| { |
| "best_global_step": 5160, |
| "best_metric": 0.39690853763350886, |
| "best_model_checkpoint": "results_indot5_train_only/checkpoint-5160", |
| "epoch": 8.0, |
| "eval_steps": 500, |
| "global_step": 5160, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.07751937984496124, |
| "grad_norm": 5.20282506942749, |
| "learning_rate": 1.012396694214876e-05, |
| "loss": 5.057557067871094, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.15503875968992248, |
| "grad_norm": 2.5345635414123535, |
| "learning_rate": 2.0454545454545457e-05, |
| "loss": 3.9355859375, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.23255813953488372, |
| "grad_norm": 2.3226816654205322, |
| "learning_rate": 3.078512396694215e-05, |
| "loss": 3.3620819091796874, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.31007751937984496, |
| "grad_norm": 1.6183891296386719, |
| "learning_rate": 4.111570247933885e-05, |
| "loss": 3.0566058349609375, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.3875968992248062, |
| "grad_norm": 1.5544594526290894, |
| "learning_rate": 5.144628099173554e-05, |
| "loss": 2.8706439208984373, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.46511627906976744, |
| "grad_norm": 2.173813819885254, |
| "learning_rate": 6.177685950413224e-05, |
| "loss": 2.714820556640625, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.5426356589147286, |
| "grad_norm": 1.6846433877944946, |
| "learning_rate": 7.210743801652892e-05, |
| "loss": 2.588794860839844, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.6201550387596899, |
| "grad_norm": 2.4052934646606445, |
| "learning_rate": 8.243801652892562e-05, |
| "loss": 2.486365966796875, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.6976744186046512, |
| "grad_norm": 1.552986979484558, |
| "learning_rate": 9.276859504132232e-05, |
| "loss": 2.437692413330078, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.7751937984496124, |
| "grad_norm": 1.4405871629714966, |
| "learning_rate": 9.983679686649984e-05, |
| "loss": 2.3887617492675783, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.8527131782945736, |
| "grad_norm": 1.3462281227111816, |
| "learning_rate": 9.929278642149929e-05, |
| "loss": 2.250489501953125, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.9302325581395349, |
| "grad_norm": 1.2261027097702026, |
| "learning_rate": 9.874877597649875e-05, |
| "loss": 2.2036700439453125, |
| "step": 600 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_loss": 2.0033321380615234, |
| "eval_rougeL": 0.28385957598269274, |
| "eval_runtime": 273.0271, |
| "eval_samples_per_second": 1.245, |
| "eval_steps_per_second": 0.157, |
| "step": 645 |
| }, |
| { |
| "epoch": 1.0077519379844961, |
| "grad_norm": 1.3980158567428589, |
| "learning_rate": 9.820476553149822e-05, |
| "loss": 2.1746539306640624, |
| "step": 650 |
| }, |
| { |
| "epoch": 1.0852713178294573, |
| "grad_norm": 1.1627949476242065, |
| "learning_rate": 9.766075508649766e-05, |
| "loss": 2.0846818542480468, |
| "step": 700 |
| }, |
| { |
| "epoch": 1.1627906976744187, |
| "grad_norm": 1.43430757522583, |
| "learning_rate": 9.711674464149713e-05, |
| "loss": 2.0509593200683596, |
| "step": 750 |
| }, |
| { |
| "epoch": 1.2403100775193798, |
| "grad_norm": 1.63274085521698, |
| "learning_rate": 9.657273419649658e-05, |
| "loss": 2.0283493041992187, |
| "step": 800 |
| }, |
| { |
| "epoch": 1.3178294573643412, |
| "grad_norm": 1.2429606914520264, |
| "learning_rate": 9.602872375149603e-05, |
| "loss": 1.9829312133789063, |
| "step": 850 |
| }, |
| { |
| "epoch": 1.3953488372093024, |
| "grad_norm": 1.3741137981414795, |
| "learning_rate": 9.548471330649549e-05, |
| "loss": 1.979344940185547, |
| "step": 900 |
| }, |
| { |
| "epoch": 1.4728682170542635, |
| "grad_norm": 7.487593173980713, |
| "learning_rate": 9.494070286149494e-05, |
| "loss": 1.94923583984375, |
| "step": 950 |
| }, |
| { |
| "epoch": 1.550387596899225, |
| "grad_norm": 1.2685985565185547, |
| "learning_rate": 9.43966924164944e-05, |
| "loss": 1.9149658203125, |
| "step": 1000 |
| }, |
| { |
| "epoch": 1.627906976744186, |
| "grad_norm": 1.346503734588623, |
| "learning_rate": 9.385268197149387e-05, |
| "loss": 1.9400949096679687, |
| "step": 1050 |
| }, |
| { |
| "epoch": 1.7054263565891472, |
| "grad_norm": 1.4422067403793335, |
| "learning_rate": 9.330867152649332e-05, |
| "loss": 1.8536480712890624, |
| "step": 1100 |
| }, |
| { |
| "epoch": 1.7829457364341086, |
| "grad_norm": 1.1799274682998657, |
| "learning_rate": 9.276466108149277e-05, |
| "loss": 1.8910017395019532, |
| "step": 1150 |
| }, |
| { |
| "epoch": 1.8604651162790697, |
| "grad_norm": 1.329649806022644, |
| "learning_rate": 9.222065063649223e-05, |
| "loss": 1.8872673034667968, |
| "step": 1200 |
| }, |
| { |
| "epoch": 1.937984496124031, |
| "grad_norm": 1.4333596229553223, |
| "learning_rate": 9.167664019149168e-05, |
| "loss": 1.8687342834472656, |
| "step": 1250 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_loss": 1.7659074068069458, |
| "eval_rougeL": 0.3637597338080381, |
| "eval_runtime": 272.0139, |
| "eval_samples_per_second": 1.25, |
| "eval_steps_per_second": 0.158, |
| "step": 1290 |
| }, |
| { |
| "epoch": 2.0155038759689923, |
| "grad_norm": 1.279874324798584, |
| "learning_rate": 9.113262974649114e-05, |
| "loss": 1.8237803649902344, |
| "step": 1300 |
| }, |
| { |
| "epoch": 2.0930232558139537, |
| "grad_norm": 1.6495519876480103, |
| "learning_rate": 9.058861930149059e-05, |
| "loss": 1.764874267578125, |
| "step": 1350 |
| }, |
| { |
| "epoch": 2.1705426356589146, |
| "grad_norm": 1.3005207777023315, |
| "learning_rate": 9.004460885649005e-05, |
| "loss": 1.7964950561523438, |
| "step": 1400 |
| }, |
| { |
| "epoch": 2.248062015503876, |
| "grad_norm": 1.2198655605316162, |
| "learning_rate": 8.95005984114895e-05, |
| "loss": 1.7576876831054689, |
| "step": 1450 |
| }, |
| { |
| "epoch": 2.3255813953488373, |
| "grad_norm": 1.3467313051223755, |
| "learning_rate": 8.895658796648895e-05, |
| "loss": 1.7860525512695313, |
| "step": 1500 |
| }, |
| { |
| "epoch": 2.4031007751937983, |
| "grad_norm": 1.2706187963485718, |
| "learning_rate": 8.841257752148842e-05, |
| "loss": 1.7707453918457032, |
| "step": 1550 |
| }, |
| { |
| "epoch": 2.4806201550387597, |
| "grad_norm": 1.3819843530654907, |
| "learning_rate": 8.786856707648788e-05, |
| "loss": 1.7504708862304688, |
| "step": 1600 |
| }, |
| { |
| "epoch": 2.558139534883721, |
| "grad_norm": 2.460967779159546, |
| "learning_rate": 8.732455663148733e-05, |
| "loss": 1.7521382141113282, |
| "step": 1650 |
| }, |
| { |
| "epoch": 2.6356589147286824, |
| "grad_norm": 1.6921333074569702, |
| "learning_rate": 8.678054618648679e-05, |
| "loss": 1.7393124389648438, |
| "step": 1700 |
| }, |
| { |
| "epoch": 2.7131782945736433, |
| "grad_norm": 1.4000757932662964, |
| "learning_rate": 8.623653574148624e-05, |
| "loss": 1.7678129577636719, |
| "step": 1750 |
| }, |
| { |
| "epoch": 2.7906976744186047, |
| "grad_norm": 1.225231409072876, |
| "learning_rate": 8.569252529648569e-05, |
| "loss": 1.74262939453125, |
| "step": 1800 |
| }, |
| { |
| "epoch": 2.8682170542635657, |
| "grad_norm": 1.2730450630187988, |
| "learning_rate": 8.514851485148515e-05, |
| "loss": 1.7095193481445312, |
| "step": 1850 |
| }, |
| { |
| "epoch": 2.945736434108527, |
| "grad_norm": 2.069585084915161, |
| "learning_rate": 8.46045044064846e-05, |
| "loss": 1.7451597595214843, |
| "step": 1900 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_loss": 1.6796072721481323, |
| "eval_rougeL": 0.372706302733224, |
| "eval_runtime": 261.7219, |
| "eval_samples_per_second": 1.299, |
| "eval_steps_per_second": 0.164, |
| "step": 1935 |
| }, |
| { |
| "epoch": 3.0232558139534884, |
| "grad_norm": 1.244084358215332, |
| "learning_rate": 8.406049396148407e-05, |
| "loss": 1.6956103515625, |
| "step": 1950 |
| }, |
| { |
| "epoch": 3.10077519379845, |
| "grad_norm": 1.3515751361846924, |
| "learning_rate": 8.351648351648353e-05, |
| "loss": 1.6509474182128907, |
| "step": 2000 |
| }, |
| { |
| "epoch": 3.1782945736434107, |
| "grad_norm": 1.5347480773925781, |
| "learning_rate": 8.297247307148298e-05, |
| "loss": 1.6638705444335937, |
| "step": 2050 |
| }, |
| { |
| "epoch": 3.255813953488372, |
| "grad_norm": 1.2425085306167603, |
| "learning_rate": 8.242846262648243e-05, |
| "loss": 1.646502227783203, |
| "step": 2100 |
| }, |
| { |
| "epoch": 3.3333333333333335, |
| "grad_norm": 1.5136678218841553, |
| "learning_rate": 8.188445218148189e-05, |
| "loss": 1.649458770751953, |
| "step": 2150 |
| }, |
| { |
| "epoch": 3.4108527131782944, |
| "grad_norm": 1.3278005123138428, |
| "learning_rate": 8.134044173648134e-05, |
| "loss": 1.6373924255371093, |
| "step": 2200 |
| }, |
| { |
| "epoch": 3.488372093023256, |
| "grad_norm": 1.1992027759552002, |
| "learning_rate": 8.07964312914808e-05, |
| "loss": 1.6213514709472656, |
| "step": 2250 |
| }, |
| { |
| "epoch": 3.565891472868217, |
| "grad_norm": 1.2497408390045166, |
| "learning_rate": 8.025242084648025e-05, |
| "loss": 1.6644189453125, |
| "step": 2300 |
| }, |
| { |
| "epoch": 3.6434108527131785, |
| "grad_norm": 1.1033735275268555, |
| "learning_rate": 7.970841040147972e-05, |
| "loss": 1.640992431640625, |
| "step": 2350 |
| }, |
| { |
| "epoch": 3.7209302325581395, |
| "grad_norm": 1.567937970161438, |
| "learning_rate": 7.916439995647917e-05, |
| "loss": 1.6204100036621094, |
| "step": 2400 |
| }, |
| { |
| "epoch": 3.798449612403101, |
| "grad_norm": 2.5253713130950928, |
| "learning_rate": 7.862038951147862e-05, |
| "loss": 1.6449601745605469, |
| "step": 2450 |
| }, |
| { |
| "epoch": 3.875968992248062, |
| "grad_norm": 1.2424848079681396, |
| "learning_rate": 7.807637906647808e-05, |
| "loss": 1.6312498474121093, |
| "step": 2500 |
| }, |
| { |
| "epoch": 3.953488372093023, |
| "grad_norm": 1.4053829908370972, |
| "learning_rate": 7.753236862147754e-05, |
| "loss": 1.6399296569824218, |
| "step": 2550 |
| }, |
| { |
| "epoch": 4.0, |
| "eval_loss": 1.6408368349075317, |
| "eval_rougeL": 0.3834982647446496, |
| "eval_runtime": 243.928, |
| "eval_samples_per_second": 1.394, |
| "eval_steps_per_second": 0.176, |
| "step": 2580 |
| }, |
| { |
| "epoch": 4.0310077519379846, |
| "grad_norm": 1.1497409343719482, |
| "learning_rate": 7.698835817647699e-05, |
| "loss": 1.5811054992675782, |
| "step": 2600 |
| }, |
| { |
| "epoch": 4.108527131782946, |
| "grad_norm": 1.2698125839233398, |
| "learning_rate": 7.644434773147645e-05, |
| "loss": 1.5394790649414063, |
| "step": 2650 |
| }, |
| { |
| "epoch": 4.186046511627907, |
| "grad_norm": 1.905889868736267, |
| "learning_rate": 7.59003372864759e-05, |
| "loss": 1.554979248046875, |
| "step": 2700 |
| }, |
| { |
| "epoch": 4.263565891472869, |
| "grad_norm": 1.3710238933563232, |
| "learning_rate": 7.535632684147535e-05, |
| "loss": 1.5750091552734375, |
| "step": 2750 |
| }, |
| { |
| "epoch": 4.341085271317829, |
| "grad_norm": 1.2228940725326538, |
| "learning_rate": 7.481231639647482e-05, |
| "loss": 1.5747805786132814, |
| "step": 2800 |
| }, |
| { |
| "epoch": 4.4186046511627906, |
| "grad_norm": 1.6240923404693604, |
| "learning_rate": 7.426830595147427e-05, |
| "loss": 1.5721925354003907, |
| "step": 2850 |
| }, |
| { |
| "epoch": 4.496124031007752, |
| "grad_norm": 1.2682580947875977, |
| "learning_rate": 7.372429550647373e-05, |
| "loss": 1.5486814880371094, |
| "step": 2900 |
| }, |
| { |
| "epoch": 4.573643410852713, |
| "grad_norm": 1.20686936378479, |
| "learning_rate": 7.318028506147319e-05, |
| "loss": 1.525863800048828, |
| "step": 2950 |
| }, |
| { |
| "epoch": 4.651162790697675, |
| "grad_norm": 1.4539456367492676, |
| "learning_rate": 7.263627461647264e-05, |
| "loss": 1.5479656982421874, |
| "step": 3000 |
| }, |
| { |
| "epoch": 4.728682170542635, |
| "grad_norm": 1.2014193534851074, |
| "learning_rate": 7.209226417147209e-05, |
| "loss": 1.571671142578125, |
| "step": 3050 |
| }, |
| { |
| "epoch": 4.8062015503875966, |
| "grad_norm": 1.336296558380127, |
| "learning_rate": 7.154825372647154e-05, |
| "loss": 1.5505433654785157, |
| "step": 3100 |
| }, |
| { |
| "epoch": 4.883720930232558, |
| "grad_norm": 1.1211562156677246, |
| "learning_rate": 7.1004243281471e-05, |
| "loss": 1.5417903137207032, |
| "step": 3150 |
| }, |
| { |
| "epoch": 4.961240310077519, |
| "grad_norm": 1.6577614545822144, |
| "learning_rate": 7.046023283647047e-05, |
| "loss": 1.5566175842285157, |
| "step": 3200 |
| }, |
| { |
| "epoch": 5.0, |
| "eval_loss": 1.6182421445846558, |
| "eval_rougeL": 0.3815934571992504, |
| "eval_runtime": 255.1085, |
| "eval_samples_per_second": 1.333, |
| "eval_steps_per_second": 0.169, |
| "step": 3225 |
| }, |
| { |
| "epoch": 5.038759689922481, |
| "grad_norm": 2.8794777393341064, |
| "learning_rate": 6.991622239146992e-05, |
| "loss": 1.528376007080078, |
| "step": 3250 |
| }, |
| { |
| "epoch": 5.116279069767442, |
| "grad_norm": 1.3899835348129272, |
| "learning_rate": 6.937221194646938e-05, |
| "loss": 1.486846923828125, |
| "step": 3300 |
| }, |
| { |
| "epoch": 5.1937984496124034, |
| "grad_norm": 1.2603890895843506, |
| "learning_rate": 6.882820150146883e-05, |
| "loss": 1.4783212280273437, |
| "step": 3350 |
| }, |
| { |
| "epoch": 5.271317829457364, |
| "grad_norm": 1.222375512123108, |
| "learning_rate": 6.828419105646828e-05, |
| "loss": 1.4778273010253906, |
| "step": 3400 |
| }, |
| { |
| "epoch": 5.348837209302325, |
| "grad_norm": 1.3104170560836792, |
| "learning_rate": 6.774018061146774e-05, |
| "loss": 1.4953509521484376, |
| "step": 3450 |
| }, |
| { |
| "epoch": 5.426356589147287, |
| "grad_norm": 1.3662341833114624, |
| "learning_rate": 6.719617016646719e-05, |
| "loss": 1.5077650451660156, |
| "step": 3500 |
| }, |
| { |
| "epoch": 5.503875968992248, |
| "grad_norm": 1.380625605583191, |
| "learning_rate": 6.665215972146666e-05, |
| "loss": 1.4780995178222656, |
| "step": 3550 |
| }, |
| { |
| "epoch": 5.5813953488372094, |
| "grad_norm": 1.374526858329773, |
| "learning_rate": 6.610814927646612e-05, |
| "loss": 1.4853521728515624, |
| "step": 3600 |
| }, |
| { |
| "epoch": 5.658914728682171, |
| "grad_norm": 1.2672253847122192, |
| "learning_rate": 6.556413883146557e-05, |
| "loss": 1.4863217163085938, |
| "step": 3650 |
| }, |
| { |
| "epoch": 5.736434108527131, |
| "grad_norm": 1.3603891134262085, |
| "learning_rate": 6.502012838646502e-05, |
| "loss": 1.4825819396972657, |
| "step": 3700 |
| }, |
| { |
| "epoch": 5.813953488372093, |
| "grad_norm": 1.4290482997894287, |
| "learning_rate": 6.447611794146448e-05, |
| "loss": 1.4989193725585936, |
| "step": 3750 |
| }, |
| { |
| "epoch": 5.891472868217054, |
| "grad_norm": 1.291866421699524, |
| "learning_rate": 6.393210749646393e-05, |
| "loss": 1.4657516479492188, |
| "step": 3800 |
| }, |
| { |
| "epoch": 5.9689922480620154, |
| "grad_norm": 1.3898322582244873, |
| "learning_rate": 6.338809705146339e-05, |
| "loss": 1.49338134765625, |
| "step": 3850 |
| }, |
| { |
| "epoch": 6.0, |
| "eval_loss": 1.5971275568008423, |
| "eval_rougeL": 0.3918850276301586, |
| "eval_runtime": 253.2975, |
| "eval_samples_per_second": 1.342, |
| "eval_steps_per_second": 0.17, |
| "step": 3870 |
| }, |
| { |
| "epoch": 6.046511627906977, |
| "grad_norm": 1.2466403245925903, |
| "learning_rate": 6.284408660646284e-05, |
| "loss": 1.4679217529296875, |
| "step": 3900 |
| }, |
| { |
| "epoch": 6.124031007751938, |
| "grad_norm": 1.2625627517700195, |
| "learning_rate": 6.23000761614623e-05, |
| "loss": 1.4217808532714844, |
| "step": 3950 |
| }, |
| { |
| "epoch": 6.2015503875969, |
| "grad_norm": 1.1851004362106323, |
| "learning_rate": 6.175606571646177e-05, |
| "loss": 1.4385812377929688, |
| "step": 4000 |
| }, |
| { |
| "epoch": 6.27906976744186, |
| "grad_norm": 1.7639460563659668, |
| "learning_rate": 6.12120552714612e-05, |
| "loss": 1.4053060913085937, |
| "step": 4050 |
| }, |
| { |
| "epoch": 6.3565891472868215, |
| "grad_norm": 1.1406972408294678, |
| "learning_rate": 6.066804482646067e-05, |
| "loss": 1.427724609375, |
| "step": 4100 |
| }, |
| { |
| "epoch": 6.434108527131783, |
| "grad_norm": 1.301949381828308, |
| "learning_rate": 6.0124034381460124e-05, |
| "loss": 1.430659637451172, |
| "step": 4150 |
| }, |
| { |
| "epoch": 6.511627906976744, |
| "grad_norm": 1.7478923797607422, |
| "learning_rate": 5.958002393645958e-05, |
| "loss": 1.4303857421875, |
| "step": 4200 |
| }, |
| { |
| "epoch": 6.589147286821706, |
| "grad_norm": 1.4526597261428833, |
| "learning_rate": 5.9036013491459044e-05, |
| "loss": 1.4464939880371093, |
| "step": 4250 |
| }, |
| { |
| "epoch": 6.666666666666667, |
| "grad_norm": 1.4695873260498047, |
| "learning_rate": 5.84920030464585e-05, |
| "loss": 1.420296173095703, |
| "step": 4300 |
| }, |
| { |
| "epoch": 6.7441860465116275, |
| "grad_norm": 1.3710230588912964, |
| "learning_rate": 5.794799260145794e-05, |
| "loss": 1.4247969055175782, |
| "step": 4350 |
| }, |
| { |
| "epoch": 6.821705426356589, |
| "grad_norm": 1.7078025341033936, |
| "learning_rate": 5.7403982156457406e-05, |
| "loss": 1.4178988647460937, |
| "step": 4400 |
| }, |
| { |
| "epoch": 6.89922480620155, |
| "grad_norm": 1.216068148612976, |
| "learning_rate": 5.685997171145686e-05, |
| "loss": 1.4220449829101562, |
| "step": 4450 |
| }, |
| { |
| "epoch": 6.976744186046512, |
| "grad_norm": 1.4590216875076294, |
| "learning_rate": 5.631596126645632e-05, |
| "loss": 1.446116943359375, |
| "step": 4500 |
| }, |
| { |
| "epoch": 7.0, |
| "eval_loss": 1.5953580141067505, |
| "eval_rougeL": 0.3897547775732232, |
| "eval_runtime": 249.3717, |
| "eval_samples_per_second": 1.363, |
| "eval_steps_per_second": 0.172, |
| "step": 4515 |
| }, |
| { |
| "epoch": 7.054263565891473, |
| "grad_norm": 1.1623547077178955, |
| "learning_rate": 5.5771950821455775e-05, |
| "loss": 1.3962391662597655, |
| "step": 4550 |
| }, |
| { |
| "epoch": 7.131782945736434, |
| "grad_norm": 1.2934234142303467, |
| "learning_rate": 5.522794037645523e-05, |
| "loss": 1.3965945434570313, |
| "step": 4600 |
| }, |
| { |
| "epoch": 7.209302325581396, |
| "grad_norm": 1.8001725673675537, |
| "learning_rate": 5.4683929931454694e-05, |
| "loss": 1.3921437072753906, |
| "step": 4650 |
| }, |
| { |
| "epoch": 7.286821705426356, |
| "grad_norm": 1.3875949382781982, |
| "learning_rate": 5.413991948645414e-05, |
| "loss": 1.372672119140625, |
| "step": 4700 |
| }, |
| { |
| "epoch": 7.364341085271318, |
| "grad_norm": 1.2738239765167236, |
| "learning_rate": 5.3595909041453593e-05, |
| "loss": 1.4048756408691405, |
| "step": 4750 |
| }, |
| { |
| "epoch": 7.441860465116279, |
| "grad_norm": 1.614818811416626, |
| "learning_rate": 5.3051898596453057e-05, |
| "loss": 1.3589810180664061, |
| "step": 4800 |
| }, |
| { |
| "epoch": 7.51937984496124, |
| "grad_norm": 1.3518016338348389, |
| "learning_rate": 5.250788815145251e-05, |
| "loss": 1.373852996826172, |
| "step": 4850 |
| }, |
| { |
| "epoch": 7.596899224806202, |
| "grad_norm": 1.4062671661376953, |
| "learning_rate": 5.196387770645197e-05, |
| "loss": 1.3677668762207031, |
| "step": 4900 |
| }, |
| { |
| "epoch": 7.674418604651163, |
| "grad_norm": 1.2638088464736938, |
| "learning_rate": 5.1419867261451426e-05, |
| "loss": 1.3687568664550782, |
| "step": 4950 |
| }, |
| { |
| "epoch": 7.751937984496124, |
| "grad_norm": 1.450870156288147, |
| "learning_rate": 5.0875856816450875e-05, |
| "loss": 1.3882672119140624, |
| "step": 5000 |
| }, |
| { |
| "epoch": 7.829457364341085, |
| "grad_norm": 1.4483550786972046, |
| "learning_rate": 5.033184637145033e-05, |
| "loss": 1.3580873107910156, |
| "step": 5050 |
| }, |
| { |
| "epoch": 7.906976744186046, |
| "grad_norm": 1.293081283569336, |
| "learning_rate": 4.978783592644979e-05, |
| "loss": 1.3714239501953125, |
| "step": 5100 |
| }, |
| { |
| "epoch": 7.984496124031008, |
| "grad_norm": 1.4119977951049805, |
| "learning_rate": 4.9243825481449244e-05, |
| "loss": 1.395829315185547, |
| "step": 5150 |
| }, |
| { |
| "epoch": 8.0, |
| "eval_loss": 1.589362382888794, |
| "eval_rougeL": 0.39690853763350886, |
| "eval_runtime": 254.3839, |
| "eval_samples_per_second": 1.337, |
| "eval_steps_per_second": 0.169, |
| "step": 5160 |
| } |
| ], |
| "logging_steps": 50, |
| "max_steps": 9675, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 15, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "EarlyStoppingCallback": { |
| "args": { |
| "early_stopping_patience": 3, |
| "early_stopping_threshold": 0.001 |
| }, |
| "attributes": { |
| "early_stopping_patience_counter": 0 |
| } |
| }, |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 8.34545321714258e+16, |
| "train_batch_size": 10, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
