{ "best_metric": 0.4608415365219116, "best_model_checkpoint": "./Vit-GPT2-COCO2017Flickr-85k-11/checkpoint-4500", "epoch": 1.959503592423253, "eval_steps": 500, "global_step": 10500, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.09330969487729775, "grad_norm": 1.14389169216156, "learning_rate": 4.8444693293517484e-05, "loss": 0.378, "step": 500 }, { "epoch": 0.09330969487729775, "eval_gen_len": 11.7725, "eval_loss": 0.469292551279068, "eval_rouge1": 40.2274, "eval_rouge2": 15.0119, "eval_rougeL": 36.4563, "eval_rougeLsum": 36.4656, "eval_runtime": 450.3853, "eval_samples_per_second": 8.881, "eval_steps_per_second": 2.22, "step": 500 }, { "epoch": 0.1866193897545955, "grad_norm": 0.8436835408210754, "learning_rate": 4.6889386587034965e-05, "loss": 0.3748, "step": 1000 }, { "epoch": 0.1866193897545955, "eval_gen_len": 12.16675, "eval_loss": 0.46400758624076843, "eval_rouge1": 40.199, "eval_rouge2": 15.321, "eval_rougeL": 36.4279, "eval_rougeLsum": 36.4457, "eval_runtime": 431.1423, "eval_samples_per_second": 9.278, "eval_steps_per_second": 2.319, "step": 1000 }, { "epoch": 0.27992908463189325, "grad_norm": 0.8730084896087646, "learning_rate": 4.5334079880552446e-05, "loss": 0.374, "step": 1500 }, { "epoch": 0.27992908463189325, "eval_gen_len": 11.8, "eval_loss": 0.4668748080730438, "eval_rouge1": 39.9523, "eval_rouge2": 15.0587, "eval_rougeL": 36.3639, "eval_rougeLsum": 36.375, "eval_runtime": 424.8986, "eval_samples_per_second": 9.414, "eval_steps_per_second": 2.354, "step": 1500 }, { "epoch": 0.373238779509191, "grad_norm": 0.8020262718200684, "learning_rate": 4.3778773174069934e-05, "loss": 0.3721, "step": 2000 }, { "epoch": 0.373238779509191, "eval_gen_len": 11.2095, "eval_loss": 0.46446970105171204, "eval_rouge1": 40.3597, "eval_rouge2": 15.2173, "eval_rougeL": 36.6938, "eval_rougeLsum": 36.705, "eval_runtime": 410.7504, "eval_samples_per_second": 9.738, "eval_steps_per_second": 2.435, "step": 2000 }, { "epoch": 0.4665484743864888, "grad_norm": 0.8208432793617249, "learning_rate": 4.222346646758741e-05, "loss": 0.3673, "step": 2500 }, { "epoch": 0.4665484743864888, "eval_gen_len": 11.93425, "eval_loss": 0.4631750285625458, "eval_rouge1": 40.3875, "eval_rouge2": 15.2532, "eval_rougeL": 36.5923, "eval_rougeLsum": 36.6182, "eval_runtime": 441.242, "eval_samples_per_second": 9.065, "eval_steps_per_second": 2.266, "step": 2500 }, { "epoch": 0.5598581692637865, "grad_norm": 0.807547926902771, "learning_rate": 4.066815976110489e-05, "loss": 0.365, "step": 3000 }, { "epoch": 0.5598581692637865, "eval_gen_len": 12.26475, "eval_loss": 0.4623391330242157, "eval_rouge1": 39.9395, "eval_rouge2": 15.0315, "eval_rougeL": 36.1682, "eval_rougeLsum": 36.1781, "eval_runtime": 440.5083, "eval_samples_per_second": 9.08, "eval_steps_per_second": 2.27, "step": 3000 }, { "epoch": 0.6531678641410843, "grad_norm": 0.6434993147850037, "learning_rate": 3.911285305462237e-05, "loss": 0.3652, "step": 3500 }, { "epoch": 0.6531678641410843, "eval_gen_len": 11.8965, "eval_loss": 0.4611285328865051, "eval_rouge1": 39.8792, "eval_rouge2": 14.9961, "eval_rougeL": 36.2488, "eval_rougeLsum": 36.2734, "eval_runtime": 429.9024, "eval_samples_per_second": 9.304, "eval_steps_per_second": 2.326, "step": 3500 }, { "epoch": 0.746477559018382, "grad_norm": 0.8740318417549133, "learning_rate": 3.755754634813985e-05, "loss": 0.3601, "step": 4000 }, { "epoch": 0.746477559018382, "eval_gen_len": 12.0545, "eval_loss": 0.4625248610973358, "eval_rouge1": 40.57, "eval_rouge2": 15.2972, "eval_rougeL": 36.8012, "eval_rougeLsum": 36.8227, "eval_runtime": 431.0422, "eval_samples_per_second": 9.28, "eval_steps_per_second": 2.32, "step": 4000 }, { "epoch": 0.8397872538956798, "grad_norm": 0.7394187450408936, "learning_rate": 3.600223964165734e-05, "loss": 0.3574, "step": 4500 }, { "epoch": 0.8397872538956798, "eval_gen_len": 11.72875, "eval_loss": 0.4608415365219116, "eval_rouge1": 40.3276, "eval_rouge2": 15.1742, "eval_rougeL": 36.7679, "eval_rougeLsum": 36.7575, "eval_runtime": 467.6774, "eval_samples_per_second": 8.553, "eval_steps_per_second": 2.138, "step": 4500 }, { "epoch": 0.9330969487729776, "grad_norm": 0.841187059879303, "learning_rate": 3.444693293517482e-05, "loss": 0.351, "step": 5000 }, { "epoch": 0.9330969487729776, "eval_gen_len": 11.76625, "eval_loss": 0.4650140404701233, "eval_rouge1": 40.7345, "eval_rouge2": 15.5295, "eval_rougeL": 37.0769, "eval_rougeLsum": 37.0911, "eval_runtime": 475.4278, "eval_samples_per_second": 8.413, "eval_steps_per_second": 2.103, "step": 5000 }, { "epoch": 1.0264066436502752, "grad_norm": 0.8910800814628601, "learning_rate": 3.28916262286923e-05, "loss": 0.3322, "step": 5500 }, { "epoch": 1.0264066436502752, "eval_gen_len": 12.06, "eval_loss": 0.48306143283843994, "eval_rouge1": 40.5582, "eval_rouge2": 15.2954, "eval_rougeL": 36.6682, "eval_rougeLsum": 36.6694, "eval_runtime": 444.1346, "eval_samples_per_second": 9.006, "eval_steps_per_second": 2.252, "step": 5500 }, { "epoch": 1.119716338527573, "grad_norm": 0.9060000777244568, "learning_rate": 3.1336319522209783e-05, "loss": 0.2914, "step": 6000 }, { "epoch": 1.119716338527573, "eval_gen_len": 11.8405, "eval_loss": 0.4901696741580963, "eval_rouge1": 40.054, "eval_rouge2": 15.019, "eval_rougeL": 36.5476, "eval_rougeLsum": 36.556, "eval_runtime": 432.4768, "eval_samples_per_second": 9.249, "eval_steps_per_second": 2.312, "step": 6000 }, { "epoch": 1.2130260334048708, "grad_norm": 0.9239944219589233, "learning_rate": 2.978101281572726e-05, "loss": 0.2945, "step": 6500 }, { "epoch": 1.2130260334048708, "eval_gen_len": 11.84225, "eval_loss": 0.48632147908210754, "eval_rouge1": 40.3126, "eval_rouge2": 15.3154, "eval_rougeL": 36.61, "eval_rougeLsum": 36.6146, "eval_runtime": 433.2065, "eval_samples_per_second": 9.233, "eval_steps_per_second": 2.308, "step": 6500 }, { "epoch": 1.3063357282821686, "grad_norm": 0.9310933351516724, "learning_rate": 2.8225706109244742e-05, "loss": 0.2845, "step": 7000 }, { "epoch": 1.3063357282821686, "eval_gen_len": 12.0445, "eval_loss": 0.48827481269836426, "eval_rouge1": 40.228, "eval_rouge2": 15.0904, "eval_rougeL": 36.3179, "eval_rougeLsum": 36.3086, "eval_runtime": 445.7913, "eval_samples_per_second": 8.973, "eval_steps_per_second": 2.243, "step": 7000 }, { "epoch": 1.3996454231594662, "grad_norm": 0.8383007645606995, "learning_rate": 2.6670399402762224e-05, "loss": 0.2879, "step": 7500 }, { "epoch": 1.3996454231594662, "eval_gen_len": 11.93575, "eval_loss": 0.4833485782146454, "eval_rouge1": 40.6501, "eval_rouge2": 15.5682, "eval_rougeL": 36.8945, "eval_rougeLsum": 36.8823, "eval_runtime": 438.3851, "eval_samples_per_second": 9.124, "eval_steps_per_second": 2.281, "step": 7500 }, { "epoch": 1.492955118036764, "grad_norm": 0.8517706990242004, "learning_rate": 2.5115092696279708e-05, "loss": 0.2859, "step": 8000 }, { "epoch": 1.492955118036764, "eval_gen_len": 12.17425, "eval_loss": 0.4833007752895355, "eval_rouge1": 40.3187, "eval_rouge2": 15.0418, "eval_rougeL": 36.3561, "eval_rougeLsum": 36.3582, "eval_runtime": 441.1227, "eval_samples_per_second": 9.068, "eval_steps_per_second": 2.267, "step": 8000 }, { "epoch": 1.5862648129140617, "grad_norm": 0.951680064201355, "learning_rate": 2.355978598979719e-05, "loss": 0.2844, "step": 8500 }, { "epoch": 1.5862648129140617, "eval_gen_len": 12.17025, "eval_loss": 0.48842066526412964, "eval_rouge1": 40.2896, "eval_rouge2": 15.1032, "eval_rougeL": 36.4039, "eval_rougeLsum": 36.3862, "eval_runtime": 458.4907, "eval_samples_per_second": 8.724, "eval_steps_per_second": 2.181, "step": 8500 }, { "epoch": 1.6795745077913595, "grad_norm": 0.6958986520767212, "learning_rate": 2.200447928331467e-05, "loss": 0.2838, "step": 9000 }, { "epoch": 1.6795745077913595, "eval_gen_len": 11.95875, "eval_loss": 0.49017783999443054, "eval_rouge1": 40.3419, "eval_rouge2": 15.1863, "eval_rougeL": 36.4631, "eval_rougeLsum": 36.4728, "eval_runtime": 462.8508, "eval_samples_per_second": 8.642, "eval_steps_per_second": 2.161, "step": 9000 }, { "epoch": 1.7728842026686573, "grad_norm": 0.876482367515564, "learning_rate": 2.0449172576832152e-05, "loss": 0.2789, "step": 9500 }, { "epoch": 1.7728842026686573, "eval_gen_len": 12.05675, "eval_loss": 0.48646843433380127, "eval_rouge1": 40.6284, "eval_rouge2": 15.3404, "eval_rougeL": 36.7035, "eval_rougeLsum": 36.6876, "eval_runtime": 447.3126, "eval_samples_per_second": 8.942, "eval_steps_per_second": 2.236, "step": 9500 }, { "epoch": 1.8661938975459549, "grad_norm": 0.8670400381088257, "learning_rate": 1.8893865870349633e-05, "loss": 0.2758, "step": 10000 }, { "epoch": 1.8661938975459549, "eval_gen_len": 11.823, "eval_loss": 0.49086475372314453, "eval_rouge1": 40.1138, "eval_rouge2": 14.9247, "eval_rougeL": 36.4884, "eval_rougeLsum": 36.4836, "eval_runtime": 445.2778, "eval_samples_per_second": 8.983, "eval_steps_per_second": 2.246, "step": 10000 }, { "epoch": 1.959503592423253, "grad_norm": 0.92488694190979, "learning_rate": 1.7338559163867114e-05, "loss": 0.2741, "step": 10500 }, { "epoch": 1.959503592423253, "eval_gen_len": 11.95375, "eval_loss": 0.4892323315143585, "eval_rouge1": 40.3204, "eval_rouge2": 14.9594, "eval_rougeL": 36.539, "eval_rougeLsum": 36.5311, "eval_runtime": 454.063, "eval_samples_per_second": 8.809, "eval_steps_per_second": 2.202, "step": 10500 } ], "logging_steps": 500, "max_steps": 16074, "num_input_tokens_seen": 0, "num_train_epochs": 3, "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": 3.0317572780399264e+19, "train_batch_size": 4, "trial_name": null, "trial_params": null }