| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "best_supernet_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "global_step": 41175, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 0.06, | |
| "learning_rate": 2.998921662946385e-05, | |
| "loss": 5.928, | |
| "step": 500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 0.12, | |
| "learning_rate": 2.9956708418044054e-05, | |
| "loss": 3.6128, | |
| "step": 1000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 0.18, | |
| "learning_rate": 2.990265252786665e-05, | |
| "loss": 3.0946, | |
| "step": 1500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 0.24, | |
| "learning_rate": 2.982673683458095e-05, | |
| "loss": 2.836, | |
| "step": 2000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 0.3, | |
| "learning_rate": 2.972924541394191e-05, | |
| "loss": 2.6325, | |
| "step": 2500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 0.36, | |
| "learning_rate": 2.9610320134543718e-05, | |
| "loss": 2.5171, | |
| "step": 3000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 0.43, | |
| "learning_rate": 2.9470134055329297e-05, | |
| "loss": 2.3989, | |
| "step": 3500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 0.49, | |
| "learning_rate": 2.9308891173756593e-05, | |
| "loss": 2.2948, | |
| "step": 4000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 0.55, | |
| "learning_rate": 2.9126826128943387e-05, | |
| "loss": 2.1684, | |
| "step": 4500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 0.61, | |
| "learning_rate": 2.892420386022268e-05, | |
| "loss": 2.1619, | |
| "step": 5000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 0.67, | |
| "learning_rate": 2.8701319221605467e-05, | |
| "loss": 2.0291, | |
| "step": 5500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 0.73, | |
| "learning_rate": 2.8458496552711964e-05, | |
| "loss": 1.9988, | |
| "step": 6000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 0.79, | |
| "learning_rate": 2.819608920679567e-05, | |
| "loss": 1.9522, | |
| "step": 6500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 0.85, | |
| "learning_rate": 2.7914479036547047e-05, | |
| "loss": 1.9296, | |
| "step": 7000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 0.91, | |
| "learning_rate": 2.7614075838425082e-05, | |
| "loss": 1.8511, | |
| "step": 7500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 0.97, | |
| "learning_rate": 2.7295316756325375e-05, | |
| "loss": 1.8184, | |
| "step": 8000 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 384, 1: 192, 2: 384, 3: 448, 4: 448, 5: 384, 6: 448, 7: 448, 8: 610, 9: 544, 10: 631, 11: 653, 12: 591, 13: 549, 14: 508, 15: 485})])", | |
| "epoch": 1.0, | |
| "eval_HasAns_exact": 62.90485829959514, | |
| "eval_HasAns_f1": 69.4564460879015, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 68.35996635828427, | |
| "eval_NoAns_f1": 68.35996635828427, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 65.64474016676493, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 68.91584371339046, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 65.6363176956119, | |
| "eval_f1": 68.90742124223719, | |
| "eval_runtime": 19.8249, | |
| "eval_samples_per_second": 612.057, | |
| "eval_steps_per_second": 4.792, | |
| "eval_total": 11873, | |
| "step": 8235 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 512, 1: 512, 2: 512, 3: 512, 4: 512, 5: 512, 6: 512, 7: 512, 8: 2048, 9: 2048, 10: 2048, 11: 2048, 12: 2048, 13: 2048, 14: 2048, 15: 2048})])", | |
| "epoch": 1.0, | |
| "eval_HasAns_exact": 67.6450742240216, | |
| "eval_HasAns_f1": 73.95858758792157, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 68.29268292682927, | |
| "eval_NoAns_f1": 68.29268292682927, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 67.97776467615599, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 71.13000145045065, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 67.96934220500295, | |
| "eval_f1": 71.12157897929758, | |
| "eval_runtime": 23.0026, | |
| "eval_samples_per_second": 527.506, | |
| "eval_steps_per_second": 4.13, | |
| "eval_total": 11873, | |
| "step": 8235 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 1.03, | |
| "learning_rate": 2.6959356482568783e-05, | |
| "loss": 1.5854, | |
| "step": 8500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 1.09, | |
| "learning_rate": 2.6605337532847057e-05, | |
| "loss": 1.4178, | |
| "step": 9000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 1.15, | |
| "learning_rate": 2.6234430605598938e-05, | |
| "loss": 1.4158, | |
| "step": 9500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 1.21, | |
| "learning_rate": 2.5847175441080488e-05, | |
| "loss": 1.3996, | |
| "step": 10000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 1.28, | |
| "learning_rate": 2.54441355693482e-05, | |
| "loss": 1.3329, | |
| "step": 10500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 1.34, | |
| "learning_rate": 2.502589749021534e-05, | |
| "loss": 1.3709, | |
| "step": 11000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 1.4, | |
| "learning_rate": 2.459394962034141e-05, | |
| "loss": 1.3598, | |
| "step": 11500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 1.46, | |
| "learning_rate": 2.4148096507143148e-05, | |
| "loss": 1.3531, | |
| "step": 12000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 1.52, | |
| "learning_rate": 2.3688051425634673e-05, | |
| "loss": 1.3325, | |
| "step": 12500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 1.58, | |
| "learning_rate": 2.321536357377945e-05, | |
| "loss": 1.3222, | |
| "step": 13000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 1.64, | |
| "learning_rate": 2.273072080247337e-05, | |
| "loss": 1.3055, | |
| "step": 13500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 1.7, | |
| "learning_rate": 2.2234828359298165e-05, | |
| "loss": 1.2822, | |
| "step": 14000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 1.76, | |
| "learning_rate": 2.172943072286878e-05, | |
| "loss": 1.2867, | |
| "step": 14500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 1.82, | |
| "learning_rate": 2.1213237947389485e-05, | |
| "loss": 1.2798, | |
| "step": 15000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 1.88, | |
| "learning_rate": 2.0688003726754053e-05, | |
| "loss": 1.2475, | |
| "step": 15500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 1.94, | |
| "learning_rate": 2.0154492376837755e-05, | |
| "loss": 1.2647, | |
| "step": 16000 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 384, 1: 192, 2: 384, 3: 448, 4: 448, 5: 384, 6: 448, 7: 448, 8: 610, 9: 544, 10: 631, 11: 653, 12: 591, 13: 549, 14: 508, 15: 485})])", | |
| "epoch": 2.0, | |
| "eval_HasAns_exact": 63.917004048582996, | |
| "eval_HasAns_f1": 70.22830220550237, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 72.0605550883095, | |
| "eval_NoAns_f1": 72.0605550883095, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 67.99460961846205, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 71.14574037515544, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 67.99460961846205, | |
| "eval_f1": 71.14574037515541, | |
| "eval_runtime": 17.7943, | |
| "eval_samples_per_second": 681.904, | |
| "eval_steps_per_second": 5.339, | |
| "eval_total": 11873, | |
| "step": 16470 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 512, 1: 512, 2: 512, 3: 512, 4: 512, 5: 512, 6: 512, 7: 512, 8: 2048, 9: 2048, 10: 2048, 11: 2048, 12: 2048, 13: 2048, 14: 2048, 15: 2048})])", | |
| "epoch": 2.0, | |
| "eval_HasAns_exact": 68.60661268556005, | |
| "eval_HasAns_f1": 74.58138647737387, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 71.0681244743482, | |
| "eval_NoAns_f1": 71.0681244743482, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 69.839130800977, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 72.82224029629188, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 69.839130800977, | |
| "eval_f1": 72.82224029629188, | |
| "eval_runtime": 18.7704, | |
| "eval_samples_per_second": 646.444, | |
| "eval_steps_per_second": 5.061, | |
| "eval_total": 11873, | |
| "step": 16470 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.0, | |
| "learning_rate": 1.961674714301095e-05, | |
| "loss": 1.2283, | |
| "step": 16500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.06, | |
| "learning_rate": 1.9069059444969563e-05, | |
| "loss": 0.9251, | |
| "step": 17000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.13, | |
| "learning_rate": 1.8515450490069793e-05, | |
| "loss": 0.913, | |
| "step": 17500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.19, | |
| "learning_rate": 1.7956725884832123e-05, | |
| "loss": 0.9281, | |
| "step": 18000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.25, | |
| "learning_rate": 1.7393698680023048e-05, | |
| "loss": 0.9135, | |
| "step": 18500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.31, | |
| "learning_rate": 1.682718818751133e-05, | |
| "loss": 0.9216, | |
| "step": 19000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.37, | |
| "learning_rate": 1.6258018788013082e-05, | |
| "loss": 0.9112, | |
| "step": 19500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.43, | |
| "learning_rate": 1.5687018731460723e-05, | |
| "loss": 0.8918, | |
| "step": 20000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.49, | |
| "learning_rate": 1.5116163375957171e-05, | |
| "loss": 0.8542, | |
| "step": 20500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.55, | |
| "learning_rate": 1.4543995705456567e-05, | |
| "loss": 0.8708, | |
| "step": 21000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.61, | |
| "learning_rate": 1.3972491608096893e-05, | |
| "loss": 0.8999, | |
| "step": 21500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.67, | |
| "learning_rate": 1.3402482731240843e-05, | |
| "loss": 0.8712, | |
| "step": 22000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.73, | |
| "learning_rate": 1.2834798546420376e-05, | |
| "loss": 0.8533, | |
| "step": 22500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.79, | |
| "learning_rate": 1.2272515909019886e-05, | |
| "loss": 0.9089, | |
| "step": 23000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.85, | |
| "learning_rate": 1.1711937270962604e-05, | |
| "loss": 0.8488, | |
| "step": 23500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.91, | |
| "learning_rate": 1.1156143390721824e-05, | |
| "loss": 0.8619, | |
| "step": 24000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 2.98, | |
| "learning_rate": 1.0605943054300711e-05, | |
| "loss": 0.8615, | |
| "step": 24500 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 384, 1: 192, 2: 384, 3: 448, 4: 448, 5: 384, 6: 448, 7: 448, 8: 610, 9: 544, 10: 631, 11: 653, 12: 591, 13: 549, 14: 508, 15: 485})])", | |
| "epoch": 3.0, | |
| "eval_HasAns_exact": 67.03778677462888, | |
| "eval_HasAns_f1": 73.62919057241987, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 68.86459209419681, | |
| "eval_NoAns_f1": 68.86459209419681, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 67.95249726269688, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 71.24348030938324, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 67.95249726269688, | |
| "eval_f1": 71.2434803093832, | |
| "eval_runtime": 20.6828, | |
| "eval_samples_per_second": 586.672, | |
| "eval_steps_per_second": 4.593, | |
| "eval_total": 11873, | |
| "step": 24705 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 512, 1: 512, 2: 512, 3: 512, 4: 512, 5: 512, 6: 512, 7: 512, 8: 2048, 9: 2048, 10: 2048, 11: 2048, 12: 2048, 13: 2048, 14: 2048, 15: 2048})])", | |
| "epoch": 3.0, | |
| "eval_HasAns_exact": 70.20917678812415, | |
| "eval_HasAns_f1": 76.31500557470775, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 70.51303616484441, | |
| "eval_NoAns_f1": 70.51303616484441, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 70.36132401246526, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 73.40986718157744, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 70.36132401246526, | |
| "eval_f1": 73.40986718157748, | |
| "eval_runtime": 24.0079, | |
| "eval_samples_per_second": 505.418, | |
| "eval_steps_per_second": 3.957, | |
| "eval_total": 11873, | |
| "step": 24705 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 3.04, | |
| "learning_rate": 1.0063217611277036e-05, | |
| "loss": 0.7353, | |
| "step": 25000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 3.1, | |
| "learning_rate": 9.526581842660798e-06, | |
| "loss": 0.6453, | |
| "step": 25500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 3.16, | |
| "learning_rate": 8.997910940432307e-06, | |
| "loss": 0.6729, | |
| "step": 26000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 3.22, | |
| "learning_rate": 8.477974221493395e-06, | |
| "loss": 0.6703, | |
| "step": 26500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 3.28, | |
| "learning_rate": 7.967528292844926e-06, | |
| "loss": 0.6633, | |
| "step": 27000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 3.34, | |
| "learning_rate": 7.467315950579387e-06, | |
| "loss": 0.6608, | |
| "step": 27500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 3.4, | |
| "learning_rate": 6.978065098971048e-06, | |
| "loss": 0.6479, | |
| "step": 28000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 3.46, | |
| "learning_rate": 6.500487691236567e-06, | |
| "loss": 0.6683, | |
| "step": 28500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 3.52, | |
| "learning_rate": 6.036196314774858e-06, | |
| "loss": 0.6597, | |
| "step": 29000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 3.58, | |
| "learning_rate": 5.584005940876061e-06, | |
| "loss": 0.6739, | |
| "step": 29500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 3.64, | |
| "learning_rate": 5.145517632566403e-06, | |
| "loss": 0.6578, | |
| "step": 30000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 3.7, | |
| "learning_rate": 4.721369473889145e-06, | |
| "loss": 0.6535, | |
| "step": 30500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 3.76, | |
| "learning_rate": 4.3129817358827764e-06, | |
| "loss": 0.6433, | |
| "step": 31000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 3.83, | |
| "learning_rate": 3.919312074840663e-06, | |
| "loss": 0.6612, | |
| "step": 31500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 3.89, | |
| "learning_rate": 3.541766926172923e-06, | |
| "loss": 0.6668, | |
| "step": 32000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 3.95, | |
| "learning_rate": 3.18089569001271e-06, | |
| "loss": 0.6456, | |
| "step": 32500 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 384, 1: 192, 2: 384, 3: 448, 4: 448, 5: 384, 6: 448, 7: 448, 8: 610, 9: 544, 10: 631, 11: 653, 12: 591, 13: 549, 14: 508, 15: 485})])", | |
| "epoch": 4.0, | |
| "eval_HasAns_exact": 68.69095816464238, | |
| "eval_HasAns_f1": 75.3675933308443, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 67.2834314550042, | |
| "eval_NoAns_f1": 67.2834314550042, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 67.98618714730902, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 71.31972486020774, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 67.98618714730902, | |
| "eval_f1": 71.31972486020777, | |
| "eval_runtime": 19.136, | |
| "eval_samples_per_second": 634.092, | |
| "eval_steps_per_second": 4.964, | |
| "eval_total": 11873, | |
| "step": 32940 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 512, 1: 512, 2: 512, 3: 512, 4: 512, 5: 512, 6: 512, 7: 512, 8: 2048, 9: 2048, 10: 2048, 11: 2048, 12: 2048, 13: 2048, 14: 2048, 15: 2048})])", | |
| "epoch": 4.0, | |
| "eval_HasAns_exact": 71.89608636977059, | |
| "eval_HasAns_f1": 78.17919334933525, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 67.70395290159799, | |
| "eval_NoAns_f1": 67.70395290159799, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 69.79701844521182, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 72.93407379557496, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 69.79701844521182, | |
| "eval_f1": 72.93407379557496, | |
| "eval_runtime": 20.2232, | |
| "eval_samples_per_second": 600.004, | |
| "eval_steps_per_second": 4.698, | |
| "eval_total": 11873, | |
| "step": 32940 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.01, | |
| "learning_rate": 2.8378933458497492e-06, | |
| "loss": 0.6269, | |
| "step": 33000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.07, | |
| "learning_rate": 2.5125184709830478e-06, | |
| "loss": 0.5658, | |
| "step": 33500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.13, | |
| "learning_rate": 2.204645347373324e-06, | |
| "loss": 0.5484, | |
| "step": 34000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.19, | |
| "learning_rate": 1.9153919029589925e-06, | |
| "loss": 0.5433, | |
| "step": 34500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.25, | |
| "learning_rate": 1.6451790566193852e-06, | |
| "loss": 0.5692, | |
| "step": 35000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.31, | |
| "learning_rate": 1.3944000195354374e-06, | |
| "loss": 0.5627, | |
| "step": 35500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.37, | |
| "learning_rate": 1.1638616961868014e-06, | |
| "loss": 0.5623, | |
| "step": 36000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.43, | |
| "learning_rate": 9.53778694922398e-07, | |
| "loss": 0.5493, | |
| "step": 36500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.49, | |
| "learning_rate": 7.632514341613572e-07, | |
| "loss": 0.5609, | |
| "step": 37000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.55, | |
| "learning_rate": 5.934413553993973e-07, | |
| "loss": 0.5642, | |
| "step": 37500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.61, | |
| "learning_rate": 4.445955646587174e-07, | |
| "loss": 0.5642, | |
| "step": 38000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.68, | |
| "learning_rate": 3.169306609276462e-07, | |
| "loss": 0.5636, | |
| "step": 38500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.74, | |
| "learning_rate": 2.1063242096787116e-07, | |
| "loss": 0.5666, | |
| "step": 39000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.8, | |
| "learning_rate": 1.258555289738389e-07, | |
| "loss": 0.5576, | |
| "step": 39500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.86, | |
| "learning_rate": 6.272335147777586e-08, | |
| "loss": 0.5597, | |
| "step": 40000 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.92, | |
| "learning_rate": 2.1327757827879212e-08, | |
| "loss": 0.5437, | |
| "step": 40500 | |
| }, | |
| { | |
| "compression_loss": 0.0, | |
| "epoch": 4.98, | |
| "learning_rate": 1.7289865009056939e-09, | |
| "loss": 0.569, | |
| "step": 41000 | |
| }, | |
| { | |
| "Minimum SubNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 384, 1: 192, 2: 384, 3: 448, 4: 448, 5: 384, 6: 448, 7: 448, 8: 610, 9: 544, 10: 631, 11: 653, 12: 591, 13: 549, 14: 508, 15: 485})])", | |
| "epoch": 5.0, | |
| "eval_HasAns_exact": 67.6450742240216, | |
| "eval_HasAns_f1": 74.25189753904719, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 68.56181665264928, | |
| "eval_NoAns_f1": 68.56181665264928, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 68.10410174345152, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 71.40278350976789, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 68.10410174345152, | |
| "eval_f1": 71.40278350976786, | |
| "eval_runtime": 15.8145, | |
| "eval_samples_per_second": 767.268, | |
| "eval_steps_per_second": 6.007, | |
| "eval_total": 11873, | |
| "step": 41175 | |
| }, | |
| { | |
| "SuperNet": "OrderedDict([(<ElasticityDim.WIDTH: 'width'>, {0: 512, 1: 512, 2: 512, 3: 512, 4: 512, 5: 512, 6: 512, 7: 512, 8: 2048, 9: 2048, 10: 2048, 11: 2048, 12: 2048, 13: 2048, 14: 2048, 15: 2048})])", | |
| "epoch": 5.0, | |
| "eval_HasAns_exact": 71.44062078272604, | |
| "eval_HasAns_f1": 77.61788590740903, | |
| "eval_HasAns_total": 5928, | |
| "eval_NoAns_exact": 69.38603868797308, | |
| "eval_NoAns_f1": 69.38603868797308, | |
| "eval_NoAns_total": 5945, | |
| "eval_best_exact": 70.41185883938347, | |
| "eval_best_exact_thresh": 0.0, | |
| "eval_best_f1": 73.49606903555316, | |
| "eval_best_f1_thresh": 0.0, | |
| "eval_exact": 70.41185883938347, | |
| "eval_f1": 73.49606903555318, | |
| "eval_runtime": 19.4373, | |
| "eval_samples_per_second": 624.265, | |
| "eval_steps_per_second": 4.888, | |
| "eval_total": 11873, | |
| "step": 41175 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 41175, | |
| "total_flos": 3.828077231602176e+16, | |
| "train_loss": 1.1978534093045001, | |
| "train_runtime": 40881.8794, | |
| "train_samples_per_second": 16.114, | |
| "train_steps_per_second": 1.007 | |
| } | |
| ], | |
| "max_steps": 41175, | |
| "min_subnet_acc": null, | |
| "min_subnet_best_acc": null, | |
| "num_train_epochs": 5, | |
| "supernet_acc": null, | |
| "supernet_best_acc": null, | |
| "total_flos": 3.828077231602176e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |