| { | |
| "best_metric": 0.22928521037101746, | |
| "best_model_checkpoint": "./checkpoint-4000", | |
| "epoch": 70.17543859649123, | |
| "global_step": 4000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.7851749999999994e-05, | |
| "loss": 9.5756, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 3.517674999999999e-05, | |
| "loss": 4.1914, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 5.250174999999999e-05, | |
| "loss": 3.3316, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 6.982674999999999e-05, | |
| "loss": 3.1471, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "eval_cer": 1.0, | |
| "eval_loss": 3.1599209308624268, | |
| "eval_runtime": 8.4621, | |
| "eval_samples_per_second": 39.588, | |
| "eval_steps_per_second": 0.709, | |
| "eval_wer": 1.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 8.77, | |
| "learning_rate": 7e-05, | |
| "loss": 3.0957, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 10.53, | |
| "learning_rate": 7e-05, | |
| "loss": 2.8518, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 12.28, | |
| "learning_rate": 7e-05, | |
| "loss": 2.3182, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 7e-05, | |
| "loss": 1.8691, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "eval_cer": 0.1685887708649469, | |
| "eval_loss": 0.767376184463501, | |
| "eval_runtime": 8.3651, | |
| "eval_samples_per_second": 40.048, | |
| "eval_steps_per_second": 0.717, | |
| "eval_wer": 0.7361436377829821, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 15.79, | |
| "learning_rate": 7e-05, | |
| "loss": 1.6182, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 17.54, | |
| "learning_rate": 7e-05, | |
| "loss": 1.4813, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 19.3, | |
| "learning_rate": 7e-05, | |
| "loss": 1.4096, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 21.05, | |
| "learning_rate": 7e-05, | |
| "loss": 1.3227, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 21.05, | |
| "eval_cer": 0.10065756196256954, | |
| "eval_loss": 0.38490724563598633, | |
| "eval_runtime": 8.5046, | |
| "eval_samples_per_second": 39.39, | |
| "eval_steps_per_second": 0.705, | |
| "eval_wer": 0.5335675253708041, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 22.81, | |
| "learning_rate": 7e-05, | |
| "loss": 1.2765, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 24.56, | |
| "learning_rate": 7e-05, | |
| "loss": 1.2296, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 26.32, | |
| "learning_rate": 7e-05, | |
| "loss": 1.1942, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 28.07, | |
| "learning_rate": 7e-05, | |
| "loss": 1.163, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 28.07, | |
| "eval_cer": 0.08229640870005059, | |
| "eval_loss": 0.30146270990371704, | |
| "eval_runtime": 8.4606, | |
| "eval_samples_per_second": 39.595, | |
| "eval_steps_per_second": 0.709, | |
| "eval_wer": 0.4558938329430133, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 29.82, | |
| "learning_rate": 7e-05, | |
| "loss": 1.1351, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 31.58, | |
| "learning_rate": 7e-05, | |
| "loss": 1.1201, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 33.33, | |
| "learning_rate": 7e-05, | |
| "loss": 1.1014, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 35.09, | |
| "learning_rate": 7e-05, | |
| "loss": 1.0768, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 35.09, | |
| "eval_cer": 0.07278705108750633, | |
| "eval_loss": 0.27209824323654175, | |
| "eval_runtime": 8.2339, | |
| "eval_samples_per_second": 40.685, | |
| "eval_steps_per_second": 0.729, | |
| "eval_wer": 0.4032006245120999, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 36.84, | |
| "learning_rate": 6.670824999999999e-05, | |
| "loss": 1.063, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 38.6, | |
| "learning_rate": 6.338324999999999e-05, | |
| "loss": 1.0538, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 40.35, | |
| "learning_rate": 6.005824999999999e-05, | |
| "loss": 1.0439, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 42.11, | |
| "learning_rate": 5.673325e-05, | |
| "loss": 1.0224, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 42.11, | |
| "eval_cer": 0.06914516944865959, | |
| "eval_loss": 0.2586216628551483, | |
| "eval_runtime": 8.3142, | |
| "eval_samples_per_second": 40.292, | |
| "eval_steps_per_second": 0.722, | |
| "eval_wer": 0.3825136612021858, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 43.86, | |
| "learning_rate": 5.340824999999999e-05, | |
| "loss": 1.0116, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 45.61, | |
| "learning_rate": 5.008325e-05, | |
| "loss": 0.9925, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 47.37, | |
| "learning_rate": 4.675824999999999e-05, | |
| "loss": 0.9877, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 49.12, | |
| "learning_rate": 4.343324999999999e-05, | |
| "loss": 0.9817, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 49.12, | |
| "eval_cer": 0.06525037936267071, | |
| "eval_loss": 0.24575529992580414, | |
| "eval_runtime": 8.5545, | |
| "eval_samples_per_second": 39.16, | |
| "eval_steps_per_second": 0.701, | |
| "eval_wer": 0.36533957845433257, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 50.88, | |
| "learning_rate": 4.010825e-05, | |
| "loss": 0.9781, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 52.63, | |
| "learning_rate": 3.6816499999999996e-05, | |
| "loss": 0.9625, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 54.39, | |
| "learning_rate": 3.34915e-05, | |
| "loss": 0.9499, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 56.14, | |
| "learning_rate": 3.01665e-05, | |
| "loss": 0.941, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 56.14, | |
| "eval_cer": 0.06054628224582701, | |
| "eval_loss": 0.23060913383960724, | |
| "eval_runtime": 8.3111, | |
| "eval_samples_per_second": 40.307, | |
| "eval_steps_per_second": 0.722, | |
| "eval_wer": 0.33879781420765026, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 57.89, | |
| "learning_rate": 2.6841499999999994e-05, | |
| "loss": 0.9375, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 59.65, | |
| "learning_rate": 2.3516499999999998e-05, | |
| "loss": 0.9297, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 61.4, | |
| "learning_rate": 2.0191499999999998e-05, | |
| "loss": 0.9218, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 63.16, | |
| "learning_rate": 1.68665e-05, | |
| "loss": 0.9235, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 63.16, | |
| "eval_cer": 0.06150733434496712, | |
| "eval_loss": 0.23152786493301392, | |
| "eval_runtime": 8.3332, | |
| "eval_samples_per_second": 40.201, | |
| "eval_steps_per_second": 0.72, | |
| "eval_wer": 0.33801717408274784, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 64.91, | |
| "learning_rate": 1.3541499999999999e-05, | |
| "loss": 0.9173, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 66.67, | |
| "learning_rate": 1.0216500000000001e-05, | |
| "loss": 0.9136, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 68.42, | |
| "learning_rate": 6.891500000000002e-06, | |
| "loss": 0.9107, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 70.18, | |
| "learning_rate": 3.566500000000003e-06, | |
| "loss": 0.9141, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 70.18, | |
| "eval_cer": 0.06019221041982802, | |
| "eval_loss": 0.22928521037101746, | |
| "eval_runtime": 8.619, | |
| "eval_samples_per_second": 38.868, | |
| "eval_steps_per_second": 0.696, | |
| "eval_wer": 0.3333333333333333, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 70.18, | |
| "step": 4000, | |
| "total_flos": 1.1026564721970925e+20, | |
| "train_loss": 1.6216355571746826, | |
| "train_runtime": 16397.7621, | |
| "train_samples_per_second": 31.224, | |
| "train_steps_per_second": 0.244 | |
| } | |
| ], | |
| "max_steps": 4000, | |
| "num_train_epochs": 71, | |
| "total_flos": 1.1026564721970925e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |