| { | |
| "best_metric": 60.62761059424247, | |
| "best_model_checkpoint": "/data/jcanete/all_results/mlqa/distillbeto/epochs_3_bs_16_lr_5e-5/checkpoint-12300", | |
| "epoch": 3.0, | |
| "global_step": 15381, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06, | |
| "eval_exact_match": 17.4, | |
| "eval_f1": 30.32932709945737, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.838437032702685e-05, | |
| "loss": 3.719, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "eval_exact_match": 20.6, | |
| "eval_f1": 37.076640043048776, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "eval_exact_match": 24.8, | |
| "eval_f1": 41.84428151388094, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.6758988362265134e-05, | |
| "loss": 2.9745, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "eval_exact_match": 24.4, | |
| "eval_f1": 42.64360101358078, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5133606397503415e-05, | |
| "loss": 2.7579, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "eval_exact_match": 26.8, | |
| "eval_f1": 48.268680757504505, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "eval_exact_match": 28.2, | |
| "eval_f1": 49.756554071203084, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.3508224432741696e-05, | |
| "loss": 2.5993, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "eval_exact_match": 29.0, | |
| "eval_f1": 48.999017405606445, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "eval_exact_match": 28.2, | |
| "eval_f1": 50.21409850637284, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.188284246797998e-05, | |
| "loss": 2.5658, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "eval_exact_match": 30.8, | |
| "eval_f1": 51.69176784487257, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.0260711267147784e-05, | |
| "loss": 2.4763, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "eval_exact_match": 32.6, | |
| "eval_f1": 53.82386360217891, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "eval_exact_match": 31.6, | |
| "eval_f1": 53.311489367004775, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.8635329302386065e-05, | |
| "loss": 2.3893, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "eval_exact_match": 32.0, | |
| "eval_f1": 53.23265973484611, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "eval_exact_match": 31.8, | |
| "eval_f1": 54.46584036020615, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.7009947337624346e-05, | |
| "loss": 2.3713, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "eval_exact_match": 30.4, | |
| "eval_f1": 53.87761850361369, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.5387816136792146e-05, | |
| "loss": 2.3229, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "eval_exact_match": 33.8, | |
| "eval_f1": 56.159600258090194, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "eval_exact_match": 34.8, | |
| "eval_f1": 57.45088180281838, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.376243417203043e-05, | |
| "loss": 2.2892, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "eval_exact_match": 33.4, | |
| "eval_f1": 57.09233458055284, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "eval_exact_match": 33.6, | |
| "eval_f1": 57.65540686906378, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.213705220726871e-05, | |
| "loss": 2.0818, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "eval_exact_match": 34.0, | |
| "eval_f1": 56.63309020916699, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.051167024250699e-05, | |
| "loss": 1.9928, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "eval_exact_match": 34.8, | |
| "eval_f1": 57.55733406275571, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "eval_exact_match": 35.4, | |
| "eval_f1": 56.60799657495974, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.888628827774527e-05, | |
| "loss": 1.9304, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "eval_exact_match": 33.8, | |
| "eval_f1": 56.337531192657984, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "eval_exact_match": 35.8, | |
| "eval_f1": 58.39477883856857, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.7260906312983555e-05, | |
| "loss": 1.984, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "eval_exact_match": 35.2, | |
| "eval_f1": 57.36112838371099, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.5635524348221836e-05, | |
| "loss": 1.9726, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "eval_exact_match": 35.6, | |
| "eval_f1": 57.98892993007854, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "eval_exact_match": 36.8, | |
| "eval_f1": 57.73167903988771, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.4010142383460113e-05, | |
| "loss": 1.9706, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "eval_exact_match": 34.2, | |
| "eval_f1": 57.1473434081718, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "eval_exact_match": 36.6, | |
| "eval_f1": 58.059032558136906, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.2384760418698394e-05, | |
| "loss": 1.9159, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "eval_exact_match": 36.8, | |
| "eval_f1": 57.88380136273412, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.07626292178662e-05, | |
| "loss": 1.9019, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "eval_exact_match": 36.6, | |
| "eval_f1": 58.80808564736181, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "eval_exact_match": 36.2, | |
| "eval_f1": 58.473738166006726, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.9137247253104482e-05, | |
| "loss": 1.9478, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "eval_exact_match": 35.6, | |
| "eval_f1": 58.20909169330069, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "eval_exact_match": 35.0, | |
| "eval_f1": 58.29322380617596, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.751186528834276e-05, | |
| "loss": 1.8909, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "eval_exact_match": 35.8, | |
| "eval_f1": 59.20379968263694, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.5889734087510566e-05, | |
| "loss": 1.7441, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "eval_exact_match": 36.6, | |
| "eval_f1": 59.70561445422752, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "eval_exact_match": 36.0, | |
| "eval_f1": 59.06336874221, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.4264352122748847e-05, | |
| "loss": 1.6342, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "eval_exact_match": 34.8, | |
| "eval_f1": 58.16179505760981, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "eval_exact_match": 35.0, | |
| "eval_f1": 58.02777082755699, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.2638970157987128e-05, | |
| "loss": 1.6129, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "eval_exact_match": 34.6, | |
| "eval_f1": 58.29141560339857, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.101358819322541e-05, | |
| "loss": 1.5814, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "eval_exact_match": 36.6, | |
| "eval_f1": 58.77641537872069, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "eval_exact_match": 37.2, | |
| "eval_f1": 60.62761059424247, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.388206228463689e-06, | |
| "loss": 1.6041, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "eval_exact_match": 36.0, | |
| "eval_f1": 59.05191316875973, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "eval_exact_match": 34.0, | |
| "eval_f1": 58.3915453062455, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.76282426370197e-06, | |
| "loss": 1.6054, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "eval_exact_match": 35.8, | |
| "eval_f1": 59.222805863165625, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.137442298940251e-06, | |
| "loss": 1.5882, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "eval_exact_match": 34.6, | |
| "eval_f1": 58.24542684372941, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "eval_exact_match": 35.4, | |
| "eval_f1": 58.46770769703503, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.512060334178532e-06, | |
| "loss": 1.5841, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "eval_exact_match": 35.8, | |
| "eval_f1": 58.797998172208, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "eval_exact_match": 35.4, | |
| "eval_f1": 59.277714258266194, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.8899291333463367e-06, | |
| "loss": 1.5605, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "eval_exact_match": 35.8, | |
| "eval_f1": 59.45930485720972, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.2645471685846174e-06, | |
| "loss": 1.5615, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "eval_exact_match": 35.2, | |
| "eval_f1": 58.96715598234879, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "eval_exact_match": 35.0, | |
| "eval_f1": 58.67411209812842, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 15381, | |
| "total_flos": 2.2092851276461056e+16, | |
| "train_loss": 2.057734964200962, | |
| "train_runtime": 14184.383, | |
| "train_samples_per_second": 17.348, | |
| "train_steps_per_second": 1.084 | |
| } | |
| ], | |
| "max_steps": 15381, | |
| "num_train_epochs": 3, | |
| "total_flos": 2.2092851276461056e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |