| { | |
| "best_metric": 0.258684903383255, | |
| "best_model_checkpoint": "/scratch/skscla001/speech/results/mms-1b-bemgen-combined-model/checkpoint-2200", | |
| "epoch": 1.2893243940175347, | |
| "eval_steps": 100, | |
| "global_step": 2500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05157297576070139, | |
| "grad_norm": 3.681674003601074, | |
| "learning_rate": 0.00028799999999999995, | |
| "loss": 6.7553, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05157297576070139, | |
| "eval_loss": 0.8774181604385376, | |
| "eval_runtime": 57.8286, | |
| "eval_samples_per_second": 16.877, | |
| "eval_steps_per_second": 2.11, | |
| "eval_wer": 0.847627340008707, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.10314595152140278, | |
| "grad_norm": 2.3965742588043213, | |
| "learning_rate": 0.00029950404684002066, | |
| "loss": 0.5648, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.10314595152140278, | |
| "eval_loss": 0.34078654646873474, | |
| "eval_runtime": 57.7359, | |
| "eval_samples_per_second": 16.905, | |
| "eval_steps_per_second": 2.113, | |
| "eval_wer": 0.5031562908141054, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.15471892728210418, | |
| "grad_norm": 12.905903816223145, | |
| "learning_rate": 0.00029898742896504216, | |
| "loss": 0.4827, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.15471892728210418, | |
| "eval_loss": 0.3261478543281555, | |
| "eval_runtime": 57.6285, | |
| "eval_samples_per_second": 16.936, | |
| "eval_steps_per_second": 2.117, | |
| "eval_wer": 0.4930343926861123, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.20629190304280556, | |
| "grad_norm": 1.9552323818206787, | |
| "learning_rate": 0.00029847081109006366, | |
| "loss": 0.4321, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.20629190304280556, | |
| "eval_loss": 0.303559273481369, | |
| "eval_runtime": 57.7126, | |
| "eval_samples_per_second": 16.911, | |
| "eval_steps_per_second": 2.114, | |
| "eval_wer": 0.48541575968654765, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.25786487880350695, | |
| "grad_norm": 2.4878430366516113, | |
| "learning_rate": 0.0002979541932150852, | |
| "loss": 0.4168, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.25786487880350695, | |
| "eval_loss": 0.2988507151603699, | |
| "eval_runtime": 57.9159, | |
| "eval_samples_per_second": 16.852, | |
| "eval_steps_per_second": 2.107, | |
| "eval_wer": 0.4783413147583805, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.30943785456420836, | |
| "grad_norm": 3.2817885875701904, | |
| "learning_rate": 0.0002974375753401067, | |
| "loss": 0.3965, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.30943785456420836, | |
| "eval_loss": 0.2906985282897949, | |
| "eval_runtime": 57.5951, | |
| "eval_samples_per_second": 16.946, | |
| "eval_steps_per_second": 2.118, | |
| "eval_wer": 0.45134958641706574, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.36101083032490977, | |
| "grad_norm": 1.416237235069275, | |
| "learning_rate": 0.0002969209574651283, | |
| "loss": 0.4199, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.36101083032490977, | |
| "eval_loss": 0.29259440302848816, | |
| "eval_runtime": 57.5412, | |
| "eval_samples_per_second": 16.962, | |
| "eval_steps_per_second": 2.12, | |
| "eval_wer": 0.4718110579016108, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.4125838060856111, | |
| "grad_norm": 2.3475992679595947, | |
| "learning_rate": 0.0002964043395901498, | |
| "loss": 0.3975, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.4125838060856111, | |
| "eval_loss": 0.288595974445343, | |
| "eval_runtime": 57.3013, | |
| "eval_samples_per_second": 17.033, | |
| "eval_steps_per_second": 2.129, | |
| "eval_wer": 0.44590770570309096, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.46415678184631254, | |
| "grad_norm": 1.5925949811935425, | |
| "learning_rate": 0.00029588772171517134, | |
| "loss": 0.3839, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.46415678184631254, | |
| "eval_loss": 0.29075172543525696, | |
| "eval_runtime": 57.8589, | |
| "eval_samples_per_second": 16.869, | |
| "eval_steps_per_second": 2.109, | |
| "eval_wer": 0.4722464083587288, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.5157297576070139, | |
| "grad_norm": 2.456458568572998, | |
| "learning_rate": 0.00029537110384019285, | |
| "loss": 0.3673, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5157297576070139, | |
| "eval_loss": 0.2836114764213562, | |
| "eval_runtime": 57.7048, | |
| "eval_samples_per_second": 16.914, | |
| "eval_steps_per_second": 2.114, | |
| "eval_wer": 0.44449281671745755, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5673027333677153, | |
| "grad_norm": 2.249444007873535, | |
| "learning_rate": 0.00029485448596521435, | |
| "loss": 0.3777, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.5673027333677153, | |
| "eval_loss": 0.27841705083847046, | |
| "eval_runtime": 57.4013, | |
| "eval_samples_per_second": 17.003, | |
| "eval_steps_per_second": 2.125, | |
| "eval_wer": 0.4365476708750544, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.6188757091284167, | |
| "grad_norm": 1.6211791038513184, | |
| "learning_rate": 0.0002943378680902359, | |
| "loss": 0.3764, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.6188757091284167, | |
| "eval_loss": 0.2790738344192505, | |
| "eval_runtime": 57.3594, | |
| "eval_samples_per_second": 17.016, | |
| "eval_steps_per_second": 2.127, | |
| "eval_wer": 0.4278406617326948, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.6704486848891181, | |
| "grad_norm": 2.047067165374756, | |
| "learning_rate": 0.0002938212502152574, | |
| "loss": 0.3918, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.6704486848891181, | |
| "eval_loss": 0.27573078870773315, | |
| "eval_runtime": 57.7956, | |
| "eval_samples_per_second": 16.887, | |
| "eval_steps_per_second": 2.111, | |
| "eval_wer": 0.42511972137570747, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.7220216606498195, | |
| "grad_norm": 1.4619171619415283, | |
| "learning_rate": 0.00029330979851902874, | |
| "loss": 0.3669, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.7220216606498195, | |
| "eval_loss": 0.2721499502658844, | |
| "eval_runtime": 57.7153, | |
| "eval_samples_per_second": 16.911, | |
| "eval_steps_per_second": 2.114, | |
| "eval_wer": 0.41815411406181974, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.7735946364105208, | |
| "grad_norm": 1.6276806592941284, | |
| "learning_rate": 0.0002927931806440503, | |
| "loss": 0.377, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.7735946364105208, | |
| "eval_loss": 0.27284711599349976, | |
| "eval_runtime": 57.3961, | |
| "eval_samples_per_second": 17.005, | |
| "eval_steps_per_second": 2.126, | |
| "eval_wer": 0.47572921201567264, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.8251676121712223, | |
| "grad_norm": 3.681490659713745, | |
| "learning_rate": 0.0002922765627690718, | |
| "loss": 0.4174, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.8251676121712223, | |
| "eval_loss": 0.2684435546398163, | |
| "eval_runtime": 57.3852, | |
| "eval_samples_per_second": 17.008, | |
| "eval_steps_per_second": 2.126, | |
| "eval_wer": 0.4242490204614715, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.8767405879319237, | |
| "grad_norm": 1.2958589792251587, | |
| "learning_rate": 0.0002917599448940933, | |
| "loss": 0.3641, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.8767405879319237, | |
| "eval_loss": 0.2649287283420563, | |
| "eval_runtime": 57.8551, | |
| "eval_samples_per_second": 16.87, | |
| "eval_steps_per_second": 2.109, | |
| "eval_wer": 0.4194601654331737, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.9283135636926251, | |
| "grad_norm": 1.8431603908538818, | |
| "learning_rate": 0.00029124332701911486, | |
| "loss": 0.3882, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.9283135636926251, | |
| "eval_loss": 0.2646693289279938, | |
| "eval_runtime": 57.7373, | |
| "eval_samples_per_second": 16.904, | |
| "eval_steps_per_second": 2.113, | |
| "eval_wer": 0.41249455811928604, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.9798865394533265, | |
| "grad_norm": 2.4119603633880615, | |
| "learning_rate": 0.00029072670914413636, | |
| "loss": 0.3861, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.9798865394533265, | |
| "eval_loss": 0.2668125033378601, | |
| "eval_runtime": 57.5111, | |
| "eval_samples_per_second": 16.971, | |
| "eval_steps_per_second": 2.121, | |
| "eval_wer": 0.44253373966042664, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.0314595152140278, | |
| "grad_norm": 1.174663782119751, | |
| "learning_rate": 0.00029021009126915787, | |
| "loss": 0.3647, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.0314595152140278, | |
| "eval_loss": 0.26750749349594116, | |
| "eval_runtime": 57.5411, | |
| "eval_samples_per_second": 16.962, | |
| "eval_steps_per_second": 2.12, | |
| "eval_wer": 0.42457553330430997, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.0830324909747293, | |
| "grad_norm": 0.9301519393920898, | |
| "learning_rate": 0.0002896934733941794, | |
| "loss": 0.3467, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.0830324909747293, | |
| "eval_loss": 0.2629115581512451, | |
| "eval_runtime": 57.7541, | |
| "eval_samples_per_second": 16.899, | |
| "eval_steps_per_second": 2.112, | |
| "eval_wer": 0.40977361776229865, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.1346054667354306, | |
| "grad_norm": 0.7093687057495117, | |
| "learning_rate": 0.00028917685551920093, | |
| "loss": 0.3579, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.1346054667354306, | |
| "eval_loss": 0.258684903383255, | |
| "eval_runtime": 57.7529, | |
| "eval_samples_per_second": 16.9, | |
| "eval_steps_per_second": 2.112, | |
| "eval_wer": 0.41858946451893775, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.1861784424961321, | |
| "grad_norm": 1.027114748954773, | |
| "learning_rate": 0.00028866023764422243, | |
| "loss": 0.3544, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.1861784424961321, | |
| "eval_loss": 0.2609032392501831, | |
| "eval_runtime": 57.5179, | |
| "eval_samples_per_second": 16.969, | |
| "eval_steps_per_second": 2.121, | |
| "eval_wer": 0.412712233347845, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.2377514182568334, | |
| "grad_norm": 0.9472237825393677, | |
| "learning_rate": 0.000288143619769244, | |
| "loss": 0.35, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.2377514182568334, | |
| "eval_loss": 0.2592073678970337, | |
| "eval_runtime": 57.1938, | |
| "eval_samples_per_second": 17.065, | |
| "eval_steps_per_second": 2.133, | |
| "eval_wer": 0.4061819764910753, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.2893243940175347, | |
| "grad_norm": 1.6566214561462402, | |
| "learning_rate": 0.00028762700189426555, | |
| "loss": 0.3519, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.2893243940175347, | |
| "eval_loss": 0.2590978443622589, | |
| "eval_runtime": 57.823, | |
| "eval_samples_per_second": 16.879, | |
| "eval_steps_per_second": 2.11, | |
| "eval_wer": 0.4134740966478015, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.2893243940175347, | |
| "step": 2500, | |
| "total_flos": 5.958212987780215e+18, | |
| "train_loss": 0.6475233551025391, | |
| "train_runtime": 3633.5111, | |
| "train_samples_per_second": 64.037, | |
| "train_steps_per_second": 16.009 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 58170, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 30, | |
| "save_steps": 400, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 3, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 2 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.958212987780215e+18, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |