| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 80.0, | |
| "global_step": 2000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.4, | |
| "eval_loss": 0.32287806272506714, | |
| "eval_runtime": 2.9676, | |
| "eval_samples_per_second": 33.697, | |
| "eval_steps_per_second": 4.381, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.4, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.35066673159599304, | |
| "eval_runtime": 3.0298, | |
| "eval_samples_per_second": 33.006, | |
| "eval_steps_per_second": 4.291, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.39, | |
| "eval_loss": 0.31648871302604675, | |
| "eval_runtime": 3.0781, | |
| "eval_samples_per_second": 32.487, | |
| "eval_steps_per_second": 4.223, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.31587904691696167, | |
| "eval_runtime": 3.1134, | |
| "eval_samples_per_second": 32.119, | |
| "eval_steps_per_second": 4.176, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.35, | |
| "eval_loss": 0.3276338279247284, | |
| "eval_runtime": 3.1438, | |
| "eval_samples_per_second": 31.809, | |
| "eval_steps_per_second": 4.135, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.37, | |
| "eval_loss": 0.3255067765712738, | |
| "eval_runtime": 3.1564, | |
| "eval_samples_per_second": 31.681, | |
| "eval_steps_per_second": 4.119, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.2893008887767792, | |
| "eval_runtime": 3.169, | |
| "eval_samples_per_second": 31.555, | |
| "eval_steps_per_second": 4.102, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.30655354261398315, | |
| "eval_runtime": 3.1806, | |
| "eval_samples_per_second": 31.441, | |
| "eval_steps_per_second": 4.087, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.3015369772911072, | |
| "eval_runtime": 3.1746, | |
| "eval_samples_per_second": 31.5, | |
| "eval_steps_per_second": 4.095, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.293341726064682, | |
| "eval_runtime": 3.1847, | |
| "eval_samples_per_second": 31.4, | |
| "eval_steps_per_second": 4.082, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.45, | |
| "eval_loss": 0.29533302783966064, | |
| "eval_runtime": 3.1804, | |
| "eval_samples_per_second": 31.443, | |
| "eval_steps_per_second": 4.088, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.29426100850105286, | |
| "eval_runtime": 3.1802, | |
| "eval_samples_per_second": 31.444, | |
| "eval_steps_per_second": 4.088, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.28667622804641724, | |
| "eval_runtime": 3.1818, | |
| "eval_samples_per_second": 31.429, | |
| "eval_steps_per_second": 4.086, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.28816643357276917, | |
| "eval_runtime": 3.183, | |
| "eval_samples_per_second": 31.417, | |
| "eval_steps_per_second": 4.084, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.29221782088279724, | |
| "eval_runtime": 3.1833, | |
| "eval_samples_per_second": 31.414, | |
| "eval_steps_per_second": 4.084, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.28952255845069885, | |
| "eval_runtime": 3.1873, | |
| "eval_samples_per_second": 31.374, | |
| "eval_steps_per_second": 4.079, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 8, | |
| "best_eval_accuracy": 0.64, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.29009494185447693, | |
| "eval_runtime": 3.1886, | |
| "eval_samples_per_second": 31.362, | |
| "eval_steps_per_second": 4.077, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.28774628043174744, | |
| "eval_runtime": 3.1894, | |
| "eval_samples_per_second": 31.354, | |
| "eval_steps_per_second": 4.076, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.2908971309661865, | |
| "eval_runtime": 3.1888, | |
| "eval_samples_per_second": 31.36, | |
| "eval_steps_per_second": 4.077, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.0075, | |
| "loss": 0.5537, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.2870675325393677, | |
| "eval_runtime": 3.1962, | |
| "eval_samples_per_second": 31.287, | |
| "eval_steps_per_second": 4.067, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.28545090556144714, | |
| "eval_runtime": 3.1985, | |
| "eval_samples_per_second": 31.264, | |
| "eval_steps_per_second": 4.064, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.286283940076828, | |
| "eval_runtime": 3.1885, | |
| "eval_samples_per_second": 31.363, | |
| "eval_steps_per_second": 4.077, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.28589385747909546, | |
| "eval_runtime": 3.1956, | |
| "eval_samples_per_second": 31.293, | |
| "eval_steps_per_second": 4.068, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.28539586067199707, | |
| "eval_runtime": 3.1953, | |
| "eval_samples_per_second": 31.296, | |
| "eval_steps_per_second": 4.068, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2839178442955017, | |
| "eval_runtime": 3.1925, | |
| "eval_samples_per_second": 31.323, | |
| "eval_steps_per_second": 4.072, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.56, | |
| "eval_loss": 0.2858905494213104, | |
| "eval_runtime": 3.1947, | |
| "eval_samples_per_second": 31.302, | |
| "eval_steps_per_second": 4.069, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.28208523988723755, | |
| "eval_runtime": 3.1973, | |
| "eval_samples_per_second": 31.276, | |
| "eval_steps_per_second": 4.066, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.2830556631088257, | |
| "eval_runtime": 3.1983, | |
| "eval_samples_per_second": 31.266, | |
| "eval_steps_per_second": 4.065, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 16, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.2813069522380829, | |
| "eval_runtime": 3.2015, | |
| "eval_samples_per_second": 31.236, | |
| "eval_steps_per_second": 4.061, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 28, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.28120794892311096, | |
| "eval_runtime": 3.1996, | |
| "eval_samples_per_second": 31.254, | |
| "eval_steps_per_second": 4.063, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.27902308106422424, | |
| "eval_runtime": 3.201, | |
| "eval_samples_per_second": 31.24, | |
| "eval_steps_per_second": 4.061, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.2801484763622284, | |
| "eval_runtime": 3.2027, | |
| "eval_samples_per_second": 31.224, | |
| "eval_steps_per_second": 4.059, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.28051745891571045, | |
| "eval_runtime": 3.1989, | |
| "eval_samples_per_second": 31.26, | |
| "eval_steps_per_second": 4.064, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.28497397899627686, | |
| "eval_runtime": 3.1984, | |
| "eval_samples_per_second": 31.265, | |
| "eval_steps_per_second": 4.064, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.2781234681606293, | |
| "eval_runtime": 3.1956, | |
| "eval_samples_per_second": 31.293, | |
| "eval_steps_per_second": 4.068, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.280036598443985, | |
| "eval_runtime": 3.1973, | |
| "eval_samples_per_second": 31.276, | |
| "eval_steps_per_second": 4.066, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.28637388348579407, | |
| "eval_runtime": 3.1965, | |
| "eval_samples_per_second": 31.284, | |
| "eval_steps_per_second": 4.067, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.28161221742630005, | |
| "eval_runtime": 3.1956, | |
| "eval_samples_per_second": 31.293, | |
| "eval_steps_per_second": 4.068, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.28857657313346863, | |
| "eval_runtime": 3.203, | |
| "eval_samples_per_second": 31.221, | |
| "eval_steps_per_second": 4.059, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.005, | |
| "loss": 0.5047, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.310051292181015, | |
| "eval_runtime": 3.1962, | |
| "eval_samples_per_second": 31.287, | |
| "eval_steps_per_second": 4.067, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.28261107206344604, | |
| "eval_runtime": 3.1977, | |
| "eval_samples_per_second": 31.273, | |
| "eval_steps_per_second": 4.065, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.2800942361354828, | |
| "eval_runtime": 3.1967, | |
| "eval_samples_per_second": 31.282, | |
| "eval_steps_per_second": 4.067, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 29, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.2907128632068634, | |
| "eval_runtime": 3.1994, | |
| "eval_samples_per_second": 31.256, | |
| "eval_steps_per_second": 4.063, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.2894176244735718, | |
| "eval_runtime": 3.1965, | |
| "eval_samples_per_second": 31.284, | |
| "eval_steps_per_second": 4.067, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.2854782044887543, | |
| "eval_runtime": 3.2014, | |
| "eval_samples_per_second": 31.236, | |
| "eval_steps_per_second": 4.061, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.28108128905296326, | |
| "eval_runtime": 3.1981, | |
| "eval_samples_per_second": 31.269, | |
| "eval_steps_per_second": 4.065, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.2946772575378418, | |
| "eval_runtime": 3.2021, | |
| "eval_samples_per_second": 31.23, | |
| "eval_steps_per_second": 4.06, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.29523247480392456, | |
| "eval_runtime": 3.2001, | |
| "eval_samples_per_second": 31.249, | |
| "eval_steps_per_second": 4.062, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.2832265794277191, | |
| "eval_runtime": 3.2085, | |
| "eval_samples_per_second": 31.167, | |
| "eval_steps_per_second": 4.052, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.29542115330696106, | |
| "eval_runtime": 3.1989, | |
| "eval_samples_per_second": 31.261, | |
| "eval_steps_per_second": 4.064, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.2839832007884979, | |
| "eval_runtime": 3.2036, | |
| "eval_samples_per_second": 31.215, | |
| "eval_steps_per_second": 4.058, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.30789753794670105, | |
| "eval_runtime": 3.2016, | |
| "eval_samples_per_second": 31.235, | |
| "eval_steps_per_second": 4.06, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.27960124611854553, | |
| "eval_runtime": 3.2024, | |
| "eval_samples_per_second": 31.226, | |
| "eval_steps_per_second": 4.059, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.286190003156662, | |
| "eval_runtime": 3.2013, | |
| "eval_samples_per_second": 31.237, | |
| "eval_steps_per_second": 4.061, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.28528231382369995, | |
| "eval_runtime": 3.2039, | |
| "eval_samples_per_second": 31.212, | |
| "eval_steps_per_second": 4.058, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.29692474007606506, | |
| "eval_runtime": 3.2026, | |
| "eval_samples_per_second": 31.225, | |
| "eval_steps_per_second": 4.059, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.2865641117095947, | |
| "eval_runtime": 3.2043, | |
| "eval_samples_per_second": 31.208, | |
| "eval_steps_per_second": 4.057, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.28954190015792847, | |
| "eval_runtime": 3.2033, | |
| "eval_samples_per_second": 31.218, | |
| "eval_steps_per_second": 4.058, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.305752158164978, | |
| "eval_runtime": 3.2026, | |
| "eval_samples_per_second": 31.225, | |
| "eval_steps_per_second": 4.059, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.0025, | |
| "loss": 0.4502, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.2998403310775757, | |
| "eval_runtime": 3.2116, | |
| "eval_samples_per_second": 31.137, | |
| "eval_steps_per_second": 4.048, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.29743996262550354, | |
| "eval_runtime": 3.2044, | |
| "eval_samples_per_second": 31.207, | |
| "eval_steps_per_second": 4.057, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.2788439095020294, | |
| "eval_runtime": 3.2048, | |
| "eval_samples_per_second": 31.203, | |
| "eval_steps_per_second": 4.056, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.2882141172885895, | |
| "eval_runtime": 3.2042, | |
| "eval_samples_per_second": 31.209, | |
| "eval_steps_per_second": 4.057, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.2892961800098419, | |
| "eval_runtime": 3.2113, | |
| "eval_samples_per_second": 31.14, | |
| "eval_steps_per_second": 4.048, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.28336700797080994, | |
| "eval_runtime": 3.2095, | |
| "eval_samples_per_second": 31.157, | |
| "eval_steps_per_second": 4.05, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 46, | |
| "best_eval_accuracy": 0.7, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.2888612747192383, | |
| "eval_runtime": 3.205, | |
| "eval_samples_per_second": 31.201, | |
| "eval_steps_per_second": 4.056, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 65, | |
| "best_eval_accuracy": 0.72, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.2850661873817444, | |
| "eval_runtime": 3.2056, | |
| "eval_samples_per_second": 31.195, | |
| "eval_steps_per_second": 4.055, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 66, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.27734076976776123, | |
| "eval_runtime": 3.2023, | |
| "eval_samples_per_second": 31.228, | |
| "eval_steps_per_second": 4.06, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 66, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.2854936420917511, | |
| "eval_runtime": 3.2033, | |
| "eval_samples_per_second": 31.218, | |
| "eval_steps_per_second": 4.058, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 66, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.2902901768684387, | |
| "eval_runtime": 3.2035, | |
| "eval_samples_per_second": 31.216, | |
| "eval_steps_per_second": 4.058, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 66, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.28505316376686096, | |
| "eval_runtime": 3.2006, | |
| "eval_samples_per_second": 31.244, | |
| "eval_steps_per_second": 4.062, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 66, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.2892363369464874, | |
| "eval_runtime": 3.2052, | |
| "eval_samples_per_second": 31.2, | |
| "eval_steps_per_second": 4.056, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 66, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.2811330556869507, | |
| "eval_runtime": 3.2021, | |
| "eval_samples_per_second": 31.23, | |
| "eval_steps_per_second": 4.06, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 66, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.2880754768848419, | |
| "eval_runtime": 3.201, | |
| "eval_samples_per_second": 31.241, | |
| "eval_steps_per_second": 4.061, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 66, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.2892223000526428, | |
| "eval_runtime": 3.2049, | |
| "eval_samples_per_second": 31.202, | |
| "eval_steps_per_second": 4.056, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 66, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.2834882438182831, | |
| "eval_runtime": 3.1993, | |
| "eval_samples_per_second": 31.257, | |
| "eval_steps_per_second": 4.063, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 66, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.2799616754055023, | |
| "eval_runtime": 3.1962, | |
| "eval_samples_per_second": 31.287, | |
| "eval_steps_per_second": 4.067, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 66, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.2808959484100342, | |
| "eval_runtime": 3.199, | |
| "eval_samples_per_second": 31.259, | |
| "eval_steps_per_second": 4.064, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 66, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.28005915880203247, | |
| "eval_runtime": 3.1973, | |
| "eval_samples_per_second": 31.277, | |
| "eval_steps_per_second": 4.066, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 66, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.4329, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.2806113362312317, | |
| "eval_runtime": 3.1967, | |
| "eval_samples_per_second": 31.282, | |
| "eval_steps_per_second": 4.067, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 66, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.4853669128417969, | |
| "train_runtime": 1702.0431, | |
| "train_samples_per_second": 18.801, | |
| "train_steps_per_second": 1.175 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |