{ "best_metric": 96.6660279747078, "best_model_checkpoint": "./whisper-small-ha/checkpoint-3000", "epoch": 19.10828025477707, "eval_steps": 500, "global_step": 3000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.1592356687898089, "grad_norm": 1.5306243896484375, "learning_rate": 5e-06, "loss": 0.0034, "step": 25 }, { "epoch": 0.3184713375796178, "grad_norm": 0.7897353172302246, "learning_rate": 1e-05, "loss": 0.0053, "step": 50 }, { "epoch": 0.47770700636942676, "grad_norm": 0.6488688588142395, "learning_rate": 1e-05, "loss": 0.0075, "step": 75 }, { "epoch": 0.6369426751592356, "grad_norm": 5.7697882652282715, "learning_rate": 1e-05, "loss": 0.0117, "step": 100 }, { "epoch": 0.7961783439490446, "grad_norm": 3.655069351196289, "learning_rate": 1e-05, "loss": 0.0112, "step": 125 }, { "epoch": 0.9554140127388535, "grad_norm": 0.6484639644622803, "learning_rate": 1e-05, "loss": 0.0119, "step": 150 }, { "epoch": 1.1146496815286624, "grad_norm": 1.6433720588684082, "learning_rate": 1e-05, "loss": 0.0078, "step": 175 }, { "epoch": 1.2738853503184713, "grad_norm": 0.5470134019851685, "learning_rate": 1e-05, "loss": 0.0111, "step": 200 }, { "epoch": 1.4331210191082802, "grad_norm": 7.9058685302734375, "learning_rate": 1e-05, "loss": 0.0105, "step": 225 }, { "epoch": 1.5923566878980893, "grad_norm": 0.44242119789123535, "learning_rate": 1e-05, "loss": 0.0083, "step": 250 }, { "epoch": 1.7515923566878981, "grad_norm": 0.560337483882904, "learning_rate": 1e-05, "loss": 0.0093, "step": 275 }, { "epoch": 1.910828025477707, "grad_norm": 0.45471709966659546, "learning_rate": 1e-05, "loss": 0.0065, "step": 300 }, { "epoch": 2.070063694267516, "grad_norm": 0.8609573841094971, "learning_rate": 1e-05, "loss": 0.0071, "step": 325 }, { "epoch": 2.229299363057325, "grad_norm": 0.27363401651382446, "learning_rate": 1e-05, "loss": 0.0044, "step": 350 }, { "epoch": 2.388535031847134, "grad_norm": 0.8000548481941223, "learning_rate": 1e-05, "loss": 0.0057, "step": 375 }, { "epoch": 2.5477707006369426, "grad_norm": 13.804436683654785, "learning_rate": 1e-05, "loss": 0.0067, "step": 400 }, { "epoch": 2.7070063694267517, "grad_norm": 0.20495767891407013, "learning_rate": 1e-05, "loss": 0.0064, "step": 425 }, { "epoch": 2.8662420382165603, "grad_norm": 1.7401798963546753, "learning_rate": 1e-05, "loss": 0.0097, "step": 450 }, { "epoch": 3.0254777070063694, "grad_norm": 2.01271653175354, "learning_rate": 1e-05, "loss": 0.0067, "step": 475 }, { "epoch": 3.1847133757961785, "grad_norm": 0.6848011016845703, "learning_rate": 1e-05, "loss": 0.0081, "step": 500 }, { "epoch": 3.1847133757961785, "eval_loss": 2.595545768737793, "eval_runtime": 162.5603, "eval_samples_per_second": 4.06, "eval_steps_per_second": 0.258, "eval_wer": 100.24908986395862, "eval_wer_ortho": 101.58203125, "step": 500 }, { "epoch": 3.343949044585987, "grad_norm": 2.8130505084991455, "learning_rate": 1e-05, "loss": 0.0062, "step": 525 }, { "epoch": 3.5031847133757963, "grad_norm": 2.8670215606689453, "learning_rate": 1e-05, "loss": 0.0049, "step": 550 }, { "epoch": 3.662420382165605, "grad_norm": 0.6676667928695679, "learning_rate": 1e-05, "loss": 0.0052, "step": 575 }, { "epoch": 3.821656050955414, "grad_norm": 0.7465265393257141, "learning_rate": 1e-05, "loss": 0.0033, "step": 600 }, { "epoch": 3.980891719745223, "grad_norm": 0.9514827132225037, "learning_rate": 1e-05, "loss": 0.0037, "step": 625 }, { "epoch": 4.140127388535032, "grad_norm": 0.12443900853395462, "learning_rate": 1e-05, "loss": 0.0024, "step": 650 }, { "epoch": 4.2993630573248405, "grad_norm": 0.19353769719600677, "learning_rate": 1e-05, "loss": 0.0034, "step": 675 }, { "epoch": 4.45859872611465, "grad_norm": 3.0788159370422363, "learning_rate": 1e-05, "loss": 0.0051, "step": 700 }, { "epoch": 4.617834394904459, "grad_norm": 0.6144561171531677, "learning_rate": 1e-05, "loss": 0.003, "step": 725 }, { "epoch": 4.777070063694268, "grad_norm": 0.4218046963214874, "learning_rate": 1e-05, "loss": 0.0046, "step": 750 }, { "epoch": 4.936305732484076, "grad_norm": 13.958199501037598, "learning_rate": 1e-05, "loss": 0.0052, "step": 775 }, { "epoch": 5.095541401273885, "grad_norm": 5.515628337860107, "learning_rate": 1e-05, "loss": 0.005, "step": 800 }, { "epoch": 5.254777070063694, "grad_norm": 0.19730611145496368, "learning_rate": 1e-05, "loss": 0.0025, "step": 825 }, { "epoch": 5.414012738853503, "grad_norm": 0.5651189684867859, "learning_rate": 1e-05, "loss": 0.0084, "step": 850 }, { "epoch": 5.573248407643312, "grad_norm": 0.14254024624824524, "learning_rate": 1e-05, "loss": 0.0045, "step": 875 }, { "epoch": 5.732484076433121, "grad_norm": 0.863792359828949, "learning_rate": 1e-05, "loss": 0.0046, "step": 900 }, { "epoch": 5.89171974522293, "grad_norm": 1.5313643217086792, "learning_rate": 1e-05, "loss": 0.0036, "step": 925 }, { "epoch": 6.050955414012739, "grad_norm": 0.1208188533782959, "learning_rate": 1e-05, "loss": 0.0041, "step": 950 }, { "epoch": 6.210191082802548, "grad_norm": 0.26082348823547363, "learning_rate": 1e-05, "loss": 0.0036, "step": 975 }, { "epoch": 6.369426751592357, "grad_norm": 0.07662803679704666, "learning_rate": 1e-05, "loss": 0.0019, "step": 1000 }, { "epoch": 6.369426751592357, "eval_loss": 2.7293362617492676, "eval_runtime": 161.9823, "eval_samples_per_second": 4.075, "eval_steps_per_second": 0.259, "eval_wer": 98.2563709522897, "eval_wer_ortho": 99.53125, "step": 1000 }, { "epoch": 6.528662420382165, "grad_norm": 0.20079679787158966, "learning_rate": 1e-05, "loss": 0.0019, "step": 1025 }, { "epoch": 6.687898089171974, "grad_norm": 0.5479994416236877, "learning_rate": 1e-05, "loss": 0.0029, "step": 1050 }, { "epoch": 6.8471337579617835, "grad_norm": 2.995877504348755, "learning_rate": 1e-05, "loss": 0.0026, "step": 1075 }, { "epoch": 7.006369426751593, "grad_norm": 0.1092665046453476, "learning_rate": 1e-05, "loss": 0.0042, "step": 1100 }, { "epoch": 7.165605095541402, "grad_norm": 0.23875755071640015, "learning_rate": 1e-05, "loss": 0.0016, "step": 1125 }, { "epoch": 7.32484076433121, "grad_norm": 0.13069689273834229, "learning_rate": 1e-05, "loss": 0.0014, "step": 1150 }, { "epoch": 7.484076433121019, "grad_norm": 0.5974874496459961, "learning_rate": 1e-05, "loss": 0.0026, "step": 1175 }, { "epoch": 7.643312101910828, "grad_norm": 0.2607993185520172, "learning_rate": 1e-05, "loss": 0.0027, "step": 1200 }, { "epoch": 7.802547770700637, "grad_norm": 0.7289318442344666, "learning_rate": 1e-05, "loss": 0.0032, "step": 1225 }, { "epoch": 7.961783439490446, "grad_norm": 0.5595282316207886, "learning_rate": 1e-05, "loss": 0.003, "step": 1250 }, { "epoch": 8.121019108280255, "grad_norm": 1.0708633661270142, "learning_rate": 1e-05, "loss": 0.0066, "step": 1275 }, { "epoch": 8.280254777070065, "grad_norm": 0.12123806029558182, "learning_rate": 1e-05, "loss": 0.003, "step": 1300 }, { "epoch": 8.439490445859873, "grad_norm": 6.941540718078613, "learning_rate": 1e-05, "loss": 0.0042, "step": 1325 }, { "epoch": 8.598726114649681, "grad_norm": 8.278300285339355, "learning_rate": 1e-05, "loss": 0.0067, "step": 1350 }, { "epoch": 8.757961783439491, "grad_norm": 0.314845472574234, "learning_rate": 1e-05, "loss": 0.004, "step": 1375 }, { "epoch": 8.9171974522293, "grad_norm": 0.5092040300369263, "learning_rate": 1e-05, "loss": 0.0034, "step": 1400 }, { "epoch": 9.07643312101911, "grad_norm": 3.4508585929870605, "learning_rate": 1e-05, "loss": 0.0033, "step": 1425 }, { "epoch": 9.235668789808917, "grad_norm": 0.20290407538414001, "learning_rate": 1e-05, "loss": 0.0029, "step": 1450 }, { "epoch": 9.394904458598726, "grad_norm": 0.2386186718940735, "learning_rate": 1e-05, "loss": 0.0044, "step": 1475 }, { "epoch": 9.554140127388536, "grad_norm": 7.686614036560059, "learning_rate": 1e-05, "loss": 0.0025, "step": 1500 }, { "epoch": 9.554140127388536, "eval_loss": 2.748263120651245, "eval_runtime": 164.2494, "eval_samples_per_second": 4.018, "eval_steps_per_second": 0.256, "eval_wer": 100.3257328990228, "eval_wer_ortho": 101.953125, "step": 1500 }, { "epoch": 9.713375796178344, "grad_norm": 4.924851417541504, "learning_rate": 1e-05, "loss": 0.0038, "step": 1525 }, { "epoch": 9.872611464968152, "grad_norm": 0.2480151355266571, "learning_rate": 1e-05, "loss": 0.0063, "step": 1550 }, { "epoch": 10.031847133757962, "grad_norm": 0.3033640384674072, "learning_rate": 1e-05, "loss": 0.0028, "step": 1575 }, { "epoch": 10.19108280254777, "grad_norm": 5.167762756347656, "learning_rate": 1e-05, "loss": 0.0055, "step": 1600 }, { "epoch": 10.35031847133758, "grad_norm": 3.810326099395752, "learning_rate": 1e-05, "loss": 0.0086, "step": 1625 }, { "epoch": 10.509554140127388, "grad_norm": 1.039038062095642, "learning_rate": 1e-05, "loss": 0.0029, "step": 1650 }, { "epoch": 10.668789808917197, "grad_norm": 0.6598480939865112, "learning_rate": 1e-05, "loss": 0.0026, "step": 1675 }, { "epoch": 10.828025477707007, "grad_norm": 0.169340118765831, "learning_rate": 1e-05, "loss": 0.0042, "step": 1700 }, { "epoch": 10.987261146496815, "grad_norm": 0.8864091634750366, "learning_rate": 1e-05, "loss": 0.0024, "step": 1725 }, { "epoch": 11.146496815286625, "grad_norm": 5.09591817855835, "learning_rate": 1e-05, "loss": 0.0035, "step": 1750 }, { "epoch": 11.305732484076433, "grad_norm": 0.39931872487068176, "learning_rate": 1e-05, "loss": 0.0042, "step": 1775 }, { "epoch": 11.464968152866241, "grad_norm": 0.2659781873226166, "learning_rate": 1e-05, "loss": 0.0025, "step": 1800 }, { "epoch": 11.624203821656051, "grad_norm": 0.09195584058761597, "learning_rate": 1e-05, "loss": 0.0041, "step": 1825 }, { "epoch": 11.78343949044586, "grad_norm": 4.084526062011719, "learning_rate": 1e-05, "loss": 0.0035, "step": 1850 }, { "epoch": 11.94267515923567, "grad_norm": 0.3604657053947449, "learning_rate": 1e-05, "loss": 0.0051, "step": 1875 }, { "epoch": 12.101910828025478, "grad_norm": 2.5682971477508545, "learning_rate": 1e-05, "loss": 0.0062, "step": 1900 }, { "epoch": 12.261146496815286, "grad_norm": 0.22079519927501678, "learning_rate": 1e-05, "loss": 0.0047, "step": 1925 }, { "epoch": 12.420382165605096, "grad_norm": 2.0606675148010254, "learning_rate": 1e-05, "loss": 0.0016, "step": 1950 }, { "epoch": 12.579617834394904, "grad_norm": 0.8950979709625244, "learning_rate": 1e-05, "loss": 0.005, "step": 1975 }, { "epoch": 12.738853503184714, "grad_norm": 0.9102997183799744, "learning_rate": 1e-05, "loss": 0.0021, "step": 2000 }, { "epoch": 12.738853503184714, "eval_loss": 2.8374807834625244, "eval_runtime": 168.5918, "eval_samples_per_second": 3.915, "eval_steps_per_second": 0.249, "eval_wer": 97.16420770262502, "eval_wer_ortho": 99.0234375, "step": 2000 }, { "epoch": 12.898089171974522, "grad_norm": 0.8730576634407043, "learning_rate": 1e-05, "loss": 0.0042, "step": 2025 }, { "epoch": 13.05732484076433, "grad_norm": 0.2256990671157837, "learning_rate": 1e-05, "loss": 0.0032, "step": 2050 }, { "epoch": 13.21656050955414, "grad_norm": 0.15921200811862946, "learning_rate": 1e-05, "loss": 0.0027, "step": 2075 }, { "epoch": 13.375796178343949, "grad_norm": 0.06783568859100342, "learning_rate": 1e-05, "loss": 0.0026, "step": 2100 }, { "epoch": 13.535031847133759, "grad_norm": 7.764928817749023, "learning_rate": 1e-05, "loss": 0.0026, "step": 2125 }, { "epoch": 13.694267515923567, "grad_norm": 2.082003116607666, "learning_rate": 1e-05, "loss": 0.0049, "step": 2150 }, { "epoch": 13.853503184713375, "grad_norm": 0.10517149418592453, "learning_rate": 1e-05, "loss": 0.0034, "step": 2175 }, { "epoch": 14.012738853503185, "grad_norm": 0.05206575617194176, "learning_rate": 1e-05, "loss": 0.0022, "step": 2200 }, { "epoch": 14.171974522292993, "grad_norm": 0.4108193814754486, "learning_rate": 1e-05, "loss": 0.0012, "step": 2225 }, { "epoch": 14.331210191082803, "grad_norm": 0.15452294051647186, "learning_rate": 1e-05, "loss": 0.0023, "step": 2250 }, { "epoch": 14.490445859872612, "grad_norm": 0.8691730499267578, "learning_rate": 1e-05, "loss": 0.0033, "step": 2275 }, { "epoch": 14.64968152866242, "grad_norm": 0.13633207976818085, "learning_rate": 1e-05, "loss": 0.003, "step": 2300 }, { "epoch": 14.80891719745223, "grad_norm": 0.18140809237957, "learning_rate": 1e-05, "loss": 0.0033, "step": 2325 }, { "epoch": 14.968152866242038, "grad_norm": 0.180851548910141, "learning_rate": 1e-05, "loss": 0.0011, "step": 2350 }, { "epoch": 15.127388535031848, "grad_norm": 0.21481122076511383, "learning_rate": 1e-05, "loss": 0.0028, "step": 2375 }, { "epoch": 15.286624203821656, "grad_norm": 0.12347543984651566, "learning_rate": 1e-05, "loss": 0.0023, "step": 2400 }, { "epoch": 15.445859872611464, "grad_norm": 0.17947587370872498, "learning_rate": 1e-05, "loss": 0.0018, "step": 2425 }, { "epoch": 15.605095541401274, "grad_norm": 3.2563202381134033, "learning_rate": 1e-05, "loss": 0.0019, "step": 2450 }, { "epoch": 15.764331210191083, "grad_norm": 1.245087742805481, "learning_rate": 1e-05, "loss": 0.0021, "step": 2475 }, { "epoch": 15.923566878980893, "grad_norm": 0.23747095465660095, "learning_rate": 1e-05, "loss": 0.0044, "step": 2500 }, { "epoch": 15.923566878980893, "eval_loss": 2.879415512084961, "eval_runtime": 165.4139, "eval_samples_per_second": 3.99, "eval_steps_per_second": 0.254, "eval_wer": 99.38685571948649, "eval_wer_ortho": 100.95703125, "step": 2500 }, { "epoch": 16.0828025477707, "grad_norm": 0.09223847836256027, "learning_rate": 1e-05, "loss": 0.0021, "step": 2525 }, { "epoch": 16.24203821656051, "grad_norm": 0.11110839247703552, "learning_rate": 1e-05, "loss": 0.0021, "step": 2550 }, { "epoch": 16.401273885350317, "grad_norm": 0.8871801495552063, "learning_rate": 1e-05, "loss": 0.0017, "step": 2575 }, { "epoch": 16.56050955414013, "grad_norm": 4.709665775299072, "learning_rate": 1e-05, "loss": 0.0026, "step": 2600 }, { "epoch": 16.719745222929937, "grad_norm": 0.05977766588330269, "learning_rate": 1e-05, "loss": 0.0023, "step": 2625 }, { "epoch": 16.878980891719745, "grad_norm": 3.9311728477478027, "learning_rate": 1e-05, "loss": 0.0024, "step": 2650 }, { "epoch": 17.038216560509554, "grad_norm": 0.7477012872695923, "learning_rate": 1e-05, "loss": 0.0046, "step": 2675 }, { "epoch": 17.197452229299362, "grad_norm": 0.2451999932527542, "learning_rate": 1e-05, "loss": 0.0019, "step": 2700 }, { "epoch": 17.356687898089174, "grad_norm": 6.275203704833984, "learning_rate": 1e-05, "loss": 0.0025, "step": 2725 }, { "epoch": 17.515923566878982, "grad_norm": 0.22062252461910248, "learning_rate": 1e-05, "loss": 0.0031, "step": 2750 }, { "epoch": 17.67515923566879, "grad_norm": 0.08334895223379135, "learning_rate": 1e-05, "loss": 0.0014, "step": 2775 }, { "epoch": 17.8343949044586, "grad_norm": 2.0435831546783447, "learning_rate": 1e-05, "loss": 0.0013, "step": 2800 }, { "epoch": 17.993630573248407, "grad_norm": 0.09641116112470627, "learning_rate": 1e-05, "loss": 0.0028, "step": 2825 }, { "epoch": 18.15286624203822, "grad_norm": 0.36049848794937134, "learning_rate": 1e-05, "loss": 0.0009, "step": 2850 }, { "epoch": 18.312101910828027, "grad_norm": 1.4279415607452393, "learning_rate": 1e-05, "loss": 0.0038, "step": 2875 }, { "epoch": 18.471337579617835, "grad_norm": 0.22950223088264465, "learning_rate": 1e-05, "loss": 0.004, "step": 2900 }, { "epoch": 18.630573248407643, "grad_norm": 1.3637959957122803, "learning_rate": 1e-05, "loss": 0.0065, "step": 2925 }, { "epoch": 18.78980891719745, "grad_norm": 0.16725656390190125, "learning_rate": 1e-05, "loss": 0.0025, "step": 2950 }, { "epoch": 18.94904458598726, "grad_norm": 0.5503889322280884, "learning_rate": 1e-05, "loss": 0.0027, "step": 2975 }, { "epoch": 19.10828025477707, "grad_norm": 0.10171041637659073, "learning_rate": 1e-05, "loss": 0.0058, "step": 3000 }, { "epoch": 19.10828025477707, "eval_loss": 2.8631205558776855, "eval_runtime": 165.0208, "eval_samples_per_second": 3.999, "eval_steps_per_second": 0.255, "eval_wer": 96.6660279747078, "eval_wer_ortho": 98.5546875, "step": 3000 } ], "logging_steps": 25, "max_steps": 3611, "num_input_tokens_seen": 0, "num_train_epochs": 23, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 3.10835045597184e+18, "train_batch_size": 16, "trial_name": null, "trial_params": null }