval_loss": 0.5617139339447021, "eval_runtime": 12.2107, "eval_samples_per_second": 15.724, "eval_steps_per_second": 15.724, "step": 700 }, { "epoch": 1.5641873278236913, "grad_norm": 0.6398480534553528, "learning_rate": 1.206958762680157e-05, "loss": 0.49640421867370604, "step": 710 }, { "epoch": 1.5862258953168045, "grad_norm": 0.7017738819122314, "learning_rate": 1.0932258646383336e-05, "loss": 0.5167774677276611, "step": 720 }, { "epoch": 1.6082644628099172, "grad_norm": 0.8112418055534363, "learning_rate": 9.844602645542584e-06, "loss": 0.4898106575012207, "step": 730 }, { "epoch": 1.6303030303030304, "grad_norm": 0.6042963862419128, "learning_rate": 8.808002532849047e-06, "loss": 0.5056878089904785, "step": 740 }, { "epoch": 1.652341597796143, "grad_norm": 0.6272949576377869, "learning_rate": 7.823776301471591e-06, "loss": 0.49520001411437986, "step": 750 }, { "epoch": 1.6743801652892563, "grad_norm": 0.6539549231529236, "learning_rate": 6.893175353405756e-06, "loss": 0.4806521892547607, "step": 760 }, { "epoch": 1.696418732782369, "grad_norm": 0.6603189706802368, "learning_rate": 6.017382908369051e-06, "loss": 0.4939537048339844, "step": 770 }, { "epoch": 1.7184573002754822, "grad_norm": 0.9547717571258545, "learning_rate": 5.197512499387175e-06, "loss": 0.5203057765960694, "step": 780 }, { "epoch": 1.7404958677685949, "grad_norm": 0.7032685279846191, "learning_rate": 4.434606556983878e-06, "loss": 0.5253971576690674, "step": 790 }, { "epoch": 1.762534435261708, "grad_norm": 0.6500975489616394, "learning_rate": 3.729635083774724e-06, "loss": 0.5071902275085449, "step": 800 }, { "epoch": 1.762534435261708, "eval_loss": 0.5592079162597656, "eval_runtime": 11.8791, "eval_samples_per_second": 16.163, "eval_steps_per_second": 16.163, "step": 800 }, { "epoch": 1.7845730027548208, "grad_norm": 0.5473243594169617, "learning_rate": 3.083494421149752e-06, "loss": 0.5045226097106934, "step": 810 }, { "epoch": 1.806611570247934, "grad_norm": 0.6320291757583618, "learning_rate": 2.49700610961342e-06, "loss": 0.5111465454101562, "step": 820 }, { "epoch": 1.8286501377410467, "grad_norm": 0.6422545909881592, "learning_rate": 1.9709158442305365e-06, "loss": 0.48850622177124026, "step": 830 }, { "epoch": 1.8506887052341598, "grad_norm": 0.7762777209281921, "learning_rate": 1.5058925265066194e-06, "loss": 0.4583238124847412, "step": 840 }, { "epoch": 1.8727272727272726, "grad_norm": 0.6142582297325134, "learning_rate": 1.1025274139079057e-06, "loss": 0.4907660484313965, "step": 850 }, { "epoch": 1.8947658402203857, "grad_norm": 0.8006341457366943, "learning_rate": 7.613333681025236e-07, "loss": 0.4667381286621094, "step": 860 }, { "epoch": 1.9168044077134985, "grad_norm": 0.8411538004875183, "learning_rate": 4.827442028785655e-07, "loss": 0.5014886379241943, "step": 870 }, { "epoch": 1.9388429752066116, "grad_norm": 0.6041914820671082, "learning_rate": 2.67114132568258e-07, "loss": 0.5158967018127442, "step": 880 }, { "epoch": 1.9608815426997244, "grad_norm": 0.6050441861152649, "learning_rate": 1.1471732167938065e-07, "loss": 0.49015021324157715, "step": 890 }, { "epoch": 1.9829201101928375, "grad_norm": 0.6653909087181091, "learning_rate": 2.57475363067472e-08, "loss": 0.5256523609161377, "step": 900 }, { "epoch": 1.9829201101928375, "eval_loss": 0.5588088035583496, "eval_runtime": 11.8638, "eval_samples_per_second": 16.184, "eval_steps_per_second": 16.184, "step": 900 }, { "epoch": 2.0, "eval_loss": 0.5586269497871399, "eval_runtime": 11.8357, "eval_samples_per_second": 16.222, "eval_steps_per_second": 16.222, "step": 908 } ], "logging_steps": 10, "max_steps": 908, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 100, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 1.1401079002855834e+17, "train_batch_size": 1, "trial_name": null, "trial_params": null }