| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "best_supernet_model_checkpoint": null, | |
| "epoch": 20.0, | |
| "global_step": 2300, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8235294117647058, | |
| "eval_combined_score": 0.8490469358475097, | |
| "eval_f1": 0.8745644599303136, | |
| "eval_loss": 0.46352434158325195, | |
| "eval_runtime": 0.5288, | |
| "eval_samples_per_second": 771.53, | |
| "eval_steps_per_second": 7.564, | |
| "step": 115 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8357843137254902, | |
| "eval_combined_score": 0.8596312872975278, | |
| "eval_f1": 0.8834782608695653, | |
| "eval_loss": 0.40375280380249023, | |
| "eval_runtime": 0.6306, | |
| "eval_samples_per_second": 647.039, | |
| "eval_steps_per_second": 6.344, | |
| "step": 115 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8235294117647058, | |
| "eval_combined_score": 0.8492647058823529, | |
| "eval_f1": 0.875, | |
| "eval_loss": 0.4734054207801819, | |
| "eval_runtime": 0.4906, | |
| "eval_samples_per_second": 831.692, | |
| "eval_steps_per_second": 8.154, | |
| "step": 230 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8480392156862745, | |
| "eval_combined_score": 0.8712985194077632, | |
| "eval_f1": 0.8945578231292517, | |
| "eval_loss": 0.3588949739933014, | |
| "eval_runtime": 0.628, | |
| "eval_samples_per_second": 649.705, | |
| "eval_steps_per_second": 6.37, | |
| "step": 230 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.803921568627451, | |
| "eval_combined_score": 0.8359541836536595, | |
| "eval_f1": 0.867986798679868, | |
| "eval_loss": 0.9021703004837036, | |
| "eval_runtime": 0.5062, | |
| "eval_samples_per_second": 806.001, | |
| "eval_steps_per_second": 7.902, | |
| "step": 345 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8504901960784313, | |
| "eval_combined_score": 0.8743269010442241, | |
| "eval_f1": 0.8981636060100167, | |
| "eval_loss": 0.6408894658088684, | |
| "eval_runtime": 0.6431, | |
| "eval_samples_per_second": 634.428, | |
| "eval_steps_per_second": 6.22, | |
| "step": 345 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.8406862745098039, | |
| "eval_combined_score": 0.8645969966031009, | |
| "eval_f1": 0.8885077186963979, | |
| "eval_loss": 0.733981728553772, | |
| "eval_runtime": 0.5322, | |
| "eval_samples_per_second": 766.697, | |
| "eval_steps_per_second": 7.517, | |
| "step": 460 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.8578431372549019, | |
| "eval_combined_score": 0.8797690262545697, | |
| "eval_f1": 0.9016949152542373, | |
| "eval_loss": 0.6668654680252075, | |
| "eval_runtime": 0.664, | |
| "eval_samples_per_second": 614.462, | |
| "eval_steps_per_second": 6.024, | |
| "step": 460 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.35, | |
| "learning_rate": 2.6738413736593183e-05, | |
| "loss": 0.6907, | |
| "step": 500 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.8235294117647058, | |
| "eval_combined_score": 0.8511586452762923, | |
| "eval_f1": 0.8787878787878788, | |
| "eval_loss": 0.839835524559021, | |
| "eval_runtime": 0.4417, | |
| "eval_samples_per_second": 923.667, | |
| "eval_steps_per_second": 9.056, | |
| "step": 575 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.8799019607843137, | |
| "eval_combined_score": 0.8976366453317077, | |
| "eval_f1": 0.9153713298791019, | |
| "eval_loss": 0.6096086502075195, | |
| "eval_runtime": 0.5485, | |
| "eval_samples_per_second": 743.792, | |
| "eval_steps_per_second": 7.292, | |
| "step": 575 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.8186274509803921, | |
| "eval_combined_score": 0.8459575611066344, | |
| "eval_f1": 0.8732876712328768, | |
| "eval_loss": 1.0232644081115723, | |
| "eval_runtime": 0.4729, | |
| "eval_samples_per_second": 862.676, | |
| "eval_steps_per_second": 8.458, | |
| "step": 690 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.8406862745098039, | |
| "eval_combined_score": 0.8645969966031009, | |
| "eval_f1": 0.8885077186963979, | |
| "eval_loss": 0.8277555108070374, | |
| "eval_runtime": 0.6413, | |
| "eval_samples_per_second": 636.161, | |
| "eval_steps_per_second": 6.237, | |
| "step": 690 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.8137254901960784, | |
| "eval_combined_score": 0.8382706873363064, | |
| "eval_f1": 0.8628158844765343, | |
| "eval_loss": 0.9062389135360718, | |
| "eval_runtime": 0.5264, | |
| "eval_samples_per_second": 775.021, | |
| "eval_steps_per_second": 7.598, | |
| "step": 805 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.8455882352941176, | |
| "eval_combined_score": 0.8670419052576783, | |
| "eval_f1": 0.8884955752212389, | |
| "eval_loss": 0.696999728679657, | |
| "eval_runtime": 0.6178, | |
| "eval_samples_per_second": 660.427, | |
| "eval_steps_per_second": 6.475, | |
| "step": 805 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.8235294117647058, | |
| "eval_combined_score": 0.8499090357792601, | |
| "eval_f1": 0.8762886597938144, | |
| "eval_loss": 0.9299286603927612, | |
| "eval_runtime": 0.5047, | |
| "eval_samples_per_second": 808.406, | |
| "eval_steps_per_second": 7.926, | |
| "step": 920 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.8455882352941176, | |
| "eval_combined_score": 0.8700302985515815, | |
| "eval_f1": 0.8944723618090453, | |
| "eval_loss": 0.7925681471824646, | |
| "eval_runtime": 0.6718, | |
| "eval_samples_per_second": 607.307, | |
| "eval_steps_per_second": 5.954, | |
| "step": 920 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 8.7, | |
| "learning_rate": 1.829214077705104e-05, | |
| "loss": 0.0889, | |
| "step": 1000 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.821078431372549, | |
| "eval_combined_score": 0.8487794864138549, | |
| "eval_f1": 0.8764805414551609, | |
| "eval_loss": 0.9827358722686768, | |
| "eval_runtime": 0.4726, | |
| "eval_samples_per_second": 863.398, | |
| "eval_steps_per_second": 8.465, | |
| "step": 1035 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.8529411764705882, | |
| "eval_combined_score": 0.8754501800720287, | |
| "eval_f1": 0.8979591836734693, | |
| "eval_loss": 0.7438695430755615, | |
| "eval_runtime": 0.6707, | |
| "eval_samples_per_second": 608.321, | |
| "eval_steps_per_second": 5.964, | |
| "step": 1035 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.8284313725490197, | |
| "eval_combined_score": 0.8546918767507002, | |
| "eval_f1": 0.8809523809523808, | |
| "eval_loss": 0.8921936750411987, | |
| "eval_runtime": 0.4539, | |
| "eval_samples_per_second": 898.928, | |
| "eval_steps_per_second": 8.813, | |
| "step": 1150 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.8602941176470589, | |
| "eval_combined_score": 0.8807536446120909, | |
| "eval_f1": 0.9012131715771231, | |
| "eval_loss": 0.6455641984939575, | |
| "eval_runtime": 0.6908, | |
| "eval_samples_per_second": 590.661, | |
| "eval_steps_per_second": 5.791, | |
| "step": 1150 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.821078431372549, | |
| "eval_combined_score": 0.8491946778711483, | |
| "eval_f1": 0.8773109243697478, | |
| "eval_loss": 0.9563024044036865, | |
| "eval_runtime": 0.5754, | |
| "eval_samples_per_second": 709.094, | |
| "eval_steps_per_second": 6.952, | |
| "step": 1265 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.8553921568627451, | |
| "eval_combined_score": 0.877611188787909, | |
| "eval_f1": 0.8998302207130731, | |
| "eval_loss": 0.6870017051696777, | |
| "eval_runtime": 0.6557, | |
| "eval_samples_per_second": 622.197, | |
| "eval_steps_per_second": 6.1, | |
| "step": 1265 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.8259803921568627, | |
| "eval_combined_score": 0.8525131943748538, | |
| "eval_f1": 0.8790459965928449, | |
| "eval_loss": 0.9068766832351685, | |
| "eval_runtime": 0.5352, | |
| "eval_samples_per_second": 762.351, | |
| "eval_steps_per_second": 7.474, | |
| "step": 1380 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.8676470588235294, | |
| "eval_combined_score": 0.8866207322089674, | |
| "eval_f1": 0.9055944055944055, | |
| "eval_loss": 0.5972063541412354, | |
| "eval_runtime": 0.6204, | |
| "eval_samples_per_second": 657.677, | |
| "eval_steps_per_second": 6.448, | |
| "step": 1380 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.8235294117647058, | |
| "eval_combined_score": 0.8515640369860318, | |
| "eval_f1": 0.8795986622073578, | |
| "eval_loss": 0.9531628489494324, | |
| "eval_runtime": 0.4389, | |
| "eval_samples_per_second": 929.512, | |
| "eval_steps_per_second": 9.113, | |
| "step": 1495 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.8651960784313726, | |
| "eval_combined_score": 0.8849377272572807, | |
| "eval_f1": 0.9046793760831888, | |
| "eval_loss": 0.641593337059021, | |
| "eval_runtime": 0.5799, | |
| "eval_samples_per_second": 703.627, | |
| "eval_steps_per_second": 6.898, | |
| "step": 1495 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 13.04, | |
| "learning_rate": 8.410113675774417e-06, | |
| "loss": 0.0339, | |
| "step": 1500 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.8259803921568627, | |
| "eval_combined_score": 0.8523064353946708, | |
| "eval_f1": 0.8786324786324787, | |
| "eval_loss": 0.9255245327949524, | |
| "eval_runtime": 0.5277, | |
| "eval_samples_per_second": 773.205, | |
| "eval_steps_per_second": 7.58, | |
| "step": 1610 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.8578431372549019, | |
| "eval_combined_score": 0.8796018407362944, | |
| "eval_f1": 0.901360544217687, | |
| "eval_loss": 0.7534880042076111, | |
| "eval_runtime": 0.6535, | |
| "eval_samples_per_second": 624.357, | |
| "eval_steps_per_second": 6.121, | |
| "step": 1610 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.821078431372549, | |
| "eval_combined_score": 0.8479320115696364, | |
| "eval_f1": 0.8747855917667238, | |
| "eval_loss": 0.9056081771850586, | |
| "eval_runtime": 0.3636, | |
| "eval_samples_per_second": 1122.037, | |
| "eval_steps_per_second": 11.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.8602941176470589, | |
| "eval_combined_score": 0.8802346244977852, | |
| "eval_f1": 0.9001751313485115, | |
| "eval_loss": 0.6561837196350098, | |
| "eval_runtime": 0.492, | |
| "eval_samples_per_second": 829.301, | |
| "eval_steps_per_second": 8.13, | |
| "step": 1725 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.8088235294117647, | |
| "eval_combined_score": 0.8396277115497361, | |
| "eval_f1": 0.8704318936877077, | |
| "eval_loss": 0.9581977725028992, | |
| "eval_runtime": 0.3672, | |
| "eval_samples_per_second": 1111.013, | |
| "eval_steps_per_second": 10.892, | |
| "step": 1840 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.8553921568627451, | |
| "eval_combined_score": 0.877440541804456, | |
| "eval_f1": 0.899488926746167, | |
| "eval_loss": 0.7179605960845947, | |
| "eval_runtime": 0.4744, | |
| "eval_samples_per_second": 860.007, | |
| "eval_steps_per_second": 8.431, | |
| "step": 1840 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.8308823529411765, | |
| "eval_combined_score": 0.856867322480775, | |
| "eval_f1": 0.8828522920203735, | |
| "eval_loss": 0.9101623892784119, | |
| "eval_runtime": 0.3973, | |
| "eval_samples_per_second": 1026.915, | |
| "eval_steps_per_second": 10.068, | |
| "step": 1955 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.8627450980392157, | |
| "eval_combined_score": 0.883262583383869, | |
| "eval_f1": 0.9037800687285222, | |
| "eval_loss": 0.6792478561401367, | |
| "eval_runtime": 0.4894, | |
| "eval_samples_per_second": 833.646, | |
| "eval_steps_per_second": 8.173, | |
| "step": 1955 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 17.39, | |
| "learning_rate": 1.41888323488484e-06, | |
| "loss": 0.0195, | |
| "step": 2000 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.8284313725490197, | |
| "eval_combined_score": 0.8546918767507002, | |
| "eval_f1": 0.8809523809523808, | |
| "eval_loss": 0.8795027136802673, | |
| "eval_runtime": 0.4269, | |
| "eval_samples_per_second": 955.795, | |
| "eval_steps_per_second": 9.371, | |
| "step": 2070 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.8651960784313726, | |
| "eval_combined_score": 0.8851023570049782, | |
| "eval_f1": 0.9050086355785838, | |
| "eval_loss": 0.6670258641242981, | |
| "eval_runtime": 0.488, | |
| "eval_samples_per_second": 836.011, | |
| "eval_steps_per_second": 8.196, | |
| "step": 2070 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.8235294117647058, | |
| "eval_combined_score": 0.8509538950715421, | |
| "eval_f1": 0.8783783783783784, | |
| "eval_loss": 0.9087016582489014, | |
| "eval_runtime": 0.4014, | |
| "eval_samples_per_second": 1016.504, | |
| "eval_steps_per_second": 9.966, | |
| "step": 2185 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.8627450980392157, | |
| "eval_combined_score": 0.883262583383869, | |
| "eval_f1": 0.9037800687285222, | |
| "eval_loss": 0.6763139367103577, | |
| "eval_runtime": 0.5343, | |
| "eval_samples_per_second": 763.583, | |
| "eval_steps_per_second": 7.486, | |
| "step": 2185 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 704, 2: 768, 3: 768, 4: 768, 5: 640, 6: 640, 7: 512, 8: 512, 9: 320, 10: 192, 11: 192, 12: 995, 13: 930, 14: 928, 15: 931, 16: 872, 17: 787, 18: 672, 19: 579, 20: 409, 21: 291, 22: 300, 23: 308})])", | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.8235294117647058, | |
| "eval_combined_score": 0.8509538950715421, | |
| "eval_f1": 0.8783783783783784, | |
| "eval_loss": 0.9081816673278809, | |
| "eval_runtime": 0.4205, | |
| "eval_samples_per_second": 970.314, | |
| "eval_steps_per_second": 9.513, | |
| "step": 2300 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 768, 1: 768, 2: 768, 3: 768, 4: 768, 5: 768, 6: 768, 7: 768, 8: 768, 9: 768, 10: 768, 11: 768, 12: 3072, 13: 3072, 14: 3072, 15: 3072, 16: 3072, 17: 3072, 18: 3072, 19: 3072, 20: 3072, 21: 3072, 22: 3072, 23: 3072})])", | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.8627450980392157, | |
| "eval_combined_score": 0.883262583383869, | |
| "eval_f1": 0.9037800687285222, | |
| "eval_loss": 0.6761664748191833, | |
| "eval_runtime": 0.4936, | |
| "eval_samples_per_second": 826.651, | |
| "eval_steps_per_second": 8.104, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "step": 2300, | |
| "total_flos": 4825456755302400.0, | |
| "train_loss": 0.18312535887179168, | |
| "train_runtime": 4630.0259, | |
| "train_samples_per_second": 15.844, | |
| "train_steps_per_second": 0.497 | |
| } | |
| ], | |
| "max_steps": 2300, | |
| "min_subnet_acc": null, | |
| "min_subnet_best_acc": null, | |
| "num_train_epochs": 20, | |
| "supernet_acc": null, | |
| "supernet_best_acc": null, | |
| "total_flos": 4825456755302400.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |