{ "best_metric": 48.94001149135019, "best_model_checkpoint": "/home/jcanete/ft-data/all_results/mlqa/albeto_base_2/epochs_4_bs_16_lr_2e-5/checkpoint-12900", "epoch": 4.0, "global_step": 20516, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.06, "eval_exact_match": 11.2, "eval_f1": 26.483192221593768, "step": 300 }, { "epoch": 0.1, "learning_rate": 1.951355039968805e-05, "loss": 4.0436, "step": 500 }, { "epoch": 0.12, "eval_exact_match": 15.0, "eval_f1": 33.51854251608216, "step": 600 }, { "epoch": 0.18, "eval_exact_match": 18.6, "eval_f1": 35.629522354460704, "step": 900 }, { "epoch": 0.19, "learning_rate": 1.902612595047768e-05, "loss": 3.2766, "step": 1000 }, { "epoch": 0.23, "eval_exact_match": 18.0, "eval_f1": 36.188007340871735, "step": 1200 }, { "epoch": 0.29, "learning_rate": 1.8538701501267304e-05, "loss": 3.0936, "step": 1500 }, { "epoch": 0.29, "eval_exact_match": 18.8, "eval_f1": 37.44597113423442, "step": 1500 }, { "epoch": 0.35, "eval_exact_match": 21.2, "eval_f1": 39.84498534299273, "step": 1800 }, { "epoch": 0.39, "learning_rate": 1.8051277052056932e-05, "loss": 2.9305, "step": 2000 }, { "epoch": 0.41, "eval_exact_match": 20.8, "eval_f1": 39.95193770849471, "step": 2100 }, { "epoch": 0.47, "eval_exact_match": 22.2, "eval_f1": 42.758832207716736, "step": 2400 }, { "epoch": 0.49, "learning_rate": 1.756482745174498e-05, "loss": 2.794, "step": 2500 }, { "epoch": 0.53, "eval_exact_match": 22.2, "eval_f1": 42.601350107858735, "step": 2700 }, { "epoch": 0.58, "learning_rate": 1.707740300253461e-05, "loss": 2.7894, "step": 3000 }, { "epoch": 0.58, "eval_exact_match": 24.0, "eval_f1": 43.534851112760435, "step": 3000 }, { "epoch": 0.64, "eval_exact_match": 22.8, "eval_f1": 43.62106996130737, "step": 3300 }, { "epoch": 0.68, "learning_rate": 1.6590953402222657e-05, "loss": 2.7014, "step": 3500 }, { "epoch": 0.7, "eval_exact_match": 22.8, "eval_f1": 44.321198455769256, "step": 3600 }, { "epoch": 0.76, "eval_exact_match": 22.6, "eval_f1": 44.57581116292914, "step": 3900 }, { "epoch": 0.78, "learning_rate": 1.6103528953012282e-05, "loss": 2.6815, "step": 4000 }, { "epoch": 0.82, "eval_exact_match": 23.4, "eval_f1": 46.30028588125735, "step": 4200 }, { "epoch": 0.88, "learning_rate": 1.561610450380191e-05, "loss": 2.6472, "step": 4500 }, { "epoch": 0.88, "eval_exact_match": 25.2, "eval_f1": 45.66133144463246, "step": 4500 }, { "epoch": 0.94, "eval_exact_match": 24.4, "eval_f1": 46.05729908209824, "step": 4800 }, { "epoch": 0.97, "learning_rate": 1.512868005459154e-05, "loss": 2.6125, "step": 5000 }, { "epoch": 0.99, "eval_exact_match": 24.6, "eval_f1": 45.667180594212915, "step": 5100 }, { "epoch": 1.05, "eval_exact_match": 23.2, "eval_f1": 45.63595896581295, "step": 5400 }, { "epoch": 1.07, "learning_rate": 1.4641255605381166e-05, "loss": 2.4544, "step": 5500 }, { "epoch": 1.11, "eval_exact_match": 23.0, "eval_f1": 45.57347635092067, "step": 5700 }, { "epoch": 1.17, "learning_rate": 1.4153831156170795e-05, "loss": 2.4307, "step": 6000 }, { "epoch": 1.17, "eval_exact_match": 22.8, "eval_f1": 45.4787090109933, "step": 6000 }, { "epoch": 1.23, "eval_exact_match": 24.2, "eval_f1": 46.144624410019276, "step": 6300 }, { "epoch": 1.27, "learning_rate": 1.3666406706960422e-05, "loss": 2.4623, "step": 6500 }, { "epoch": 1.29, "eval_exact_match": 23.2, "eval_f1": 46.0243228203072, "step": 6600 }, { "epoch": 1.35, "eval_exact_match": 24.2, "eval_f1": 46.67180386550054, "step": 6900 }, { "epoch": 1.36, "learning_rate": 1.3179957106648472e-05, "loss": 2.4077, "step": 7000 }, { "epoch": 1.4, "eval_exact_match": 25.4, "eval_f1": 47.08102740483588, "step": 7200 }, { "epoch": 1.46, "learning_rate": 1.2692532657438098e-05, "loss": 2.3967, "step": 7500 }, { "epoch": 1.46, "eval_exact_match": 25.2, "eval_f1": 46.94730205752481, "step": 7500 }, { "epoch": 1.52, "eval_exact_match": 22.6, "eval_f1": 45.74979168773037, "step": 7800 }, { "epoch": 1.56, "learning_rate": 1.2205108208227725e-05, "loss": 2.3936, "step": 8000 }, { "epoch": 1.58, "eval_exact_match": 22.8, "eval_f1": 45.53841968646729, "step": 8100 }, { "epoch": 1.64, "eval_exact_match": 23.8, "eval_f1": 46.600652318202485, "step": 8400 }, { "epoch": 1.66, "learning_rate": 1.1717683759017354e-05, "loss": 2.3773, "step": 8500 }, { "epoch": 1.7, "eval_exact_match": 22.4, "eval_f1": 46.012756869173074, "step": 8700 }, { "epoch": 1.75, "learning_rate": 1.123025930980698e-05, "loss": 2.3893, "step": 9000 }, { "epoch": 1.75, "eval_exact_match": 23.8, "eval_f1": 45.9605328124833, "step": 9000 }, { "epoch": 1.81, "eval_exact_match": 24.4, "eval_f1": 47.31331525220247, "step": 9300 }, { "epoch": 1.85, "learning_rate": 1.0742834860596609e-05, "loss": 2.4066, "step": 9500 }, { "epoch": 1.87, "eval_exact_match": 25.8, "eval_f1": 48.09576595590226, "step": 9600 }, { "epoch": 1.93, "eval_exact_match": 24.2, "eval_f1": 46.333298341879264, "step": 9900 }, { "epoch": 1.95, "learning_rate": 1.0256385260284656e-05, "loss": 2.3616, "step": 10000 }, { "epoch": 1.99, "eval_exact_match": 24.4, "eval_f1": 47.86325870979021, "step": 10200 }, { "epoch": 2.05, "learning_rate": 9.768960811074284e-06, "loss": 2.3018, "step": 10500 }, { "epoch": 2.05, "eval_exact_match": 23.4, "eval_f1": 46.869345750688815, "step": 10500 }, { "epoch": 2.11, "eval_exact_match": 23.2, "eval_f1": 46.75900565743204, "step": 10800 }, { "epoch": 2.14, "learning_rate": 9.281536361863913e-06, "loss": 2.2207, "step": 11000 }, { "epoch": 2.16, "eval_exact_match": 23.0, "eval_f1": 47.11768387241494, "step": 11100 }, { "epoch": 2.22, "eval_exact_match": 24.8, "eval_f1": 47.824589572947076, "step": 11400 }, { "epoch": 2.24, "learning_rate": 8.79411191265354e-06, "loss": 2.2133, "step": 11500 }, { "epoch": 2.28, "eval_exact_match": 22.6, "eval_f1": 46.81847093965424, "step": 11700 }, { "epoch": 2.34, "learning_rate": 8.306687463443166e-06, "loss": 2.1887, "step": 12000 }, { "epoch": 2.34, "eval_exact_match": 25.4, "eval_f1": 48.70213586118982, "step": 12000 }, { "epoch": 2.4, "eval_exact_match": 24.0, "eval_f1": 47.907936732879136, "step": 12300 }, { "epoch": 2.44, "learning_rate": 7.819263014232795e-06, "loss": 2.2084, "step": 12500 }, { "epoch": 2.46, "eval_exact_match": 24.2, "eval_f1": 47.50117367525339, "step": 12600 }, { "epoch": 2.52, "eval_exact_match": 25.6, "eval_f1": 48.94001149135019, "step": 12900 }, { "epoch": 2.53, "learning_rate": 7.3318385650224225e-06, "loss": 2.205, "step": 13000 }, { "epoch": 2.57, "eval_exact_match": 24.8, "eval_f1": 48.65674792844071, "step": 13200 }, { "epoch": 2.63, "learning_rate": 6.844414115812049e-06, "loss": 2.1895, "step": 13500 }, { "epoch": 2.63, "eval_exact_match": 25.6, "eval_f1": 48.791085379512644, "step": 13500 }, { "epoch": 2.69, "eval_exact_match": 24.4, "eval_f1": 46.9179679859715, "step": 13800 }, { "epoch": 2.73, "learning_rate": 6.357964515500098e-06, "loss": 2.2337, "step": 14000 }, { "epoch": 2.75, "eval_exact_match": 23.6, "eval_f1": 46.28350673077509, "step": 14100 }, { "epoch": 2.81, "eval_exact_match": 23.6, "eval_f1": 47.41817849560017, "step": 14400 }, { "epoch": 2.83, "learning_rate": 5.871514915188147e-06, "loss": 2.1879, "step": 14500 }, { "epoch": 2.87, "eval_exact_match": 22.8, "eval_f1": 46.458213496270524, "step": 14700 }, { "epoch": 2.92, "learning_rate": 5.384090465977774e-06, "loss": 2.2071, "step": 15000 }, { "epoch": 2.92, "eval_exact_match": 24.2, "eval_f1": 47.532040364746656, "step": 15000 }, { "epoch": 2.98, "eval_exact_match": 24.0, "eval_f1": 47.252480052113945, "step": 15300 }, { "epoch": 3.02, "learning_rate": 4.896666016767401e-06, "loss": 2.1374, "step": 15500 }, { "epoch": 3.04, "eval_exact_match": 23.4, "eval_f1": 46.915969224210855, "step": 15600 }, { "epoch": 3.1, "eval_exact_match": 23.4, "eval_f1": 47.627048464769736, "step": 15900 }, { "epoch": 3.12, "learning_rate": 4.409241567557029e-06, "loss": 2.1177, "step": 16000 }, { "epoch": 3.16, "eval_exact_match": 22.2, "eval_f1": 46.544258456129775, "step": 16200 }, { "epoch": 3.22, "learning_rate": 3.921817118346657e-06, "loss": 2.0913, "step": 16500 }, { "epoch": 3.22, "eval_exact_match": 23.0, "eval_f1": 46.50691177045911, "step": 16500 }, { "epoch": 3.28, "eval_exact_match": 23.2, "eval_f1": 47.050804068875436, "step": 16800 }, { "epoch": 3.31, "learning_rate": 3.435367518034705e-06, "loss": 2.0779, "step": 17000 }, { "epoch": 3.33, "eval_exact_match": 23.2, "eval_f1": 46.33502363014144, "step": 17100 }, { "epoch": 3.39, "eval_exact_match": 23.2, "eval_f1": 46.765575747268464, "step": 17400 }, { "epoch": 3.41, "learning_rate": 2.9479430688243326e-06, "loss": 2.0847, "step": 17500 }, { "epoch": 3.45, "eval_exact_match": 24.2, "eval_f1": 47.26636124409616, "step": 17700 }, { "epoch": 3.51, "learning_rate": 2.46051861961396e-06, "loss": 2.0733, "step": 18000 }, { "epoch": 3.51, "eval_exact_match": 23.0, "eval_f1": 46.90041208739893, "step": 18000 }, { "epoch": 3.57, "eval_exact_match": 24.2, "eval_f1": 47.60954713139801, "step": 18300 }, { "epoch": 3.61, "learning_rate": 1.9730941704035875e-06, "loss": 2.0735, "step": 18500 }, { "epoch": 3.63, "eval_exact_match": 23.4, "eval_f1": 47.03174405348153, "step": 18600 }, { "epoch": 3.68, "eval_exact_match": 22.4, "eval_f1": 46.133959338075734, "step": 18900 }, { "epoch": 3.7, "learning_rate": 1.4856697211932152e-06, "loss": 2.0766, "step": 19000 }, { "epoch": 3.74, "eval_exact_match": 23.0, "eval_f1": 46.661523832404924, "step": 19200 }, { "epoch": 3.8, "learning_rate": 9.982452719828426e-07, "loss": 2.0837, "step": 19500 }, { "epoch": 3.8, "eval_exact_match": 23.2, "eval_f1": 46.75773872861982, "step": 19500 }, { "epoch": 3.86, "eval_exact_match": 23.0, "eval_f1": 46.675643262204126, "step": 19800 }, { "epoch": 3.9, "learning_rate": 5.108208227724703e-07, "loss": 2.0645, "step": 20000 }, { "epoch": 3.92, "eval_exact_match": 22.8, "eval_f1": 46.09897714824699, "step": 20100 }, { "epoch": 3.98, "eval_exact_match": 22.8, "eval_f1": 46.31081325560941, "step": 20400 }, { "epoch": 4.0, "learning_rate": 2.437122246051862e-08, "loss": 2.05, "step": 20500 }, { "epoch": 4.0, "step": 20516, "total_flos": 5015301739248672.0, "train_loss": 2.4078880540807246, "train_runtime": 515.9797, "train_samples_per_second": 636.079, "train_steps_per_second": 39.761 } ], "max_steps": 20516, "num_train_epochs": 4, "total_flos": 5015301739248672.0, "trial_name": null, "trial_params": null }