| { | |
| "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": 0.41, | |
| "eval_loss": 0.7174346446990967, | |
| "eval_runtime": 2.8888, | |
| "eval_samples_per_second": 34.616, | |
| "eval_steps_per_second": 4.5, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.41, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.48, | |
| "eval_loss": 0.7533286809921265, | |
| "eval_runtime": 2.9185, | |
| "eval_samples_per_second": 34.265, | |
| "eval_steps_per_second": 4.454, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 1, | |
| "best_eval_accuracy": 0.48, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.7732255458831787, | |
| "eval_runtime": 2.9588, | |
| "eval_samples_per_second": 33.798, | |
| "eval_steps_per_second": 4.394, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.6637663245201111, | |
| "eval_runtime": 2.9949, | |
| "eval_samples_per_second": 33.39, | |
| "eval_steps_per_second": 4.341, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.49, | |
| "eval_loss": 0.6406551599502563, | |
| "eval_runtime": 3.0121, | |
| "eval_samples_per_second": 33.199, | |
| "eval_steps_per_second": 4.316, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.37, | |
| "eval_loss": 0.7757716178894043, | |
| "eval_runtime": 3.0199, | |
| "eval_samples_per_second": 33.114, | |
| "eval_steps_per_second": 4.305, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 2, | |
| "best_eval_accuracy": 0.61, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.5933136940002441, | |
| "eval_runtime": 3.0209, | |
| "eval_samples_per_second": 33.103, | |
| "eval_steps_per_second": 4.303, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6226850152015686, | |
| "eval_runtime": 3.022, | |
| "eval_samples_per_second": 33.09, | |
| "eval_steps_per_second": 4.302, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6461802124977112, | |
| "eval_runtime": 3.0241, | |
| "eval_samples_per_second": 33.067, | |
| "eval_steps_per_second": 4.299, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6519502997398376, | |
| "eval_runtime": 3.0256, | |
| "eval_samples_per_second": 33.051, | |
| "eval_steps_per_second": 4.297, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.5907068252563477, | |
| "eval_runtime": 3.0374, | |
| "eval_samples_per_second": 32.923, | |
| "eval_steps_per_second": 4.28, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6254033446311951, | |
| "eval_runtime": 3.0382, | |
| "eval_samples_per_second": 32.914, | |
| "eval_steps_per_second": 4.279, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6456963419914246, | |
| "eval_runtime": 3.0326, | |
| "eval_samples_per_second": 32.975, | |
| "eval_steps_per_second": 4.287, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5731363892555237, | |
| "eval_runtime": 3.0341, | |
| "eval_samples_per_second": 32.959, | |
| "eval_steps_per_second": 4.285, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6087689995765686, | |
| "eval_runtime": 3.0329, | |
| "eval_samples_per_second": 32.972, | |
| "eval_steps_per_second": 4.286, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.5721583962440491, | |
| "eval_runtime": 3.0324, | |
| "eval_samples_per_second": 32.977, | |
| "eval_steps_per_second": 4.287, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6186696887016296, | |
| "eval_runtime": 3.0346, | |
| "eval_samples_per_second": 32.953, | |
| "eval_steps_per_second": 4.284, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6932394504547119, | |
| "eval_runtime": 3.0352, | |
| "eval_samples_per_second": 32.947, | |
| "eval_steps_per_second": 4.283, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6067947149276733, | |
| "eval_runtime": 3.0404, | |
| "eval_samples_per_second": 32.89, | |
| "eval_steps_per_second": 4.276, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 6, | |
| "best_eval_accuracy": 0.67, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.0075, | |
| "loss": 0.7336, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.5740391612052917, | |
| "eval_runtime": 3.0379, | |
| "eval_samples_per_second": 32.918, | |
| "eval_steps_per_second": 4.279, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.5791218280792236, | |
| "eval_runtime": 3.0433, | |
| "eval_samples_per_second": 32.859, | |
| "eval_steps_per_second": 4.272, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.7414693236351013, | |
| "eval_runtime": 3.0449, | |
| "eval_samples_per_second": 32.841, | |
| "eval_steps_per_second": 4.269, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6274610757827759, | |
| "eval_runtime": 3.0405, | |
| "eval_samples_per_second": 32.889, | |
| "eval_steps_per_second": 4.276, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.651451051235199, | |
| "eval_runtime": 3.0416, | |
| "eval_samples_per_second": 32.878, | |
| "eval_steps_per_second": 4.274, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6618791818618774, | |
| "eval_runtime": 3.038, | |
| "eval_samples_per_second": 32.916, | |
| "eval_steps_per_second": 4.279, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.7296016216278076, | |
| "eval_runtime": 3.0371, | |
| "eval_samples_per_second": 32.926, | |
| "eval_steps_per_second": 4.28, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6984413862228394, | |
| "eval_runtime": 3.0395, | |
| "eval_samples_per_second": 32.9, | |
| "eval_steps_per_second": 4.277, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.7812673449516296, | |
| "eval_runtime": 3.0404, | |
| "eval_samples_per_second": 32.891, | |
| "eval_steps_per_second": 4.276, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.7498847842216492, | |
| "eval_runtime": 3.0347, | |
| "eval_samples_per_second": 32.952, | |
| "eval_steps_per_second": 4.284, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.8273061513900757, | |
| "eval_runtime": 3.0399, | |
| "eval_samples_per_second": 32.896, | |
| "eval_steps_per_second": 4.277, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.7841309905052185, | |
| "eval_runtime": 3.0369, | |
| "eval_samples_per_second": 32.928, | |
| "eval_steps_per_second": 4.281, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.739948570728302, | |
| "eval_runtime": 3.0364, | |
| "eval_samples_per_second": 32.934, | |
| "eval_steps_per_second": 4.281, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6789054274559021, | |
| "eval_runtime": 3.039, | |
| "eval_samples_per_second": 32.905, | |
| "eval_steps_per_second": 4.278, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.7219232320785522, | |
| "eval_runtime": 3.0362, | |
| "eval_samples_per_second": 32.936, | |
| "eval_steps_per_second": 4.282, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.7322627305984497, | |
| "eval_runtime": 3.046, | |
| "eval_samples_per_second": 32.83, | |
| "eval_steps_per_second": 4.268, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 19, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.7056111097335815, | |
| "eval_runtime": 3.0461, | |
| "eval_samples_per_second": 32.829, | |
| "eval_steps_per_second": 4.268, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.6668945550918579, | |
| "eval_runtime": 3.0403, | |
| "eval_samples_per_second": 32.892, | |
| "eval_steps_per_second": 4.276, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6746095418930054, | |
| "eval_runtime": 3.04, | |
| "eval_samples_per_second": 32.894, | |
| "eval_steps_per_second": 4.276, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.6932029128074646, | |
| "eval_runtime": 3.041, | |
| "eval_samples_per_second": 32.884, | |
| "eval_steps_per_second": 4.275, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.005, | |
| "loss": 0.371, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.6695249080657959, | |
| "eval_runtime": 3.0405, | |
| "eval_samples_per_second": 32.89, | |
| "eval_steps_per_second": 4.276, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.709083080291748, | |
| "eval_runtime": 3.0423, | |
| "eval_samples_per_second": 32.87, | |
| "eval_steps_per_second": 4.273, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6841559410095215, | |
| "eval_runtime": 3.042, | |
| "eval_samples_per_second": 32.873, | |
| "eval_steps_per_second": 4.274, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6723530292510986, | |
| "eval_runtime": 3.0406, | |
| "eval_samples_per_second": 32.888, | |
| "eval_steps_per_second": 4.275, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6937786936759949, | |
| "eval_runtime": 3.0427, | |
| "eval_samples_per_second": 32.866, | |
| "eval_steps_per_second": 4.273, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6779208183288574, | |
| "eval_runtime": 3.04, | |
| "eval_samples_per_second": 32.895, | |
| "eval_steps_per_second": 4.276, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6893536448478699, | |
| "eval_runtime": 3.0459, | |
| "eval_samples_per_second": 32.831, | |
| "eval_steps_per_second": 4.268, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.6746262311935425, | |
| "eval_runtime": 3.0395, | |
| "eval_samples_per_second": 32.9, | |
| "eval_steps_per_second": 4.277, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.7162497043609619, | |
| "eval_runtime": 3.043, | |
| "eval_samples_per_second": 32.862, | |
| "eval_steps_per_second": 4.272, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6891855597496033, | |
| "eval_runtime": 3.0445, | |
| "eval_samples_per_second": 32.846, | |
| "eval_steps_per_second": 4.27, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.6887782216072083, | |
| "eval_runtime": 3.0443, | |
| "eval_samples_per_second": 32.848, | |
| "eval_steps_per_second": 4.27, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.649270236492157, | |
| "eval_runtime": 3.0392, | |
| "eval_samples_per_second": 32.903, | |
| "eval_steps_per_second": 4.277, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6620053648948669, | |
| "eval_runtime": 3.0493, | |
| "eval_samples_per_second": 32.794, | |
| "eval_steps_per_second": 4.263, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.661324143409729, | |
| "eval_runtime": 3.0426, | |
| "eval_samples_per_second": 32.866, | |
| "eval_steps_per_second": 4.273, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6567308902740479, | |
| "eval_runtime": 3.042, | |
| "eval_samples_per_second": 32.874, | |
| "eval_steps_per_second": 4.274, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.688966691493988, | |
| "eval_runtime": 3.0474, | |
| "eval_samples_per_second": 32.815, | |
| "eval_steps_per_second": 4.266, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6883623003959656, | |
| "eval_runtime": 3.0444, | |
| "eval_samples_per_second": 32.847, | |
| "eval_steps_per_second": 4.27, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6547176837921143, | |
| "eval_runtime": 3.0476, | |
| "eval_samples_per_second": 32.813, | |
| "eval_steps_per_second": 4.266, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6830791234970093, | |
| "eval_runtime": 3.0462, | |
| "eval_samples_per_second": 32.828, | |
| "eval_steps_per_second": 4.268, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6529337167739868, | |
| "eval_runtime": 3.0477, | |
| "eval_samples_per_second": 32.812, | |
| "eval_steps_per_second": 4.266, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.0025, | |
| "loss": 0.2458, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6792705655097961, | |
| "eval_runtime": 3.0475, | |
| "eval_samples_per_second": 32.814, | |
| "eval_steps_per_second": 4.266, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6768913269042969, | |
| "eval_runtime": 3.0472, | |
| "eval_samples_per_second": 32.817, | |
| "eval_steps_per_second": 4.266, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6765817403793335, | |
| "eval_runtime": 3.0486, | |
| "eval_samples_per_second": 32.802, | |
| "eval_steps_per_second": 4.264, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6510599255561829, | |
| "eval_runtime": 3.0483, | |
| "eval_samples_per_second": 32.806, | |
| "eval_steps_per_second": 4.265, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6574416160583496, | |
| "eval_runtime": 3.0488, | |
| "eval_samples_per_second": 32.8, | |
| "eval_steps_per_second": 4.264, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6445488929748535, | |
| "eval_runtime": 3.049, | |
| "eval_samples_per_second": 32.797, | |
| "eval_steps_per_second": 4.264, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.646786093711853, | |
| "eval_runtime": 3.0497, | |
| "eval_samples_per_second": 32.79, | |
| "eval_steps_per_second": 4.263, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6413030028343201, | |
| "eval_runtime": 3.05, | |
| "eval_samples_per_second": 32.786, | |
| "eval_steps_per_second": 4.262, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6591331362724304, | |
| "eval_runtime": 3.0505, | |
| "eval_samples_per_second": 32.782, | |
| "eval_steps_per_second": 4.262, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.6374020576477051, | |
| "eval_runtime": 3.0507, | |
| "eval_samples_per_second": 32.779, | |
| "eval_steps_per_second": 4.261, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6688305735588074, | |
| "eval_runtime": 3.0527, | |
| "eval_samples_per_second": 32.758, | |
| "eval_steps_per_second": 4.259, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6512439250946045, | |
| "eval_runtime": 3.0522, | |
| "eval_samples_per_second": 32.763, | |
| "eval_steps_per_second": 4.259, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6465023756027222, | |
| "eval_runtime": 3.0527, | |
| "eval_samples_per_second": 32.757, | |
| "eval_steps_per_second": 4.258, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6601850390434265, | |
| "eval_runtime": 3.055, | |
| "eval_samples_per_second": 32.733, | |
| "eval_steps_per_second": 4.255, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6481786966323853, | |
| "eval_runtime": 3.0527, | |
| "eval_samples_per_second": 32.757, | |
| "eval_steps_per_second": 4.258, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6433829665184021, | |
| "eval_runtime": 3.0536, | |
| "eval_samples_per_second": 32.749, | |
| "eval_steps_per_second": 4.257, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6523444652557373, | |
| "eval_runtime": 3.0546, | |
| "eval_samples_per_second": 32.737, | |
| "eval_steps_per_second": 4.256, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6502225399017334, | |
| "eval_runtime": 3.0521, | |
| "eval_samples_per_second": 32.764, | |
| "eval_steps_per_second": 4.259, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6447280049324036, | |
| "eval_runtime": 3.0523, | |
| "eval_samples_per_second": 32.763, | |
| "eval_steps_per_second": 4.259, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6426615715026855, | |
| "eval_runtime": 3.0559, | |
| "eval_samples_per_second": 32.723, | |
| "eval_steps_per_second": 4.254, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.2218, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.6424870491027832, | |
| "eval_runtime": 3.0595, | |
| "eval_samples_per_second": 32.685, | |
| "eval_steps_per_second": 4.249, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 35, | |
| "best_eval_accuracy": 0.69, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.39303060150146485, | |
| "train_runtime": 1626.5542, | |
| "train_samples_per_second": 19.673, | |
| "train_steps_per_second": 1.23 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |