diff --git a/checkpoint-1000/config.json b/checkpoint-1000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..337642e18341c758364c32c6b6159a090615329b --- /dev/null +++ b/checkpoint-1000/config.json @@ -0,0 +1,47 @@ +{ + "activation_dropout": 0.0, + "activation_function": "gelu", + "apply_spec_augment": false, + "architectures": [ + "WhisperForConditionalGeneration" + ], + "attention_dropout": 0.0, + "begin_suppress_tokens": null, + "bos_token_id": 50257, + "classifier_proj_size": 256, + "d_model": 768, + "decoder_attention_heads": 12, + "decoder_ffn_dim": 3072, + "decoder_layerdrop": 0.0, + "decoder_layers": 12, + "decoder_start_token_id": 50258, + "dropout": 0.0, + "encoder_attention_heads": 12, + "encoder_ffn_dim": 3072, + "encoder_layerdrop": 0.0, + "encoder_layers": 12, + "eos_token_id": 50257, + "forced_decoder_ids": null, + "init_std": 0.02, + "is_encoder_decoder": true, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "max_length": null, + "max_source_positions": 1500, + "max_target_positions": 448, + "median_filter_width": 7, + "model_type": "whisper", + "num_hidden_layers": 12, + "num_mel_bins": 80, + "pad_token_id": 50257, + "scale_embedding": false, + "torch_dtype": "float32", + "transformers_version": "4.51.3", + "use_cache": false, + "use_weighted_layer_sum": false, + "vocab_size": 51865 +} diff --git a/checkpoint-1000/generation_config.json b/checkpoint-1000/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..e1f4120825a303544ca23708d9dc52bbaa1dce16 --- /dev/null +++ b/checkpoint-1000/generation_config.json @@ -0,0 +1,9 @@ +{ + "bos_token_id": 50257, + "decoder_start_token_id": 50258, + "eos_token_id": 50257, + "pad_token_id": 50257, + "suppress_tokens": [], + "transformers_version": "4.51.3", + "use_cache": false +} diff --git a/checkpoint-1000/model.safetensors b/checkpoint-1000/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ace031620c744f631ac5277f4d6223a21cb2b214 --- /dev/null +++ b/checkpoint-1000/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd44d89d0147eedf670e3f3dba96f1821950c516fd7f57b976543e0256c67f1c +size 966995080 diff --git a/checkpoint-1000/optimizer.pt b/checkpoint-1000/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..934cd591805aa2483275e3ddb9d301812bdbec09 --- /dev/null +++ b/checkpoint-1000/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e446455b26d8c3490f570cceb19fd1d68ec10cd8adda69fb8c9716facc3d454c +size 1925064044 diff --git a/checkpoint-1000/preprocessor_config.json b/checkpoint-1000/preprocessor_config.json new file mode 100644 index 0000000000000000000000000000000000000000..d7c5b1dfd466c815664b16139e8f605620d22e4e --- /dev/null +++ b/checkpoint-1000/preprocessor_config.json @@ -0,0 +1,15 @@ +{ + "chunk_length": 30, + "dither": 0.0, + "feature_extractor_type": "WhisperFeatureExtractor", + "feature_size": 80, + "hop_length": 160, + "n_fft": 400, + "n_samples": 480000, + "nb_max_frames": 3000, + "padding_side": "right", + "padding_value": 0.0, + "processor_class": "WhisperProcessor", + "return_attention_mask": false, + "sampling_rate": 16000 +} diff --git a/checkpoint-1000/rng_state.pth b/checkpoint-1000/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..17484613e1929f873bb7b720fe16342c295bc23b --- /dev/null +++ b/checkpoint-1000/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d4f95eca64235b7f040ceb7c0bbe77a9a52bb931b9122581a1cf284ad8efef39 +size 14244 diff --git a/checkpoint-1000/scaler.pt b/checkpoint-1000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..b20c7ce1edd2d6d447db6c7d2fa54c9412144136 --- /dev/null +++ b/checkpoint-1000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b20682d68cd70b8b0b208c3bd5f62edfdf7322c345d5f8fb0fc9912581d14875 +size 988 diff --git a/checkpoint-1000/scheduler.pt b/checkpoint-1000/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..bebcd674f987ebfd03eeb7b42059f3bb5bbfc0b8 --- /dev/null +++ b/checkpoint-1000/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:746d3afa0d117ef9f30a06dd95b2e34c08a35c2d0208de96d411b73e96ff1923 +size 1064 diff --git a/checkpoint-1000/trainer_state.json b/checkpoint-1000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..7794b80d429a7255f33c3a306da7192b2347486a --- /dev/null +++ b/checkpoint-1000/trainer_state.json @@ -0,0 +1,402 @@ +{ + "best_global_step": 1000, + "best_metric": 14.071299035983584, + "best_model_checkpoint": "./SALAMA_C3/checkpoint-1000", + "epoch": 1.675041876046901, + "eval_steps": 500, + "global_step": 1000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.03350083752093802, + "grad_norm": 11.757207870483398, + "learning_rate": 6.333333333333334e-07, + "loss": 0.5112, + "step": 20 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 10.197953224182129, + "learning_rate": 1.3e-06, + "loss": 0.5283, + "step": 40 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 12.811074256896973, + "learning_rate": 1.9666666666666668e-06, + "loss": 0.474, + "step": 60 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 8.559342384338379, + "learning_rate": 2.6e-06, + "loss": 0.424, + "step": 80 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 6.9629225730896, + "learning_rate": 3.266666666666667e-06, + "loss": 0.4465, + "step": 100 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 7.833058834075928, + "learning_rate": 3.9333333333333335e-06, + "loss": 0.3764, + "step": 120 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 6.882424831390381, + "learning_rate": 4.600000000000001e-06, + "loss": 0.3705, + "step": 140 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 5.4183244705200195, + "learning_rate": 5.2666666666666665e-06, + "loss": 0.4116, + "step": 160 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 6.103787899017334, + "learning_rate": 5.933333333333335e-06, + "loss": 0.3903, + "step": 180 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 6.084160804748535, + "learning_rate": 6.600000000000001e-06, + "loss": 0.3794, + "step": 200 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 4.531406879425049, + "learning_rate": 7.266666666666668e-06, + "loss": 0.3904, + "step": 220 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 6.702854156494141, + "learning_rate": 7.933333333333334e-06, + "loss": 0.3672, + "step": 240 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 6.386377811431885, + "learning_rate": 8.6e-06, + "loss": 0.3379, + "step": 260 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 5.33281135559082, + "learning_rate": 9.266666666666667e-06, + "loss": 0.4102, + "step": 280 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 6.031435012817383, + "learning_rate": 9.933333333333334e-06, + "loss": 0.3815, + "step": 300 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 4.8638176918029785, + "learning_rate": 9.968253968253969e-06, + "loss": 0.358, + "step": 320 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 5.352113723754883, + "learning_rate": 9.932980599647268e-06, + "loss": 0.3578, + "step": 340 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 5.560739040374756, + "learning_rate": 9.897707231040565e-06, + "loss": 0.3795, + "step": 360 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 6.186940670013428, + "learning_rate": 9.862433862433864e-06, + "loss": 0.3852, + "step": 380 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 5.37507438659668, + "learning_rate": 9.827160493827161e-06, + "loss": 0.3761, + "step": 400 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 6.373449802398682, + "learning_rate": 9.79188712522046e-06, + "loss": 0.3674, + "step": 420 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 5.702625274658203, + "learning_rate": 9.756613756613757e-06, + "loss": 0.4013, + "step": 440 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 5.5675153732299805, + "learning_rate": 9.721340388007056e-06, + "loss": 0.3495, + "step": 460 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 6.296374320983887, + "learning_rate": 9.686067019400353e-06, + "loss": 0.3704, + "step": 480 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 4.830463886260986, + "learning_rate": 9.650793650793652e-06, + "loss": 0.36, + "step": 500 + }, + { + "epoch": 0.8375209380234506, + "eval_loss": 0.27764827013015747, + "eval_runtime": 1752.3826, + "eval_samples_per_second": 2.725, + "eval_steps_per_second": 0.341, + "eval_wer": 19.59291781998664, + "step": 500 + }, + { + "epoch": 0.8710217755443886, + "grad_norm": 4.863058090209961, + "learning_rate": 9.61552028218695e-06, + "loss": 0.3874, + "step": 520 + }, + { + "epoch": 0.9045226130653267, + "grad_norm": 6.121027946472168, + "learning_rate": 9.580246913580248e-06, + "loss": 0.382, + "step": 540 + }, + { + "epoch": 0.9380234505862647, + "grad_norm": 4.538082599639893, + "learning_rate": 9.544973544973546e-06, + "loss": 0.3611, + "step": 560 + }, + { + "epoch": 0.9715242881072027, + "grad_norm": 4.8772382736206055, + "learning_rate": 9.509700176366844e-06, + "loss": 0.3584, + "step": 580 + }, + { + "epoch": 1.0050251256281406, + "grad_norm": 4.039212703704834, + "learning_rate": 9.474426807760142e-06, + "loss": 0.3701, + "step": 600 + }, + { + "epoch": 1.0385259631490786, + "grad_norm": 4.787687301635742, + "learning_rate": 9.43915343915344e-06, + "loss": 0.2058, + "step": 620 + }, + { + "epoch": 1.0720268006700167, + "grad_norm": 4.503021717071533, + "learning_rate": 9.403880070546738e-06, + "loss": 0.2455, + "step": 640 + }, + { + "epoch": 1.1055276381909547, + "grad_norm": 6.663857936859131, + "learning_rate": 9.368606701940036e-06, + "loss": 0.2179, + "step": 660 + }, + { + "epoch": 1.1390284757118927, + "grad_norm": 4.486196041107178, + "learning_rate": 9.333333333333334e-06, + "loss": 0.2196, + "step": 680 + }, + { + "epoch": 1.1725293132328307, + "grad_norm": 4.667060852050781, + "learning_rate": 9.298059964726633e-06, + "loss": 0.2183, + "step": 700 + }, + { + "epoch": 1.2060301507537687, + "grad_norm": 5.608316898345947, + "learning_rate": 9.26278659611993e-06, + "loss": 0.2161, + "step": 720 + }, + { + "epoch": 1.2395309882747068, + "grad_norm": 4.2184271812438965, + "learning_rate": 9.227513227513229e-06, + "loss": 0.2382, + "step": 740 + }, + { + "epoch": 1.2730318257956448, + "grad_norm": 3.871945381164551, + "learning_rate": 9.192239858906526e-06, + "loss": 0.2214, + "step": 760 + }, + { + "epoch": 1.3065326633165828, + "grad_norm": 3.730222225189209, + "learning_rate": 9.156966490299825e-06, + "loss": 0.2213, + "step": 780 + }, + { + "epoch": 1.3400335008375208, + "grad_norm": 4.740777015686035, + "learning_rate": 9.121693121693122e-06, + "loss": 0.2503, + "step": 800 + }, + { + "epoch": 1.3735343383584588, + "grad_norm": 4.123469352722168, + "learning_rate": 9.086419753086421e-06, + "loss": 0.2443, + "step": 820 + }, + { + "epoch": 1.4070351758793969, + "grad_norm": 3.988917827606201, + "learning_rate": 9.051146384479718e-06, + "loss": 0.241, + "step": 840 + }, + { + "epoch": 1.4405360134003349, + "grad_norm": 5.4730305671691895, + "learning_rate": 9.015873015873017e-06, + "loss": 0.2163, + "step": 860 + }, + { + "epoch": 1.474036850921273, + "grad_norm": 5.020652770996094, + "learning_rate": 8.980599647266314e-06, + "loss": 0.2387, + "step": 880 + }, + { + "epoch": 1.507537688442211, + "grad_norm": 4.231486797332764, + "learning_rate": 8.945326278659613e-06, + "loss": 0.2523, + "step": 900 + }, + { + "epoch": 1.541038525963149, + "grad_norm": 6.197975158691406, + "learning_rate": 8.910052910052912e-06, + "loss": 0.2255, + "step": 920 + }, + { + "epoch": 1.574539363484087, + "grad_norm": 5.489210605621338, + "learning_rate": 8.874779541446209e-06, + "loss": 0.2397, + "step": 940 + }, + { + "epoch": 1.608040201005025, + "grad_norm": 4.32737398147583, + "learning_rate": 8.839506172839508e-06, + "loss": 0.2298, + "step": 960 + }, + { + "epoch": 1.641541038525963, + "grad_norm": 4.504214763641357, + "learning_rate": 8.804232804232805e-06, + "loss": 0.2301, + "step": 980 + }, + { + "epoch": 1.675041876046901, + "grad_norm": 3.9694759845733643, + "learning_rate": 8.768959435626104e-06, + "loss": 0.2338, + "step": 1000 + }, + { + "epoch": 1.675041876046901, + "eval_loss": 0.18838582932949066, + "eval_runtime": 1770.3449, + "eval_samples_per_second": 2.698, + "eval_steps_per_second": 0.337, + "eval_wer": 14.071299035983584, + "step": 1000 + } + ], + "logging_steps": 20, + "max_steps": 5970, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "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": 9.23386705403904e+18, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-1000/training_args.bin b/checkpoint-1000/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6ffec31ca17b32691fb4cd1278f73919f3259cf4 --- /dev/null +++ b/checkpoint-1000/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41a3c645624f7463029286b7d5e1ad6924cfc44a0d153880431b77c40e5deb73 +size 5432 diff --git a/checkpoint-1500/config.json b/checkpoint-1500/config.json new file mode 100644 index 0000000000000000000000000000000000000000..337642e18341c758364c32c6b6159a090615329b --- /dev/null +++ b/checkpoint-1500/config.json @@ -0,0 +1,47 @@ +{ + "activation_dropout": 0.0, + "activation_function": "gelu", + "apply_spec_augment": false, + "architectures": [ + "WhisperForConditionalGeneration" + ], + "attention_dropout": 0.0, + "begin_suppress_tokens": null, + "bos_token_id": 50257, + "classifier_proj_size": 256, + "d_model": 768, + "decoder_attention_heads": 12, + "decoder_ffn_dim": 3072, + "decoder_layerdrop": 0.0, + "decoder_layers": 12, + "decoder_start_token_id": 50258, + "dropout": 0.0, + "encoder_attention_heads": 12, + "encoder_ffn_dim": 3072, + "encoder_layerdrop": 0.0, + "encoder_layers": 12, + "eos_token_id": 50257, + "forced_decoder_ids": null, + "init_std": 0.02, + "is_encoder_decoder": true, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "max_length": null, + "max_source_positions": 1500, + "max_target_positions": 448, + "median_filter_width": 7, + "model_type": "whisper", + "num_hidden_layers": 12, + "num_mel_bins": 80, + "pad_token_id": 50257, + "scale_embedding": false, + "torch_dtype": "float32", + "transformers_version": "4.51.3", + "use_cache": false, + "use_weighted_layer_sum": false, + "vocab_size": 51865 +} diff --git a/checkpoint-1500/generation_config.json b/checkpoint-1500/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..e1f4120825a303544ca23708d9dc52bbaa1dce16 --- /dev/null +++ b/checkpoint-1500/generation_config.json @@ -0,0 +1,9 @@ +{ + "bos_token_id": 50257, + "decoder_start_token_id": 50258, + "eos_token_id": 50257, + "pad_token_id": 50257, + "suppress_tokens": [], + "transformers_version": "4.51.3", + "use_cache": false +} diff --git a/checkpoint-1500/model.safetensors b/checkpoint-1500/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..4627192ae19d9f1c4f6b42510b7d8bb7a30a5fa3 --- /dev/null +++ b/checkpoint-1500/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1383039729a87311da55da8166aab4f70de2077cf353120cbe4019867d195421 +size 966995080 diff --git a/checkpoint-1500/optimizer.pt b/checkpoint-1500/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..a897d8085a5cfe2d0ea60d1a4e6d0d1a5fad78bd --- /dev/null +++ b/checkpoint-1500/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:59853d5c8ad3a215d900e6fe789b56aa9301c83f2f20ecda92d4504fea9a52e7 +size 1925064044 diff --git a/checkpoint-1500/preprocessor_config.json b/checkpoint-1500/preprocessor_config.json new file mode 100644 index 0000000000000000000000000000000000000000..d7c5b1dfd466c815664b16139e8f605620d22e4e --- /dev/null +++ b/checkpoint-1500/preprocessor_config.json @@ -0,0 +1,15 @@ +{ + "chunk_length": 30, + "dither": 0.0, + "feature_extractor_type": "WhisperFeatureExtractor", + "feature_size": 80, + "hop_length": 160, + "n_fft": 400, + "n_samples": 480000, + "nb_max_frames": 3000, + "padding_side": "right", + "padding_value": 0.0, + "processor_class": "WhisperProcessor", + "return_attention_mask": false, + "sampling_rate": 16000 +} diff --git a/checkpoint-1500/rng_state.pth b/checkpoint-1500/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..bde50de1a420e33e65e7867e5b9b1773046de316 --- /dev/null +++ b/checkpoint-1500/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:641ea6afeeb91ac119db140ccc4b56a307b61b309a6e3509792e0c955cb04c57 +size 14244 diff --git a/checkpoint-1500/scaler.pt b/checkpoint-1500/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..29cb42d020711c35dc0b3cf3b90118ada25de4b6 --- /dev/null +++ b/checkpoint-1500/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c4a0fb91661a47fbe0e37f999cf89e4b80985f1c000b17f3f08e9e22356182db +size 988 diff --git a/checkpoint-1500/scheduler.pt b/checkpoint-1500/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..904a8cfd781801ca1ace9cb9c65ba0a8335b885a --- /dev/null +++ b/checkpoint-1500/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c9fa73076154129d09f32d7071b6e9a88bc719841a440c20606e35875e46afcd +size 1064 diff --git a/checkpoint-1500/trainer_state.json b/checkpoint-1500/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..0a9c16fd5db4f0c950cee1c47244328865d1ec87 --- /dev/null +++ b/checkpoint-1500/trainer_state.json @@ -0,0 +1,586 @@ +{ + "best_global_step": 1500, + "best_metric": 11.713753937195762, + "best_model_checkpoint": "./SALAMA_C3/checkpoint-1500", + "epoch": 2.5125628140703515, + "eval_steps": 500, + "global_step": 1500, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.03350083752093802, + "grad_norm": 11.757207870483398, + "learning_rate": 6.333333333333334e-07, + "loss": 0.5112, + "step": 20 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 10.197953224182129, + "learning_rate": 1.3e-06, + "loss": 0.5283, + "step": 40 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 12.811074256896973, + "learning_rate": 1.9666666666666668e-06, + "loss": 0.474, + "step": 60 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 8.559342384338379, + "learning_rate": 2.6e-06, + "loss": 0.424, + "step": 80 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 6.9629225730896, + "learning_rate": 3.266666666666667e-06, + "loss": 0.4465, + "step": 100 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 7.833058834075928, + "learning_rate": 3.9333333333333335e-06, + "loss": 0.3764, + "step": 120 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 6.882424831390381, + "learning_rate": 4.600000000000001e-06, + "loss": 0.3705, + "step": 140 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 5.4183244705200195, + "learning_rate": 5.2666666666666665e-06, + "loss": 0.4116, + "step": 160 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 6.103787899017334, + "learning_rate": 5.933333333333335e-06, + "loss": 0.3903, + "step": 180 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 6.084160804748535, + "learning_rate": 6.600000000000001e-06, + "loss": 0.3794, + "step": 200 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 4.531406879425049, + "learning_rate": 7.266666666666668e-06, + "loss": 0.3904, + "step": 220 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 6.702854156494141, + "learning_rate": 7.933333333333334e-06, + "loss": 0.3672, + "step": 240 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 6.386377811431885, + "learning_rate": 8.6e-06, + "loss": 0.3379, + "step": 260 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 5.33281135559082, + "learning_rate": 9.266666666666667e-06, + "loss": 0.4102, + "step": 280 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 6.031435012817383, + "learning_rate": 9.933333333333334e-06, + "loss": 0.3815, + "step": 300 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 4.8638176918029785, + "learning_rate": 9.968253968253969e-06, + "loss": 0.358, + "step": 320 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 5.352113723754883, + "learning_rate": 9.932980599647268e-06, + "loss": 0.3578, + "step": 340 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 5.560739040374756, + "learning_rate": 9.897707231040565e-06, + "loss": 0.3795, + "step": 360 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 6.186940670013428, + "learning_rate": 9.862433862433864e-06, + "loss": 0.3852, + "step": 380 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 5.37507438659668, + "learning_rate": 9.827160493827161e-06, + "loss": 0.3761, + "step": 400 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 6.373449802398682, + "learning_rate": 9.79188712522046e-06, + "loss": 0.3674, + "step": 420 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 5.702625274658203, + "learning_rate": 9.756613756613757e-06, + "loss": 0.4013, + "step": 440 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 5.5675153732299805, + "learning_rate": 9.721340388007056e-06, + "loss": 0.3495, + "step": 460 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 6.296374320983887, + "learning_rate": 9.686067019400353e-06, + "loss": 0.3704, + "step": 480 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 4.830463886260986, + "learning_rate": 9.650793650793652e-06, + "loss": 0.36, + "step": 500 + }, + { + "epoch": 0.8375209380234506, + "eval_loss": 0.27764827013015747, + "eval_runtime": 1752.3826, + "eval_samples_per_second": 2.725, + "eval_steps_per_second": 0.341, + "eval_wer": 19.59291781998664, + "step": 500 + }, + { + "epoch": 0.8710217755443886, + "grad_norm": 4.863058090209961, + "learning_rate": 9.61552028218695e-06, + "loss": 0.3874, + "step": 520 + }, + { + "epoch": 0.9045226130653267, + "grad_norm": 6.121027946472168, + "learning_rate": 9.580246913580248e-06, + "loss": 0.382, + "step": 540 + }, + { + "epoch": 0.9380234505862647, + "grad_norm": 4.538082599639893, + "learning_rate": 9.544973544973546e-06, + "loss": 0.3611, + "step": 560 + }, + { + "epoch": 0.9715242881072027, + "grad_norm": 4.8772382736206055, + "learning_rate": 9.509700176366844e-06, + "loss": 0.3584, + "step": 580 + }, + { + "epoch": 1.0050251256281406, + "grad_norm": 4.039212703704834, + "learning_rate": 9.474426807760142e-06, + "loss": 0.3701, + "step": 600 + }, + { + "epoch": 1.0385259631490786, + "grad_norm": 4.787687301635742, + "learning_rate": 9.43915343915344e-06, + "loss": 0.2058, + "step": 620 + }, + { + "epoch": 1.0720268006700167, + "grad_norm": 4.503021717071533, + "learning_rate": 9.403880070546738e-06, + "loss": 0.2455, + "step": 640 + }, + { + "epoch": 1.1055276381909547, + "grad_norm": 6.663857936859131, + "learning_rate": 9.368606701940036e-06, + "loss": 0.2179, + "step": 660 + }, + { + "epoch": 1.1390284757118927, + "grad_norm": 4.486196041107178, + "learning_rate": 9.333333333333334e-06, + "loss": 0.2196, + "step": 680 + }, + { + "epoch": 1.1725293132328307, + "grad_norm": 4.667060852050781, + "learning_rate": 9.298059964726633e-06, + "loss": 0.2183, + "step": 700 + }, + { + "epoch": 1.2060301507537687, + "grad_norm": 5.608316898345947, + "learning_rate": 9.26278659611993e-06, + "loss": 0.2161, + "step": 720 + }, + { + "epoch": 1.2395309882747068, + "grad_norm": 4.2184271812438965, + "learning_rate": 9.227513227513229e-06, + "loss": 0.2382, + "step": 740 + }, + { + "epoch": 1.2730318257956448, + "grad_norm": 3.871945381164551, + "learning_rate": 9.192239858906526e-06, + "loss": 0.2214, + "step": 760 + }, + { + "epoch": 1.3065326633165828, + "grad_norm": 3.730222225189209, + "learning_rate": 9.156966490299825e-06, + "loss": 0.2213, + "step": 780 + }, + { + "epoch": 1.3400335008375208, + "grad_norm": 4.740777015686035, + "learning_rate": 9.121693121693122e-06, + "loss": 0.2503, + "step": 800 + }, + { + "epoch": 1.3735343383584588, + "grad_norm": 4.123469352722168, + "learning_rate": 9.086419753086421e-06, + "loss": 0.2443, + "step": 820 + }, + { + "epoch": 1.4070351758793969, + "grad_norm": 3.988917827606201, + "learning_rate": 9.051146384479718e-06, + "loss": 0.241, + "step": 840 + }, + { + "epoch": 1.4405360134003349, + "grad_norm": 5.4730305671691895, + "learning_rate": 9.015873015873017e-06, + "loss": 0.2163, + "step": 860 + }, + { + "epoch": 1.474036850921273, + "grad_norm": 5.020652770996094, + "learning_rate": 8.980599647266314e-06, + "loss": 0.2387, + "step": 880 + }, + { + "epoch": 1.507537688442211, + "grad_norm": 4.231486797332764, + "learning_rate": 8.945326278659613e-06, + "loss": 0.2523, + "step": 900 + }, + { + "epoch": 1.541038525963149, + "grad_norm": 6.197975158691406, + "learning_rate": 8.910052910052912e-06, + "loss": 0.2255, + "step": 920 + }, + { + "epoch": 1.574539363484087, + "grad_norm": 5.489210605621338, + "learning_rate": 8.874779541446209e-06, + "loss": 0.2397, + "step": 940 + }, + { + "epoch": 1.608040201005025, + "grad_norm": 4.32737398147583, + "learning_rate": 8.839506172839508e-06, + "loss": 0.2298, + "step": 960 + }, + { + "epoch": 1.641541038525963, + "grad_norm": 4.504214763641357, + "learning_rate": 8.804232804232805e-06, + "loss": 0.2301, + "step": 980 + }, + { + "epoch": 1.675041876046901, + "grad_norm": 3.9694759845733643, + "learning_rate": 8.768959435626104e-06, + "loss": 0.2338, + "step": 1000 + }, + { + "epoch": 1.675041876046901, + "eval_loss": 0.18838582932949066, + "eval_runtime": 1770.3449, + "eval_samples_per_second": 2.698, + "eval_steps_per_second": 0.337, + "eval_wer": 14.071299035983584, + "step": 1000 + }, + { + "epoch": 1.708542713567839, + "grad_norm": 4.972934722900391, + "learning_rate": 8.7336860670194e-06, + "loss": 0.2388, + "step": 1020 + }, + { + "epoch": 1.742043551088777, + "grad_norm": 5.178994655609131, + "learning_rate": 8.6984126984127e-06, + "loss": 0.2531, + "step": 1040 + }, + { + "epoch": 1.775544388609715, + "grad_norm": 3.560372829437256, + "learning_rate": 8.663139329805997e-06, + "loss": 0.2187, + "step": 1060 + }, + { + "epoch": 1.809045226130653, + "grad_norm": 4.647324562072754, + "learning_rate": 8.627865961199296e-06, + "loss": 0.2413, + "step": 1080 + }, + { + "epoch": 1.8425460636515911, + "grad_norm": 3.6784422397613525, + "learning_rate": 8.592592592592593e-06, + "loss": 0.2242, + "step": 1100 + }, + { + "epoch": 1.8760469011725294, + "grad_norm": 5.928018093109131, + "learning_rate": 8.557319223985891e-06, + "loss": 0.2406, + "step": 1120 + }, + { + "epoch": 1.9095477386934674, + "grad_norm": 6.358487606048584, + "learning_rate": 8.52204585537919e-06, + "loss": 0.2482, + "step": 1140 + }, + { + "epoch": 1.9430485762144054, + "grad_norm": 4.409506797790527, + "learning_rate": 8.486772486772487e-06, + "loss": 0.2415, + "step": 1160 + }, + { + "epoch": 1.9765494137353434, + "grad_norm": 5.169639587402344, + "learning_rate": 8.451499118165786e-06, + "loss": 0.222, + "step": 1180 + }, + { + "epoch": 2.0100502512562812, + "grad_norm": 2.8841328620910645, + "learning_rate": 8.416225749559083e-06, + "loss": 0.2109, + "step": 1200 + }, + { + "epoch": 2.0435510887772192, + "grad_norm": 3.835498332977295, + "learning_rate": 8.380952380952382e-06, + "loss": 0.1284, + "step": 1220 + }, + { + "epoch": 2.0770519262981573, + "grad_norm": 2.584859848022461, + "learning_rate": 8.34567901234568e-06, + "loss": 0.1059, + "step": 1240 + }, + { + "epoch": 2.1105527638190953, + "grad_norm": 2.797682285308838, + "learning_rate": 8.310405643738978e-06, + "loss": 0.0967, + "step": 1260 + }, + { + "epoch": 2.1440536013400333, + "grad_norm": 3.2937309741973877, + "learning_rate": 8.275132275132275e-06, + "loss": 0.1267, + "step": 1280 + }, + { + "epoch": 2.1775544388609713, + "grad_norm": 3.344325065612793, + "learning_rate": 8.239858906525574e-06, + "loss": 0.1354, + "step": 1300 + }, + { + "epoch": 2.2110552763819094, + "grad_norm": 2.955522060394287, + "learning_rate": 8.204585537918873e-06, + "loss": 0.104, + "step": 1320 + }, + { + "epoch": 2.2445561139028474, + "grad_norm": 5.180908203125, + "learning_rate": 8.16931216931217e-06, + "loss": 0.1229, + "step": 1340 + }, + { + "epoch": 2.2780569514237854, + "grad_norm": 3.639417886734009, + "learning_rate": 8.134038800705469e-06, + "loss": 0.104, + "step": 1360 + }, + { + "epoch": 2.3115577889447234, + "grad_norm": 3.7380902767181396, + "learning_rate": 8.098765432098766e-06, + "loss": 0.1141, + "step": 1380 + }, + { + "epoch": 2.3450586264656614, + "grad_norm": 4.100318908691406, + "learning_rate": 8.063492063492065e-06, + "loss": 0.1123, + "step": 1400 + }, + { + "epoch": 2.3785594639865995, + "grad_norm": 2.8147048950195312, + "learning_rate": 8.028218694885362e-06, + "loss": 0.1039, + "step": 1420 + }, + { + "epoch": 2.4120603015075375, + "grad_norm": 3.7376410961151123, + "learning_rate": 7.99294532627866e-06, + "loss": 0.1179, + "step": 1440 + }, + { + "epoch": 2.4455611390284755, + "grad_norm": 4.320065975189209, + "learning_rate": 7.957671957671958e-06, + "loss": 0.1167, + "step": 1460 + }, + { + "epoch": 2.4790619765494135, + "grad_norm": 4.415127277374268, + "learning_rate": 7.922398589065257e-06, + "loss": 0.125, + "step": 1480 + }, + { + "epoch": 2.5125628140703515, + "grad_norm": 2.631763219833374, + "learning_rate": 7.887125220458554e-06, + "loss": 0.1188, + "step": 1500 + }, + { + "epoch": 2.5125628140703515, + "eval_loss": 0.13635103404521942, + "eval_runtime": 1806.1099, + "eval_samples_per_second": 2.644, + "eval_steps_per_second": 0.331, + "eval_wer": 11.713753937195762, + "step": 1500 + } + ], + "logging_steps": 20, + "max_steps": 5970, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "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": 1.385036770295808e+19, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-1500/training_args.bin b/checkpoint-1500/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6ffec31ca17b32691fb4cd1278f73919f3259cf4 --- /dev/null +++ b/checkpoint-1500/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41a3c645624f7463029286b7d5e1ad6924cfc44a0d153880431b77c40e5deb73 +size 5432 diff --git a/checkpoint-2000/config.json b/checkpoint-2000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..337642e18341c758364c32c6b6159a090615329b --- /dev/null +++ b/checkpoint-2000/config.json @@ -0,0 +1,47 @@ +{ + "activation_dropout": 0.0, + "activation_function": "gelu", + "apply_spec_augment": false, + "architectures": [ + "WhisperForConditionalGeneration" + ], + "attention_dropout": 0.0, + "begin_suppress_tokens": null, + "bos_token_id": 50257, + "classifier_proj_size": 256, + "d_model": 768, + "decoder_attention_heads": 12, + "decoder_ffn_dim": 3072, + "decoder_layerdrop": 0.0, + "decoder_layers": 12, + "decoder_start_token_id": 50258, + "dropout": 0.0, + "encoder_attention_heads": 12, + "encoder_ffn_dim": 3072, + "encoder_layerdrop": 0.0, + "encoder_layers": 12, + "eos_token_id": 50257, + "forced_decoder_ids": null, + "init_std": 0.02, + "is_encoder_decoder": true, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "max_length": null, + "max_source_positions": 1500, + "max_target_positions": 448, + "median_filter_width": 7, + "model_type": "whisper", + "num_hidden_layers": 12, + "num_mel_bins": 80, + "pad_token_id": 50257, + "scale_embedding": false, + "torch_dtype": "float32", + "transformers_version": "4.51.3", + "use_cache": false, + "use_weighted_layer_sum": false, + "vocab_size": 51865 +} diff --git a/checkpoint-2000/generation_config.json b/checkpoint-2000/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..e1f4120825a303544ca23708d9dc52bbaa1dce16 --- /dev/null +++ b/checkpoint-2000/generation_config.json @@ -0,0 +1,9 @@ +{ + "bos_token_id": 50257, + "decoder_start_token_id": 50258, + "eos_token_id": 50257, + "pad_token_id": 50257, + "suppress_tokens": [], + "transformers_version": "4.51.3", + "use_cache": false +} diff --git a/checkpoint-2000/model.safetensors b/checkpoint-2000/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..82a6ab0b4bc51c5c8c9d454ff4b6b3a9b11dd4b5 --- /dev/null +++ b/checkpoint-2000/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c498ee290ccc95138b06ba4e9df6e40f71bbcd1469a11d559d1fda78500860d6 +size 966995080 diff --git a/checkpoint-2000/optimizer.pt b/checkpoint-2000/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..8b600ad0538018d1e2eb327077a0bcbbb41faa02 --- /dev/null +++ b/checkpoint-2000/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d55d01888dd6a30defbf104a69ca6ce62571bbebec904e0246886e33bea442ef +size 1925064044 diff --git a/checkpoint-2000/preprocessor_config.json b/checkpoint-2000/preprocessor_config.json new file mode 100644 index 0000000000000000000000000000000000000000..d7c5b1dfd466c815664b16139e8f605620d22e4e --- /dev/null +++ b/checkpoint-2000/preprocessor_config.json @@ -0,0 +1,15 @@ +{ + "chunk_length": 30, + "dither": 0.0, + "feature_extractor_type": "WhisperFeatureExtractor", + "feature_size": 80, + "hop_length": 160, + "n_fft": 400, + "n_samples": 480000, + "nb_max_frames": 3000, + "padding_side": "right", + "padding_value": 0.0, + "processor_class": "WhisperProcessor", + "return_attention_mask": false, + "sampling_rate": 16000 +} diff --git a/checkpoint-2000/rng_state.pth b/checkpoint-2000/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..0e4e80dfebd8cd6f1b955cb9deba13c0c8586b61 --- /dev/null +++ b/checkpoint-2000/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:91d9a982d0a102bb658e38ae6685c0a7ee27a2849f974142ce6a350968204948 +size 14244 diff --git a/checkpoint-2000/scaler.pt b/checkpoint-2000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..74e543ea7a0ec66296bd08e55d3216f555fe6846 --- /dev/null +++ b/checkpoint-2000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fd89000c206de186c97dd611bd50d2dd7f764488fa801cf3f469d639f7c1b66f +size 988 diff --git a/checkpoint-2000/scheduler.pt b/checkpoint-2000/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..92fbe43c26d23adf5a9ebda30a727b3290e7a7ab --- /dev/null +++ b/checkpoint-2000/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c19a31a6a8b6605fe25dc22b645addbe28ac65d7d2baa70210b9ac1d12e57193 +size 1064 diff --git a/checkpoint-2000/trainer_state.json b/checkpoint-2000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..339c755e511d04cb56a0f5bbde270a5c6d8d6fde --- /dev/null +++ b/checkpoint-2000/trainer_state.json @@ -0,0 +1,770 @@ +{ + "best_global_step": 2000, + "best_metric": 7.172854824854443, + "best_model_checkpoint": "./SALAMA_C3/checkpoint-2000", + "epoch": 3.3500837520938025, + "eval_steps": 500, + "global_step": 2000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.03350083752093802, + "grad_norm": 11.757207870483398, + "learning_rate": 6.333333333333334e-07, + "loss": 0.5112, + "step": 20 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 10.197953224182129, + "learning_rate": 1.3e-06, + "loss": 0.5283, + "step": 40 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 12.811074256896973, + "learning_rate": 1.9666666666666668e-06, + "loss": 0.474, + "step": 60 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 8.559342384338379, + "learning_rate": 2.6e-06, + "loss": 0.424, + "step": 80 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 6.9629225730896, + "learning_rate": 3.266666666666667e-06, + "loss": 0.4465, + "step": 100 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 7.833058834075928, + "learning_rate": 3.9333333333333335e-06, + "loss": 0.3764, + "step": 120 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 6.882424831390381, + "learning_rate": 4.600000000000001e-06, + "loss": 0.3705, + "step": 140 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 5.4183244705200195, + "learning_rate": 5.2666666666666665e-06, + "loss": 0.4116, + "step": 160 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 6.103787899017334, + "learning_rate": 5.933333333333335e-06, + "loss": 0.3903, + "step": 180 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 6.084160804748535, + "learning_rate": 6.600000000000001e-06, + "loss": 0.3794, + "step": 200 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 4.531406879425049, + "learning_rate": 7.266666666666668e-06, + "loss": 0.3904, + "step": 220 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 6.702854156494141, + "learning_rate": 7.933333333333334e-06, + "loss": 0.3672, + "step": 240 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 6.386377811431885, + "learning_rate": 8.6e-06, + "loss": 0.3379, + "step": 260 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 5.33281135559082, + "learning_rate": 9.266666666666667e-06, + "loss": 0.4102, + "step": 280 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 6.031435012817383, + "learning_rate": 9.933333333333334e-06, + "loss": 0.3815, + "step": 300 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 4.8638176918029785, + "learning_rate": 9.968253968253969e-06, + "loss": 0.358, + "step": 320 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 5.352113723754883, + "learning_rate": 9.932980599647268e-06, + "loss": 0.3578, + "step": 340 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 5.560739040374756, + "learning_rate": 9.897707231040565e-06, + "loss": 0.3795, + "step": 360 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 6.186940670013428, + "learning_rate": 9.862433862433864e-06, + "loss": 0.3852, + "step": 380 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 5.37507438659668, + "learning_rate": 9.827160493827161e-06, + "loss": 0.3761, + "step": 400 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 6.373449802398682, + "learning_rate": 9.79188712522046e-06, + "loss": 0.3674, + "step": 420 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 5.702625274658203, + "learning_rate": 9.756613756613757e-06, + "loss": 0.4013, + "step": 440 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 5.5675153732299805, + "learning_rate": 9.721340388007056e-06, + "loss": 0.3495, + "step": 460 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 6.296374320983887, + "learning_rate": 9.686067019400353e-06, + "loss": 0.3704, + "step": 480 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 4.830463886260986, + "learning_rate": 9.650793650793652e-06, + "loss": 0.36, + "step": 500 + }, + { + "epoch": 0.8375209380234506, + "eval_loss": 0.27764827013015747, + "eval_runtime": 1752.3826, + "eval_samples_per_second": 2.725, + "eval_steps_per_second": 0.341, + "eval_wer": 19.59291781998664, + "step": 500 + }, + { + "epoch": 0.8710217755443886, + "grad_norm": 4.863058090209961, + "learning_rate": 9.61552028218695e-06, + "loss": 0.3874, + "step": 520 + }, + { + "epoch": 0.9045226130653267, + "grad_norm": 6.121027946472168, + "learning_rate": 9.580246913580248e-06, + "loss": 0.382, + "step": 540 + }, + { + "epoch": 0.9380234505862647, + "grad_norm": 4.538082599639893, + "learning_rate": 9.544973544973546e-06, + "loss": 0.3611, + "step": 560 + }, + { + "epoch": 0.9715242881072027, + "grad_norm": 4.8772382736206055, + "learning_rate": 9.509700176366844e-06, + "loss": 0.3584, + "step": 580 + }, + { + "epoch": 1.0050251256281406, + "grad_norm": 4.039212703704834, + "learning_rate": 9.474426807760142e-06, + "loss": 0.3701, + "step": 600 + }, + { + "epoch": 1.0385259631490786, + "grad_norm": 4.787687301635742, + "learning_rate": 9.43915343915344e-06, + "loss": 0.2058, + "step": 620 + }, + { + "epoch": 1.0720268006700167, + "grad_norm": 4.503021717071533, + "learning_rate": 9.403880070546738e-06, + "loss": 0.2455, + "step": 640 + }, + { + "epoch": 1.1055276381909547, + "grad_norm": 6.663857936859131, + "learning_rate": 9.368606701940036e-06, + "loss": 0.2179, + "step": 660 + }, + { + "epoch": 1.1390284757118927, + "grad_norm": 4.486196041107178, + "learning_rate": 9.333333333333334e-06, + "loss": 0.2196, + "step": 680 + }, + { + "epoch": 1.1725293132328307, + "grad_norm": 4.667060852050781, + "learning_rate": 9.298059964726633e-06, + "loss": 0.2183, + "step": 700 + }, + { + "epoch": 1.2060301507537687, + "grad_norm": 5.608316898345947, + "learning_rate": 9.26278659611993e-06, + "loss": 0.2161, + "step": 720 + }, + { + "epoch": 1.2395309882747068, + "grad_norm": 4.2184271812438965, + "learning_rate": 9.227513227513229e-06, + "loss": 0.2382, + "step": 740 + }, + { + "epoch": 1.2730318257956448, + "grad_norm": 3.871945381164551, + "learning_rate": 9.192239858906526e-06, + "loss": 0.2214, + "step": 760 + }, + { + "epoch": 1.3065326633165828, + "grad_norm": 3.730222225189209, + "learning_rate": 9.156966490299825e-06, + "loss": 0.2213, + "step": 780 + }, + { + "epoch": 1.3400335008375208, + "grad_norm": 4.740777015686035, + "learning_rate": 9.121693121693122e-06, + "loss": 0.2503, + "step": 800 + }, + { + "epoch": 1.3735343383584588, + "grad_norm": 4.123469352722168, + "learning_rate": 9.086419753086421e-06, + "loss": 0.2443, + "step": 820 + }, + { + "epoch": 1.4070351758793969, + "grad_norm": 3.988917827606201, + "learning_rate": 9.051146384479718e-06, + "loss": 0.241, + "step": 840 + }, + { + "epoch": 1.4405360134003349, + "grad_norm": 5.4730305671691895, + "learning_rate": 9.015873015873017e-06, + "loss": 0.2163, + "step": 860 + }, + { + "epoch": 1.474036850921273, + "grad_norm": 5.020652770996094, + "learning_rate": 8.980599647266314e-06, + "loss": 0.2387, + "step": 880 + }, + { + "epoch": 1.507537688442211, + "grad_norm": 4.231486797332764, + "learning_rate": 8.945326278659613e-06, + "loss": 0.2523, + "step": 900 + }, + { + "epoch": 1.541038525963149, + "grad_norm": 6.197975158691406, + "learning_rate": 8.910052910052912e-06, + "loss": 0.2255, + "step": 920 + }, + { + "epoch": 1.574539363484087, + "grad_norm": 5.489210605621338, + "learning_rate": 8.874779541446209e-06, + "loss": 0.2397, + "step": 940 + }, + { + "epoch": 1.608040201005025, + "grad_norm": 4.32737398147583, + "learning_rate": 8.839506172839508e-06, + "loss": 0.2298, + "step": 960 + }, + { + "epoch": 1.641541038525963, + "grad_norm": 4.504214763641357, + "learning_rate": 8.804232804232805e-06, + "loss": 0.2301, + "step": 980 + }, + { + "epoch": 1.675041876046901, + "grad_norm": 3.9694759845733643, + "learning_rate": 8.768959435626104e-06, + "loss": 0.2338, + "step": 1000 + }, + { + "epoch": 1.675041876046901, + "eval_loss": 0.18838582932949066, + "eval_runtime": 1770.3449, + "eval_samples_per_second": 2.698, + "eval_steps_per_second": 0.337, + "eval_wer": 14.071299035983584, + "step": 1000 + }, + { + "epoch": 1.708542713567839, + "grad_norm": 4.972934722900391, + "learning_rate": 8.7336860670194e-06, + "loss": 0.2388, + "step": 1020 + }, + { + "epoch": 1.742043551088777, + "grad_norm": 5.178994655609131, + "learning_rate": 8.6984126984127e-06, + "loss": 0.2531, + "step": 1040 + }, + { + "epoch": 1.775544388609715, + "grad_norm": 3.560372829437256, + "learning_rate": 8.663139329805997e-06, + "loss": 0.2187, + "step": 1060 + }, + { + "epoch": 1.809045226130653, + "grad_norm": 4.647324562072754, + "learning_rate": 8.627865961199296e-06, + "loss": 0.2413, + "step": 1080 + }, + { + "epoch": 1.8425460636515911, + "grad_norm": 3.6784422397613525, + "learning_rate": 8.592592592592593e-06, + "loss": 0.2242, + "step": 1100 + }, + { + "epoch": 1.8760469011725294, + "grad_norm": 5.928018093109131, + "learning_rate": 8.557319223985891e-06, + "loss": 0.2406, + "step": 1120 + }, + { + "epoch": 1.9095477386934674, + "grad_norm": 6.358487606048584, + "learning_rate": 8.52204585537919e-06, + "loss": 0.2482, + "step": 1140 + }, + { + "epoch": 1.9430485762144054, + "grad_norm": 4.409506797790527, + "learning_rate": 8.486772486772487e-06, + "loss": 0.2415, + "step": 1160 + }, + { + "epoch": 1.9765494137353434, + "grad_norm": 5.169639587402344, + "learning_rate": 8.451499118165786e-06, + "loss": 0.222, + "step": 1180 + }, + { + "epoch": 2.0100502512562812, + "grad_norm": 2.8841328620910645, + "learning_rate": 8.416225749559083e-06, + "loss": 0.2109, + "step": 1200 + }, + { + "epoch": 2.0435510887772192, + "grad_norm": 3.835498332977295, + "learning_rate": 8.380952380952382e-06, + "loss": 0.1284, + "step": 1220 + }, + { + "epoch": 2.0770519262981573, + "grad_norm": 2.584859848022461, + "learning_rate": 8.34567901234568e-06, + "loss": 0.1059, + "step": 1240 + }, + { + "epoch": 2.1105527638190953, + "grad_norm": 2.797682285308838, + "learning_rate": 8.310405643738978e-06, + "loss": 0.0967, + "step": 1260 + }, + { + "epoch": 2.1440536013400333, + "grad_norm": 3.2937309741973877, + "learning_rate": 8.275132275132275e-06, + "loss": 0.1267, + "step": 1280 + }, + { + "epoch": 2.1775544388609713, + "grad_norm": 3.344325065612793, + "learning_rate": 8.239858906525574e-06, + "loss": 0.1354, + "step": 1300 + }, + { + "epoch": 2.2110552763819094, + "grad_norm": 2.955522060394287, + "learning_rate": 8.204585537918873e-06, + "loss": 0.104, + "step": 1320 + }, + { + "epoch": 2.2445561139028474, + "grad_norm": 5.180908203125, + "learning_rate": 8.16931216931217e-06, + "loss": 0.1229, + "step": 1340 + }, + { + "epoch": 2.2780569514237854, + "grad_norm": 3.639417886734009, + "learning_rate": 8.134038800705469e-06, + "loss": 0.104, + "step": 1360 + }, + { + "epoch": 2.3115577889447234, + "grad_norm": 3.7380902767181396, + "learning_rate": 8.098765432098766e-06, + "loss": 0.1141, + "step": 1380 + }, + { + "epoch": 2.3450586264656614, + "grad_norm": 4.100318908691406, + "learning_rate": 8.063492063492065e-06, + "loss": 0.1123, + "step": 1400 + }, + { + "epoch": 2.3785594639865995, + "grad_norm": 2.8147048950195312, + "learning_rate": 8.028218694885362e-06, + "loss": 0.1039, + "step": 1420 + }, + { + "epoch": 2.4120603015075375, + "grad_norm": 3.7376410961151123, + "learning_rate": 7.99294532627866e-06, + "loss": 0.1179, + "step": 1440 + }, + { + "epoch": 2.4455611390284755, + "grad_norm": 4.320065975189209, + "learning_rate": 7.957671957671958e-06, + "loss": 0.1167, + "step": 1460 + }, + { + "epoch": 2.4790619765494135, + "grad_norm": 4.415127277374268, + "learning_rate": 7.922398589065257e-06, + "loss": 0.125, + "step": 1480 + }, + { + "epoch": 2.5125628140703515, + "grad_norm": 2.631763219833374, + "learning_rate": 7.887125220458554e-06, + "loss": 0.1188, + "step": 1500 + }, + { + "epoch": 2.5125628140703515, + "eval_loss": 0.13635103404521942, + "eval_runtime": 1806.1099, + "eval_samples_per_second": 2.644, + "eval_steps_per_second": 0.331, + "eval_wer": 11.713753937195762, + "step": 1500 + }, + { + "epoch": 2.5460636515912896, + "grad_norm": 3.608773946762085, + "learning_rate": 7.851851851851853e-06, + "loss": 0.1222, + "step": 1520 + }, + { + "epoch": 2.5795644891122276, + "grad_norm": 3.4494569301605225, + "learning_rate": 7.816578483245151e-06, + "loss": 0.1087, + "step": 1540 + }, + { + "epoch": 2.6130653266331656, + "grad_norm": 3.6112048625946045, + "learning_rate": 7.781305114638449e-06, + "loss": 0.1097, + "step": 1560 + }, + { + "epoch": 2.6465661641541036, + "grad_norm": 3.1978707313537598, + "learning_rate": 7.746031746031747e-06, + "loss": 0.1101, + "step": 1580 + }, + { + "epoch": 2.6800670016750416, + "grad_norm": 3.6376733779907227, + "learning_rate": 7.710758377425045e-06, + "loss": 0.1056, + "step": 1600 + }, + { + "epoch": 2.7135678391959797, + "grad_norm": 3.754915237426758, + "learning_rate": 7.675485008818343e-06, + "loss": 0.1103, + "step": 1620 + }, + { + "epoch": 2.7470686767169177, + "grad_norm": 3.329240322113037, + "learning_rate": 7.64021164021164e-06, + "loss": 0.1193, + "step": 1640 + }, + { + "epoch": 2.7805695142378557, + "grad_norm": 3.9300169944763184, + "learning_rate": 7.604938271604939e-06, + "loss": 0.1232, + "step": 1660 + }, + { + "epoch": 2.8140703517587937, + "grad_norm": 4.290626049041748, + "learning_rate": 7.569664902998237e-06, + "loss": 0.1374, + "step": 1680 + }, + { + "epoch": 2.8475711892797317, + "grad_norm": 3.8987998962402344, + "learning_rate": 7.534391534391535e-06, + "loss": 0.1062, + "step": 1700 + }, + { + "epoch": 2.8810720268006698, + "grad_norm": 4.234960079193115, + "learning_rate": 7.499118165784833e-06, + "loss": 0.111, + "step": 1720 + }, + { + "epoch": 2.914572864321608, + "grad_norm": 3.103458881378174, + "learning_rate": 7.463844797178131e-06, + "loss": 0.1076, + "step": 1740 + }, + { + "epoch": 2.948073701842546, + "grad_norm": 3.270204782485962, + "learning_rate": 7.428571428571429e-06, + "loss": 0.1224, + "step": 1760 + }, + { + "epoch": 2.981574539363484, + "grad_norm": 4.261337757110596, + "learning_rate": 7.393298059964727e-06, + "loss": 0.1224, + "step": 1780 + }, + { + "epoch": 3.0150753768844223, + "grad_norm": 1.9531301259994507, + "learning_rate": 7.358024691358025e-06, + "loss": 0.0913, + "step": 1800 + }, + { + "epoch": 3.0485762144053603, + "grad_norm": 1.866215467453003, + "learning_rate": 7.322751322751324e-06, + "loss": 0.0466, + "step": 1820 + }, + { + "epoch": 3.0820770519262983, + "grad_norm": 3.9456610679626465, + "learning_rate": 7.287477954144622e-06, + "loss": 0.0451, + "step": 1840 + }, + { + "epoch": 3.1155778894472363, + "grad_norm": 1.9518849849700928, + "learning_rate": 7.25220458553792e-06, + "loss": 0.0414, + "step": 1860 + }, + { + "epoch": 3.1490787269681744, + "grad_norm": 2.232792854309082, + "learning_rate": 7.216931216931218e-06, + "loss": 0.0502, + "step": 1880 + }, + { + "epoch": 3.1825795644891124, + "grad_norm": 2.074127674102783, + "learning_rate": 7.181657848324516e-06, + "loss": 0.057, + "step": 1900 + }, + { + "epoch": 3.2160804020100504, + "grad_norm": 3.5036733150482178, + "learning_rate": 7.146384479717814e-06, + "loss": 0.0515, + "step": 1920 + }, + { + "epoch": 3.2495812395309884, + "grad_norm": 2.324014186859131, + "learning_rate": 7.111111111111112e-06, + "loss": 0.0454, + "step": 1940 + }, + { + "epoch": 3.2830820770519265, + "grad_norm": 2.710326671600342, + "learning_rate": 7.07583774250441e-06, + "loss": 0.0473, + "step": 1960 + }, + { + "epoch": 3.3165829145728645, + "grad_norm": 2.8943896293640137, + "learning_rate": 7.040564373897708e-06, + "loss": 0.046, + "step": 1980 + }, + { + "epoch": 3.3500837520938025, + "grad_norm": 2.319986581802368, + "learning_rate": 7.005291005291006e-06, + "loss": 0.0449, + "step": 2000 + }, + { + "epoch": 3.3500837520938025, + "eval_loss": 0.10934468358755112, + "eval_runtime": 1762.4506, + "eval_samples_per_second": 2.71, + "eval_steps_per_second": 0.339, + "eval_wer": 7.172854824854443, + "step": 2000 + } + ], + "logging_steps": 20, + "max_steps": 5970, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "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": 1.846686835187712e+19, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-2000/training_args.bin b/checkpoint-2000/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6ffec31ca17b32691fb4cd1278f73919f3259cf4 --- /dev/null +++ b/checkpoint-2000/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41a3c645624f7463029286b7d5e1ad6924cfc44a0d153880431b77c40e5deb73 +size 5432 diff --git a/checkpoint-2500/config.json b/checkpoint-2500/config.json new file mode 100644 index 0000000000000000000000000000000000000000..337642e18341c758364c32c6b6159a090615329b --- /dev/null +++ b/checkpoint-2500/config.json @@ -0,0 +1,47 @@ +{ + "activation_dropout": 0.0, + "activation_function": "gelu", + "apply_spec_augment": false, + "architectures": [ + "WhisperForConditionalGeneration" + ], + "attention_dropout": 0.0, + "begin_suppress_tokens": null, + "bos_token_id": 50257, + "classifier_proj_size": 256, + "d_model": 768, + "decoder_attention_heads": 12, + "decoder_ffn_dim": 3072, + "decoder_layerdrop": 0.0, + "decoder_layers": 12, + "decoder_start_token_id": 50258, + "dropout": 0.0, + "encoder_attention_heads": 12, + "encoder_ffn_dim": 3072, + "encoder_layerdrop": 0.0, + "encoder_layers": 12, + "eos_token_id": 50257, + "forced_decoder_ids": null, + "init_std": 0.02, + "is_encoder_decoder": true, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "max_length": null, + "max_source_positions": 1500, + "max_target_positions": 448, + "median_filter_width": 7, + "model_type": "whisper", + "num_hidden_layers": 12, + "num_mel_bins": 80, + "pad_token_id": 50257, + "scale_embedding": false, + "torch_dtype": "float32", + "transformers_version": "4.51.3", + "use_cache": false, + "use_weighted_layer_sum": false, + "vocab_size": 51865 +} diff --git a/checkpoint-2500/generation_config.json b/checkpoint-2500/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..e1f4120825a303544ca23708d9dc52bbaa1dce16 --- /dev/null +++ b/checkpoint-2500/generation_config.json @@ -0,0 +1,9 @@ +{ + "bos_token_id": 50257, + "decoder_start_token_id": 50258, + "eos_token_id": 50257, + "pad_token_id": 50257, + "suppress_tokens": [], + "transformers_version": "4.51.3", + "use_cache": false +} diff --git a/checkpoint-2500/model.safetensors b/checkpoint-2500/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f04716ed71b6b3a64d9312ec7fb280e6e5c98c3c --- /dev/null +++ b/checkpoint-2500/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:983ad47253d47f33288d03a6dc02f69916cf46fab753a6c1fc58916fd035349a +size 966995080 diff --git a/checkpoint-2500/optimizer.pt b/checkpoint-2500/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..052321f1ceaa0aa664026cdcce526d877944e7d3 --- /dev/null +++ b/checkpoint-2500/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:30d507a0591cd28124a7b5e2ceb3f1b6db701163d043cdf64538d377ff5de8c1 +size 1925064044 diff --git a/checkpoint-2500/preprocessor_config.json b/checkpoint-2500/preprocessor_config.json new file mode 100644 index 0000000000000000000000000000000000000000..d7c5b1dfd466c815664b16139e8f605620d22e4e --- /dev/null +++ b/checkpoint-2500/preprocessor_config.json @@ -0,0 +1,15 @@ +{ + "chunk_length": 30, + "dither": 0.0, + "feature_extractor_type": "WhisperFeatureExtractor", + "feature_size": 80, + "hop_length": 160, + "n_fft": 400, + "n_samples": 480000, + "nb_max_frames": 3000, + "padding_side": "right", + "padding_value": 0.0, + "processor_class": "WhisperProcessor", + "return_attention_mask": false, + "sampling_rate": 16000 +} diff --git a/checkpoint-2500/rng_state.pth b/checkpoint-2500/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..db516a5a46005a07bccbbbba0fe53dc98dbbdb19 --- /dev/null +++ b/checkpoint-2500/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:846458c7462562ffbdb4648906c3b912a1d942419c6cf6acd73bf7bd018aae2c +size 14244 diff --git a/checkpoint-2500/scaler.pt b/checkpoint-2500/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..02419576029802ae0337f5daa5a6dcdb4e8b5f30 --- /dev/null +++ b/checkpoint-2500/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f518493be2847dc24c07ed29ea8a55ff05bdfa05590734e42ae275f087cf2300 +size 988 diff --git a/checkpoint-2500/scheduler.pt b/checkpoint-2500/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..676ad5fb6bdc95de1aed27a24a26d4968aeb4809 --- /dev/null +++ b/checkpoint-2500/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2d4dcde2bb7ed571bf37cb3c2010a2bd3c91906a89169b61a4be9525501656fe +size 1064 diff --git a/checkpoint-2500/trainer_state.json b/checkpoint-2500/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..53d5434bdff15da0a8c53ece433711ca7a8b837e --- /dev/null +++ b/checkpoint-2500/trainer_state.json @@ -0,0 +1,954 @@ +{ + "best_global_step": 2500, + "best_metric": 6.707549871146321, + "best_model_checkpoint": "./SALAMA_C3/checkpoint-2500", + "epoch": 4.187604690117253, + "eval_steps": 500, + "global_step": 2500, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.03350083752093802, + "grad_norm": 11.757207870483398, + "learning_rate": 6.333333333333334e-07, + "loss": 0.5112, + "step": 20 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 10.197953224182129, + "learning_rate": 1.3e-06, + "loss": 0.5283, + "step": 40 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 12.811074256896973, + "learning_rate": 1.9666666666666668e-06, + "loss": 0.474, + "step": 60 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 8.559342384338379, + "learning_rate": 2.6e-06, + "loss": 0.424, + "step": 80 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 6.9629225730896, + "learning_rate": 3.266666666666667e-06, + "loss": 0.4465, + "step": 100 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 7.833058834075928, + "learning_rate": 3.9333333333333335e-06, + "loss": 0.3764, + "step": 120 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 6.882424831390381, + "learning_rate": 4.600000000000001e-06, + "loss": 0.3705, + "step": 140 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 5.4183244705200195, + "learning_rate": 5.2666666666666665e-06, + "loss": 0.4116, + "step": 160 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 6.103787899017334, + "learning_rate": 5.933333333333335e-06, + "loss": 0.3903, + "step": 180 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 6.084160804748535, + "learning_rate": 6.600000000000001e-06, + "loss": 0.3794, + "step": 200 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 4.531406879425049, + "learning_rate": 7.266666666666668e-06, + "loss": 0.3904, + "step": 220 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 6.702854156494141, + "learning_rate": 7.933333333333334e-06, + "loss": 0.3672, + "step": 240 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 6.386377811431885, + "learning_rate": 8.6e-06, + "loss": 0.3379, + "step": 260 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 5.33281135559082, + "learning_rate": 9.266666666666667e-06, + "loss": 0.4102, + "step": 280 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 6.031435012817383, + "learning_rate": 9.933333333333334e-06, + "loss": 0.3815, + "step": 300 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 4.8638176918029785, + "learning_rate": 9.968253968253969e-06, + "loss": 0.358, + "step": 320 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 5.352113723754883, + "learning_rate": 9.932980599647268e-06, + "loss": 0.3578, + "step": 340 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 5.560739040374756, + "learning_rate": 9.897707231040565e-06, + "loss": 0.3795, + "step": 360 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 6.186940670013428, + "learning_rate": 9.862433862433864e-06, + "loss": 0.3852, + "step": 380 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 5.37507438659668, + "learning_rate": 9.827160493827161e-06, + "loss": 0.3761, + "step": 400 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 6.373449802398682, + "learning_rate": 9.79188712522046e-06, + "loss": 0.3674, + "step": 420 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 5.702625274658203, + "learning_rate": 9.756613756613757e-06, + "loss": 0.4013, + "step": 440 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 5.5675153732299805, + "learning_rate": 9.721340388007056e-06, + "loss": 0.3495, + "step": 460 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 6.296374320983887, + "learning_rate": 9.686067019400353e-06, + "loss": 0.3704, + "step": 480 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 4.830463886260986, + "learning_rate": 9.650793650793652e-06, + "loss": 0.36, + "step": 500 + }, + { + "epoch": 0.8375209380234506, + "eval_loss": 0.27764827013015747, + "eval_runtime": 1752.3826, + "eval_samples_per_second": 2.725, + "eval_steps_per_second": 0.341, + "eval_wer": 19.59291781998664, + "step": 500 + }, + { + "epoch": 0.8710217755443886, + "grad_norm": 4.863058090209961, + "learning_rate": 9.61552028218695e-06, + "loss": 0.3874, + "step": 520 + }, + { + "epoch": 0.9045226130653267, + "grad_norm": 6.121027946472168, + "learning_rate": 9.580246913580248e-06, + "loss": 0.382, + "step": 540 + }, + { + "epoch": 0.9380234505862647, + "grad_norm": 4.538082599639893, + "learning_rate": 9.544973544973546e-06, + "loss": 0.3611, + "step": 560 + }, + { + "epoch": 0.9715242881072027, + "grad_norm": 4.8772382736206055, + "learning_rate": 9.509700176366844e-06, + "loss": 0.3584, + "step": 580 + }, + { + "epoch": 1.0050251256281406, + "grad_norm": 4.039212703704834, + "learning_rate": 9.474426807760142e-06, + "loss": 0.3701, + "step": 600 + }, + { + "epoch": 1.0385259631490786, + "grad_norm": 4.787687301635742, + "learning_rate": 9.43915343915344e-06, + "loss": 0.2058, + "step": 620 + }, + { + "epoch": 1.0720268006700167, + "grad_norm": 4.503021717071533, + "learning_rate": 9.403880070546738e-06, + "loss": 0.2455, + "step": 640 + }, + { + "epoch": 1.1055276381909547, + "grad_norm": 6.663857936859131, + "learning_rate": 9.368606701940036e-06, + "loss": 0.2179, + "step": 660 + }, + { + "epoch": 1.1390284757118927, + "grad_norm": 4.486196041107178, + "learning_rate": 9.333333333333334e-06, + "loss": 0.2196, + "step": 680 + }, + { + "epoch": 1.1725293132328307, + "grad_norm": 4.667060852050781, + "learning_rate": 9.298059964726633e-06, + "loss": 0.2183, + "step": 700 + }, + { + "epoch": 1.2060301507537687, + "grad_norm": 5.608316898345947, + "learning_rate": 9.26278659611993e-06, + "loss": 0.2161, + "step": 720 + }, + { + "epoch": 1.2395309882747068, + "grad_norm": 4.2184271812438965, + "learning_rate": 9.227513227513229e-06, + "loss": 0.2382, + "step": 740 + }, + { + "epoch": 1.2730318257956448, + "grad_norm": 3.871945381164551, + "learning_rate": 9.192239858906526e-06, + "loss": 0.2214, + "step": 760 + }, + { + "epoch": 1.3065326633165828, + "grad_norm": 3.730222225189209, + "learning_rate": 9.156966490299825e-06, + "loss": 0.2213, + "step": 780 + }, + { + "epoch": 1.3400335008375208, + "grad_norm": 4.740777015686035, + "learning_rate": 9.121693121693122e-06, + "loss": 0.2503, + "step": 800 + }, + { + "epoch": 1.3735343383584588, + "grad_norm": 4.123469352722168, + "learning_rate": 9.086419753086421e-06, + "loss": 0.2443, + "step": 820 + }, + { + "epoch": 1.4070351758793969, + "grad_norm": 3.988917827606201, + "learning_rate": 9.051146384479718e-06, + "loss": 0.241, + "step": 840 + }, + { + "epoch": 1.4405360134003349, + "grad_norm": 5.4730305671691895, + "learning_rate": 9.015873015873017e-06, + "loss": 0.2163, + "step": 860 + }, + { + "epoch": 1.474036850921273, + "grad_norm": 5.020652770996094, + "learning_rate": 8.980599647266314e-06, + "loss": 0.2387, + "step": 880 + }, + { + "epoch": 1.507537688442211, + "grad_norm": 4.231486797332764, + "learning_rate": 8.945326278659613e-06, + "loss": 0.2523, + "step": 900 + }, + { + "epoch": 1.541038525963149, + "grad_norm": 6.197975158691406, + "learning_rate": 8.910052910052912e-06, + "loss": 0.2255, + "step": 920 + }, + { + "epoch": 1.574539363484087, + "grad_norm": 5.489210605621338, + "learning_rate": 8.874779541446209e-06, + "loss": 0.2397, + "step": 940 + }, + { + "epoch": 1.608040201005025, + "grad_norm": 4.32737398147583, + "learning_rate": 8.839506172839508e-06, + "loss": 0.2298, + "step": 960 + }, + { + "epoch": 1.641541038525963, + "grad_norm": 4.504214763641357, + "learning_rate": 8.804232804232805e-06, + "loss": 0.2301, + "step": 980 + }, + { + "epoch": 1.675041876046901, + "grad_norm": 3.9694759845733643, + "learning_rate": 8.768959435626104e-06, + "loss": 0.2338, + "step": 1000 + }, + { + "epoch": 1.675041876046901, + "eval_loss": 0.18838582932949066, + "eval_runtime": 1770.3449, + "eval_samples_per_second": 2.698, + "eval_steps_per_second": 0.337, + "eval_wer": 14.071299035983584, + "step": 1000 + }, + { + "epoch": 1.708542713567839, + "grad_norm": 4.972934722900391, + "learning_rate": 8.7336860670194e-06, + "loss": 0.2388, + "step": 1020 + }, + { + "epoch": 1.742043551088777, + "grad_norm": 5.178994655609131, + "learning_rate": 8.6984126984127e-06, + "loss": 0.2531, + "step": 1040 + }, + { + "epoch": 1.775544388609715, + "grad_norm": 3.560372829437256, + "learning_rate": 8.663139329805997e-06, + "loss": 0.2187, + "step": 1060 + }, + { + "epoch": 1.809045226130653, + "grad_norm": 4.647324562072754, + "learning_rate": 8.627865961199296e-06, + "loss": 0.2413, + "step": 1080 + }, + { + "epoch": 1.8425460636515911, + "grad_norm": 3.6784422397613525, + "learning_rate": 8.592592592592593e-06, + "loss": 0.2242, + "step": 1100 + }, + { + "epoch": 1.8760469011725294, + "grad_norm": 5.928018093109131, + "learning_rate": 8.557319223985891e-06, + "loss": 0.2406, + "step": 1120 + }, + { + "epoch": 1.9095477386934674, + "grad_norm": 6.358487606048584, + "learning_rate": 8.52204585537919e-06, + "loss": 0.2482, + "step": 1140 + }, + { + "epoch": 1.9430485762144054, + "grad_norm": 4.409506797790527, + "learning_rate": 8.486772486772487e-06, + "loss": 0.2415, + "step": 1160 + }, + { + "epoch": 1.9765494137353434, + "grad_norm": 5.169639587402344, + "learning_rate": 8.451499118165786e-06, + "loss": 0.222, + "step": 1180 + }, + { + "epoch": 2.0100502512562812, + "grad_norm": 2.8841328620910645, + "learning_rate": 8.416225749559083e-06, + "loss": 0.2109, + "step": 1200 + }, + { + "epoch": 2.0435510887772192, + "grad_norm": 3.835498332977295, + "learning_rate": 8.380952380952382e-06, + "loss": 0.1284, + "step": 1220 + }, + { + "epoch": 2.0770519262981573, + "grad_norm": 2.584859848022461, + "learning_rate": 8.34567901234568e-06, + "loss": 0.1059, + "step": 1240 + }, + { + "epoch": 2.1105527638190953, + "grad_norm": 2.797682285308838, + "learning_rate": 8.310405643738978e-06, + "loss": 0.0967, + "step": 1260 + }, + { + "epoch": 2.1440536013400333, + "grad_norm": 3.2937309741973877, + "learning_rate": 8.275132275132275e-06, + "loss": 0.1267, + "step": 1280 + }, + { + "epoch": 2.1775544388609713, + "grad_norm": 3.344325065612793, + "learning_rate": 8.239858906525574e-06, + "loss": 0.1354, + "step": 1300 + }, + { + "epoch": 2.2110552763819094, + "grad_norm": 2.955522060394287, + "learning_rate": 8.204585537918873e-06, + "loss": 0.104, + "step": 1320 + }, + { + "epoch": 2.2445561139028474, + "grad_norm": 5.180908203125, + "learning_rate": 8.16931216931217e-06, + "loss": 0.1229, + "step": 1340 + }, + { + "epoch": 2.2780569514237854, + "grad_norm": 3.639417886734009, + "learning_rate": 8.134038800705469e-06, + "loss": 0.104, + "step": 1360 + }, + { + "epoch": 2.3115577889447234, + "grad_norm": 3.7380902767181396, + "learning_rate": 8.098765432098766e-06, + "loss": 0.1141, + "step": 1380 + }, + { + "epoch": 2.3450586264656614, + "grad_norm": 4.100318908691406, + "learning_rate": 8.063492063492065e-06, + "loss": 0.1123, + "step": 1400 + }, + { + "epoch": 2.3785594639865995, + "grad_norm": 2.8147048950195312, + "learning_rate": 8.028218694885362e-06, + "loss": 0.1039, + "step": 1420 + }, + { + "epoch": 2.4120603015075375, + "grad_norm": 3.7376410961151123, + "learning_rate": 7.99294532627866e-06, + "loss": 0.1179, + "step": 1440 + }, + { + "epoch": 2.4455611390284755, + "grad_norm": 4.320065975189209, + "learning_rate": 7.957671957671958e-06, + "loss": 0.1167, + "step": 1460 + }, + { + "epoch": 2.4790619765494135, + "grad_norm": 4.415127277374268, + "learning_rate": 7.922398589065257e-06, + "loss": 0.125, + "step": 1480 + }, + { + "epoch": 2.5125628140703515, + "grad_norm": 2.631763219833374, + "learning_rate": 7.887125220458554e-06, + "loss": 0.1188, + "step": 1500 + }, + { + "epoch": 2.5125628140703515, + "eval_loss": 0.13635103404521942, + "eval_runtime": 1806.1099, + "eval_samples_per_second": 2.644, + "eval_steps_per_second": 0.331, + "eval_wer": 11.713753937195762, + "step": 1500 + }, + { + "epoch": 2.5460636515912896, + "grad_norm": 3.608773946762085, + "learning_rate": 7.851851851851853e-06, + "loss": 0.1222, + "step": 1520 + }, + { + "epoch": 2.5795644891122276, + "grad_norm": 3.4494569301605225, + "learning_rate": 7.816578483245151e-06, + "loss": 0.1087, + "step": 1540 + }, + { + "epoch": 2.6130653266331656, + "grad_norm": 3.6112048625946045, + "learning_rate": 7.781305114638449e-06, + "loss": 0.1097, + "step": 1560 + }, + { + "epoch": 2.6465661641541036, + "grad_norm": 3.1978707313537598, + "learning_rate": 7.746031746031747e-06, + "loss": 0.1101, + "step": 1580 + }, + { + "epoch": 2.6800670016750416, + "grad_norm": 3.6376733779907227, + "learning_rate": 7.710758377425045e-06, + "loss": 0.1056, + "step": 1600 + }, + { + "epoch": 2.7135678391959797, + "grad_norm": 3.754915237426758, + "learning_rate": 7.675485008818343e-06, + "loss": 0.1103, + "step": 1620 + }, + { + "epoch": 2.7470686767169177, + "grad_norm": 3.329240322113037, + "learning_rate": 7.64021164021164e-06, + "loss": 0.1193, + "step": 1640 + }, + { + "epoch": 2.7805695142378557, + "grad_norm": 3.9300169944763184, + "learning_rate": 7.604938271604939e-06, + "loss": 0.1232, + "step": 1660 + }, + { + "epoch": 2.8140703517587937, + "grad_norm": 4.290626049041748, + "learning_rate": 7.569664902998237e-06, + "loss": 0.1374, + "step": 1680 + }, + { + "epoch": 2.8475711892797317, + "grad_norm": 3.8987998962402344, + "learning_rate": 7.534391534391535e-06, + "loss": 0.1062, + "step": 1700 + }, + { + "epoch": 2.8810720268006698, + "grad_norm": 4.234960079193115, + "learning_rate": 7.499118165784833e-06, + "loss": 0.111, + "step": 1720 + }, + { + "epoch": 2.914572864321608, + "grad_norm": 3.103458881378174, + "learning_rate": 7.463844797178131e-06, + "loss": 0.1076, + "step": 1740 + }, + { + "epoch": 2.948073701842546, + "grad_norm": 3.270204782485962, + "learning_rate": 7.428571428571429e-06, + "loss": 0.1224, + "step": 1760 + }, + { + "epoch": 2.981574539363484, + "grad_norm": 4.261337757110596, + "learning_rate": 7.393298059964727e-06, + "loss": 0.1224, + "step": 1780 + }, + { + "epoch": 3.0150753768844223, + "grad_norm": 1.9531301259994507, + "learning_rate": 7.358024691358025e-06, + "loss": 0.0913, + "step": 1800 + }, + { + "epoch": 3.0485762144053603, + "grad_norm": 1.866215467453003, + "learning_rate": 7.322751322751324e-06, + "loss": 0.0466, + "step": 1820 + }, + { + "epoch": 3.0820770519262983, + "grad_norm": 3.9456610679626465, + "learning_rate": 7.287477954144622e-06, + "loss": 0.0451, + "step": 1840 + }, + { + "epoch": 3.1155778894472363, + "grad_norm": 1.9518849849700928, + "learning_rate": 7.25220458553792e-06, + "loss": 0.0414, + "step": 1860 + }, + { + "epoch": 3.1490787269681744, + "grad_norm": 2.232792854309082, + "learning_rate": 7.216931216931218e-06, + "loss": 0.0502, + "step": 1880 + }, + { + "epoch": 3.1825795644891124, + "grad_norm": 2.074127674102783, + "learning_rate": 7.181657848324516e-06, + "loss": 0.057, + "step": 1900 + }, + { + "epoch": 3.2160804020100504, + "grad_norm": 3.5036733150482178, + "learning_rate": 7.146384479717814e-06, + "loss": 0.0515, + "step": 1920 + }, + { + "epoch": 3.2495812395309884, + "grad_norm": 2.324014186859131, + "learning_rate": 7.111111111111112e-06, + "loss": 0.0454, + "step": 1940 + }, + { + "epoch": 3.2830820770519265, + "grad_norm": 2.710326671600342, + "learning_rate": 7.07583774250441e-06, + "loss": 0.0473, + "step": 1960 + }, + { + "epoch": 3.3165829145728645, + "grad_norm": 2.8943896293640137, + "learning_rate": 7.040564373897708e-06, + "loss": 0.046, + "step": 1980 + }, + { + "epoch": 3.3500837520938025, + "grad_norm": 2.319986581802368, + "learning_rate": 7.005291005291006e-06, + "loss": 0.0449, + "step": 2000 + }, + { + "epoch": 3.3500837520938025, + "eval_loss": 0.10934468358755112, + "eval_runtime": 1762.4506, + "eval_samples_per_second": 2.71, + "eval_steps_per_second": 0.339, + "eval_wer": 7.172854824854443, + "step": 2000 + }, + { + "epoch": 3.3835845896147405, + "grad_norm": 3.317129135131836, + "learning_rate": 6.9700176366843046e-06, + "loss": 0.0645, + "step": 2020 + }, + { + "epoch": 3.4170854271356785, + "grad_norm": 1.9533768892288208, + "learning_rate": 6.9347442680776025e-06, + "loss": 0.0432, + "step": 2040 + }, + { + "epoch": 3.4505862646566166, + "grad_norm": 2.0508453845977783, + "learning_rate": 6.8994708994709005e-06, + "loss": 0.0521, + "step": 2060 + }, + { + "epoch": 3.4840871021775546, + "grad_norm": 2.163236141204834, + "learning_rate": 6.8641975308641985e-06, + "loss": 0.0529, + "step": 2080 + }, + { + "epoch": 3.5175879396984926, + "grad_norm": 2.7154581546783447, + "learning_rate": 6.8289241622574965e-06, + "loss": 0.0452, + "step": 2100 + }, + { + "epoch": 3.5510887772194306, + "grad_norm": 3.0822432041168213, + "learning_rate": 6.7936507936507944e-06, + "loss": 0.0546, + "step": 2120 + }, + { + "epoch": 3.5845896147403686, + "grad_norm": 4.19010591506958, + "learning_rate": 6.758377425044092e-06, + "loss": 0.0529, + "step": 2140 + }, + { + "epoch": 3.6180904522613067, + "grad_norm": 2.9883594512939453, + "learning_rate": 6.72310405643739e-06, + "loss": 0.0503, + "step": 2160 + }, + { + "epoch": 3.6515912897822447, + "grad_norm": 2.3664371967315674, + "learning_rate": 6.687830687830688e-06, + "loss": 0.0498, + "step": 2180 + }, + { + "epoch": 3.6850921273031827, + "grad_norm": 2.0549991130828857, + "learning_rate": 6.652557319223986e-06, + "loss": 0.051, + "step": 2200 + }, + { + "epoch": 3.7185929648241207, + "grad_norm": 2.5339038372039795, + "learning_rate": 6.617283950617285e-06, + "loss": 0.0568, + "step": 2220 + }, + { + "epoch": 3.7520938023450587, + "grad_norm": 1.9988099336624146, + "learning_rate": 6.582010582010583e-06, + "loss": 0.051, + "step": 2240 + }, + { + "epoch": 3.7855946398659968, + "grad_norm": 2.5243782997131348, + "learning_rate": 6.546737213403881e-06, + "loss": 0.056, + "step": 2260 + }, + { + "epoch": 3.819095477386935, + "grad_norm": 3.157158136367798, + "learning_rate": 6.511463844797179e-06, + "loss": 0.0497, + "step": 2280 + }, + { + "epoch": 3.852596314907873, + "grad_norm": 1.9286202192306519, + "learning_rate": 6.476190476190477e-06, + "loss": 0.0426, + "step": 2300 + }, + { + "epoch": 3.886097152428811, + "grad_norm": 3.808802604675293, + "learning_rate": 6.440917107583775e-06, + "loss": 0.0499, + "step": 2320 + }, + { + "epoch": 3.919597989949749, + "grad_norm": 2.506671667098999, + "learning_rate": 6.405643738977073e-06, + "loss": 0.052, + "step": 2340 + }, + { + "epoch": 3.953098827470687, + "grad_norm": 2.9451920986175537, + "learning_rate": 6.370370370370371e-06, + "loss": 0.0552, + "step": 2360 + }, + { + "epoch": 3.986599664991625, + "grad_norm": 2.592744827270508, + "learning_rate": 6.335097001763669e-06, + "loss": 0.0527, + "step": 2380 + }, + { + "epoch": 4.0201005025125625, + "grad_norm": 1.8891575336456299, + "learning_rate": 6.299823633156967e-06, + "loss": 0.0289, + "step": 2400 + }, + { + "epoch": 4.0536013400335005, + "grad_norm": 1.8053243160247803, + "learning_rate": 6.264550264550266e-06, + "loss": 0.0192, + "step": 2420 + }, + { + "epoch": 4.0871021775544385, + "grad_norm": 2.0084407329559326, + "learning_rate": 6.229276895943564e-06, + "loss": 0.0242, + "step": 2440 + }, + { + "epoch": 4.1206030150753765, + "grad_norm": 1.5919119119644165, + "learning_rate": 6.194003527336862e-06, + "loss": 0.0211, + "step": 2460 + }, + { + "epoch": 4.1541038525963145, + "grad_norm": 1.9214613437652588, + "learning_rate": 6.15873015873016e-06, + "loss": 0.0233, + "step": 2480 + }, + { + "epoch": 4.187604690117253, + "grad_norm": 1.2652311325073242, + "learning_rate": 6.123456790123458e-06, + "loss": 0.0199, + "step": 2500 + }, + { + "epoch": 4.187604690117253, + "eval_loss": 0.0981329157948494, + "eval_runtime": 1779.0213, + "eval_samples_per_second": 2.685, + "eval_steps_per_second": 0.336, + "eval_wer": 6.707549871146321, + "step": 2500 + } + ], + "logging_steps": 20, + "max_steps": 5970, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "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": 2.308336900079616e+19, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-2500/training_args.bin b/checkpoint-2500/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6ffec31ca17b32691fb4cd1278f73919f3259cf4 --- /dev/null +++ b/checkpoint-2500/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41a3c645624f7463029286b7d5e1ad6924cfc44a0d153880431b77c40e5deb73 +size 5432 diff --git a/checkpoint-3000/config.json b/checkpoint-3000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..337642e18341c758364c32c6b6159a090615329b --- /dev/null +++ b/checkpoint-3000/config.json @@ -0,0 +1,47 @@ +{ + "activation_dropout": 0.0, + "activation_function": "gelu", + "apply_spec_augment": false, + "architectures": [ + "WhisperForConditionalGeneration" + ], + "attention_dropout": 0.0, + "begin_suppress_tokens": null, + "bos_token_id": 50257, + "classifier_proj_size": 256, + "d_model": 768, + "decoder_attention_heads": 12, + "decoder_ffn_dim": 3072, + "decoder_layerdrop": 0.0, + "decoder_layers": 12, + "decoder_start_token_id": 50258, + "dropout": 0.0, + "encoder_attention_heads": 12, + "encoder_ffn_dim": 3072, + "encoder_layerdrop": 0.0, + "encoder_layers": 12, + "eos_token_id": 50257, + "forced_decoder_ids": null, + "init_std": 0.02, + "is_encoder_decoder": true, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "max_length": null, + "max_source_positions": 1500, + "max_target_positions": 448, + "median_filter_width": 7, + "model_type": "whisper", + "num_hidden_layers": 12, + "num_mel_bins": 80, + "pad_token_id": 50257, + "scale_embedding": false, + "torch_dtype": "float32", + "transformers_version": "4.51.3", + "use_cache": false, + "use_weighted_layer_sum": false, + "vocab_size": 51865 +} diff --git a/checkpoint-3000/generation_config.json b/checkpoint-3000/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..e1f4120825a303544ca23708d9dc52bbaa1dce16 --- /dev/null +++ b/checkpoint-3000/generation_config.json @@ -0,0 +1,9 @@ +{ + "bos_token_id": 50257, + "decoder_start_token_id": 50258, + "eos_token_id": 50257, + "pad_token_id": 50257, + "suppress_tokens": [], + "transformers_version": "4.51.3", + "use_cache": false +} diff --git a/checkpoint-3000/model.safetensors b/checkpoint-3000/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..21e309140a06ae0615a3d3a17fb5bc803af82c46 --- /dev/null +++ b/checkpoint-3000/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0abcf9469cdebe6dab6eca1ce89a6fe70fb85e3e8364817748f1b9707ee82afd +size 966995080 diff --git a/checkpoint-3000/optimizer.pt b/checkpoint-3000/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..7796fef992026636b43a7a682c67e15be7d3b7bc --- /dev/null +++ b/checkpoint-3000/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:833b2c174e0af52976ea986fa4f02df78ec80fedb5801e9aaf6ac3a26d0689c5 +size 1925064044 diff --git a/checkpoint-3000/preprocessor_config.json b/checkpoint-3000/preprocessor_config.json new file mode 100644 index 0000000000000000000000000000000000000000..d7c5b1dfd466c815664b16139e8f605620d22e4e --- /dev/null +++ b/checkpoint-3000/preprocessor_config.json @@ -0,0 +1,15 @@ +{ + "chunk_length": 30, + "dither": 0.0, + "feature_extractor_type": "WhisperFeatureExtractor", + "feature_size": 80, + "hop_length": 160, + "n_fft": 400, + "n_samples": 480000, + "nb_max_frames": 3000, + "padding_side": "right", + "padding_value": 0.0, + "processor_class": "WhisperProcessor", + "return_attention_mask": false, + "sampling_rate": 16000 +} diff --git a/checkpoint-3000/rng_state.pth b/checkpoint-3000/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..604a49cb91e7800f3e236c7a9951bdd298546099 --- /dev/null +++ b/checkpoint-3000/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:27c93067937ab1be348de455c016b3ce31c80733910a246e9c9ccaeebb34c39a +size 14244 diff --git a/checkpoint-3000/scaler.pt b/checkpoint-3000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..ed2e8137c905230d4e6c1d87af745d88e7b8cf35 --- /dev/null +++ b/checkpoint-3000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:67811e0c62d0a1a53aaafe03633e26905661076c1c0120aa29a847c54a088eed +size 988 diff --git a/checkpoint-3000/scheduler.pt b/checkpoint-3000/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..6a0959364ea9ba046f85e52c2e8c755ee982c132 --- /dev/null +++ b/checkpoint-3000/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b8d74175fdadc4a37aa16ac28556cb53a4295de37309a86bc03e28b0ad0b32e +size 1064 diff --git a/checkpoint-3000/trainer_state.json b/checkpoint-3000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..c932cf762a49c6fea393630c1da6b293740546b6 --- /dev/null +++ b/checkpoint-3000/trainer_state.json @@ -0,0 +1,1138 @@ +{ + "best_global_step": 3000, + "best_metric": 5.500143170754987, + "best_model_checkpoint": "./SALAMA_C3/checkpoint-3000", + "epoch": 5.025125628140704, + "eval_steps": 500, + "global_step": 3000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.03350083752093802, + "grad_norm": 11.757207870483398, + "learning_rate": 6.333333333333334e-07, + "loss": 0.5112, + "step": 20 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 10.197953224182129, + "learning_rate": 1.3e-06, + "loss": 0.5283, + "step": 40 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 12.811074256896973, + "learning_rate": 1.9666666666666668e-06, + "loss": 0.474, + "step": 60 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 8.559342384338379, + "learning_rate": 2.6e-06, + "loss": 0.424, + "step": 80 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 6.9629225730896, + "learning_rate": 3.266666666666667e-06, + "loss": 0.4465, + "step": 100 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 7.833058834075928, + "learning_rate": 3.9333333333333335e-06, + "loss": 0.3764, + "step": 120 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 6.882424831390381, + "learning_rate": 4.600000000000001e-06, + "loss": 0.3705, + "step": 140 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 5.4183244705200195, + "learning_rate": 5.2666666666666665e-06, + "loss": 0.4116, + "step": 160 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 6.103787899017334, + "learning_rate": 5.933333333333335e-06, + "loss": 0.3903, + "step": 180 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 6.084160804748535, + "learning_rate": 6.600000000000001e-06, + "loss": 0.3794, + "step": 200 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 4.531406879425049, + "learning_rate": 7.266666666666668e-06, + "loss": 0.3904, + "step": 220 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 6.702854156494141, + "learning_rate": 7.933333333333334e-06, + "loss": 0.3672, + "step": 240 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 6.386377811431885, + "learning_rate": 8.6e-06, + "loss": 0.3379, + "step": 260 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 5.33281135559082, + "learning_rate": 9.266666666666667e-06, + "loss": 0.4102, + "step": 280 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 6.031435012817383, + "learning_rate": 9.933333333333334e-06, + "loss": 0.3815, + "step": 300 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 4.8638176918029785, + "learning_rate": 9.968253968253969e-06, + "loss": 0.358, + "step": 320 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 5.352113723754883, + "learning_rate": 9.932980599647268e-06, + "loss": 0.3578, + "step": 340 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 5.560739040374756, + "learning_rate": 9.897707231040565e-06, + "loss": 0.3795, + "step": 360 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 6.186940670013428, + "learning_rate": 9.862433862433864e-06, + "loss": 0.3852, + "step": 380 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 5.37507438659668, + "learning_rate": 9.827160493827161e-06, + "loss": 0.3761, + "step": 400 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 6.373449802398682, + "learning_rate": 9.79188712522046e-06, + "loss": 0.3674, + "step": 420 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 5.702625274658203, + "learning_rate": 9.756613756613757e-06, + "loss": 0.4013, + "step": 440 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 5.5675153732299805, + "learning_rate": 9.721340388007056e-06, + "loss": 0.3495, + "step": 460 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 6.296374320983887, + "learning_rate": 9.686067019400353e-06, + "loss": 0.3704, + "step": 480 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 4.830463886260986, + "learning_rate": 9.650793650793652e-06, + "loss": 0.36, + "step": 500 + }, + { + "epoch": 0.8375209380234506, + "eval_loss": 0.27764827013015747, + "eval_runtime": 1752.3826, + "eval_samples_per_second": 2.725, + "eval_steps_per_second": 0.341, + "eval_wer": 19.59291781998664, + "step": 500 + }, + { + "epoch": 0.8710217755443886, + "grad_norm": 4.863058090209961, + "learning_rate": 9.61552028218695e-06, + "loss": 0.3874, + "step": 520 + }, + { + "epoch": 0.9045226130653267, + "grad_norm": 6.121027946472168, + "learning_rate": 9.580246913580248e-06, + "loss": 0.382, + "step": 540 + }, + { + "epoch": 0.9380234505862647, + "grad_norm": 4.538082599639893, + "learning_rate": 9.544973544973546e-06, + "loss": 0.3611, + "step": 560 + }, + { + "epoch": 0.9715242881072027, + "grad_norm": 4.8772382736206055, + "learning_rate": 9.509700176366844e-06, + "loss": 0.3584, + "step": 580 + }, + { + "epoch": 1.0050251256281406, + "grad_norm": 4.039212703704834, + "learning_rate": 9.474426807760142e-06, + "loss": 0.3701, + "step": 600 + }, + { + "epoch": 1.0385259631490786, + "grad_norm": 4.787687301635742, + "learning_rate": 9.43915343915344e-06, + "loss": 0.2058, + "step": 620 + }, + { + "epoch": 1.0720268006700167, + "grad_norm": 4.503021717071533, + "learning_rate": 9.403880070546738e-06, + "loss": 0.2455, + "step": 640 + }, + { + "epoch": 1.1055276381909547, + "grad_norm": 6.663857936859131, + "learning_rate": 9.368606701940036e-06, + "loss": 0.2179, + "step": 660 + }, + { + "epoch": 1.1390284757118927, + "grad_norm": 4.486196041107178, + "learning_rate": 9.333333333333334e-06, + "loss": 0.2196, + "step": 680 + }, + { + "epoch": 1.1725293132328307, + "grad_norm": 4.667060852050781, + "learning_rate": 9.298059964726633e-06, + "loss": 0.2183, + "step": 700 + }, + { + "epoch": 1.2060301507537687, + "grad_norm": 5.608316898345947, + "learning_rate": 9.26278659611993e-06, + "loss": 0.2161, + "step": 720 + }, + { + "epoch": 1.2395309882747068, + "grad_norm": 4.2184271812438965, + "learning_rate": 9.227513227513229e-06, + "loss": 0.2382, + "step": 740 + }, + { + "epoch": 1.2730318257956448, + "grad_norm": 3.871945381164551, + "learning_rate": 9.192239858906526e-06, + "loss": 0.2214, + "step": 760 + }, + { + "epoch": 1.3065326633165828, + "grad_norm": 3.730222225189209, + "learning_rate": 9.156966490299825e-06, + "loss": 0.2213, + "step": 780 + }, + { + "epoch": 1.3400335008375208, + "grad_norm": 4.740777015686035, + "learning_rate": 9.121693121693122e-06, + "loss": 0.2503, + "step": 800 + }, + { + "epoch": 1.3735343383584588, + "grad_norm": 4.123469352722168, + "learning_rate": 9.086419753086421e-06, + "loss": 0.2443, + "step": 820 + }, + { + "epoch": 1.4070351758793969, + "grad_norm": 3.988917827606201, + "learning_rate": 9.051146384479718e-06, + "loss": 0.241, + "step": 840 + }, + { + "epoch": 1.4405360134003349, + "grad_norm": 5.4730305671691895, + "learning_rate": 9.015873015873017e-06, + "loss": 0.2163, + "step": 860 + }, + { + "epoch": 1.474036850921273, + "grad_norm": 5.020652770996094, + "learning_rate": 8.980599647266314e-06, + "loss": 0.2387, + "step": 880 + }, + { + "epoch": 1.507537688442211, + "grad_norm": 4.231486797332764, + "learning_rate": 8.945326278659613e-06, + "loss": 0.2523, + "step": 900 + }, + { + "epoch": 1.541038525963149, + "grad_norm": 6.197975158691406, + "learning_rate": 8.910052910052912e-06, + "loss": 0.2255, + "step": 920 + }, + { + "epoch": 1.574539363484087, + "grad_norm": 5.489210605621338, + "learning_rate": 8.874779541446209e-06, + "loss": 0.2397, + "step": 940 + }, + { + "epoch": 1.608040201005025, + "grad_norm": 4.32737398147583, + "learning_rate": 8.839506172839508e-06, + "loss": 0.2298, + "step": 960 + }, + { + "epoch": 1.641541038525963, + "grad_norm": 4.504214763641357, + "learning_rate": 8.804232804232805e-06, + "loss": 0.2301, + "step": 980 + }, + { + "epoch": 1.675041876046901, + "grad_norm": 3.9694759845733643, + "learning_rate": 8.768959435626104e-06, + "loss": 0.2338, + "step": 1000 + }, + { + "epoch": 1.675041876046901, + "eval_loss": 0.18838582932949066, + "eval_runtime": 1770.3449, + "eval_samples_per_second": 2.698, + "eval_steps_per_second": 0.337, + "eval_wer": 14.071299035983584, + "step": 1000 + }, + { + "epoch": 1.708542713567839, + "grad_norm": 4.972934722900391, + "learning_rate": 8.7336860670194e-06, + "loss": 0.2388, + "step": 1020 + }, + { + "epoch": 1.742043551088777, + "grad_norm": 5.178994655609131, + "learning_rate": 8.6984126984127e-06, + "loss": 0.2531, + "step": 1040 + }, + { + "epoch": 1.775544388609715, + "grad_norm": 3.560372829437256, + "learning_rate": 8.663139329805997e-06, + "loss": 0.2187, + "step": 1060 + }, + { + "epoch": 1.809045226130653, + "grad_norm": 4.647324562072754, + "learning_rate": 8.627865961199296e-06, + "loss": 0.2413, + "step": 1080 + }, + { + "epoch": 1.8425460636515911, + "grad_norm": 3.6784422397613525, + "learning_rate": 8.592592592592593e-06, + "loss": 0.2242, + "step": 1100 + }, + { + "epoch": 1.8760469011725294, + "grad_norm": 5.928018093109131, + "learning_rate": 8.557319223985891e-06, + "loss": 0.2406, + "step": 1120 + }, + { + "epoch": 1.9095477386934674, + "grad_norm": 6.358487606048584, + "learning_rate": 8.52204585537919e-06, + "loss": 0.2482, + "step": 1140 + }, + { + "epoch": 1.9430485762144054, + "grad_norm": 4.409506797790527, + "learning_rate": 8.486772486772487e-06, + "loss": 0.2415, + "step": 1160 + }, + { + "epoch": 1.9765494137353434, + "grad_norm": 5.169639587402344, + "learning_rate": 8.451499118165786e-06, + "loss": 0.222, + "step": 1180 + }, + { + "epoch": 2.0100502512562812, + "grad_norm": 2.8841328620910645, + "learning_rate": 8.416225749559083e-06, + "loss": 0.2109, + "step": 1200 + }, + { + "epoch": 2.0435510887772192, + "grad_norm": 3.835498332977295, + "learning_rate": 8.380952380952382e-06, + "loss": 0.1284, + "step": 1220 + }, + { + "epoch": 2.0770519262981573, + "grad_norm": 2.584859848022461, + "learning_rate": 8.34567901234568e-06, + "loss": 0.1059, + "step": 1240 + }, + { + "epoch": 2.1105527638190953, + "grad_norm": 2.797682285308838, + "learning_rate": 8.310405643738978e-06, + "loss": 0.0967, + "step": 1260 + }, + { + "epoch": 2.1440536013400333, + "grad_norm": 3.2937309741973877, + "learning_rate": 8.275132275132275e-06, + "loss": 0.1267, + "step": 1280 + }, + { + "epoch": 2.1775544388609713, + "grad_norm": 3.344325065612793, + "learning_rate": 8.239858906525574e-06, + "loss": 0.1354, + "step": 1300 + }, + { + "epoch": 2.2110552763819094, + "grad_norm": 2.955522060394287, + "learning_rate": 8.204585537918873e-06, + "loss": 0.104, + "step": 1320 + }, + { + "epoch": 2.2445561139028474, + "grad_norm": 5.180908203125, + "learning_rate": 8.16931216931217e-06, + "loss": 0.1229, + "step": 1340 + }, + { + "epoch": 2.2780569514237854, + "grad_norm": 3.639417886734009, + "learning_rate": 8.134038800705469e-06, + "loss": 0.104, + "step": 1360 + }, + { + "epoch": 2.3115577889447234, + "grad_norm": 3.7380902767181396, + "learning_rate": 8.098765432098766e-06, + "loss": 0.1141, + "step": 1380 + }, + { + "epoch": 2.3450586264656614, + "grad_norm": 4.100318908691406, + "learning_rate": 8.063492063492065e-06, + "loss": 0.1123, + "step": 1400 + }, + { + "epoch": 2.3785594639865995, + "grad_norm": 2.8147048950195312, + "learning_rate": 8.028218694885362e-06, + "loss": 0.1039, + "step": 1420 + }, + { + "epoch": 2.4120603015075375, + "grad_norm": 3.7376410961151123, + "learning_rate": 7.99294532627866e-06, + "loss": 0.1179, + "step": 1440 + }, + { + "epoch": 2.4455611390284755, + "grad_norm": 4.320065975189209, + "learning_rate": 7.957671957671958e-06, + "loss": 0.1167, + "step": 1460 + }, + { + "epoch": 2.4790619765494135, + "grad_norm": 4.415127277374268, + "learning_rate": 7.922398589065257e-06, + "loss": 0.125, + "step": 1480 + }, + { + "epoch": 2.5125628140703515, + "grad_norm": 2.631763219833374, + "learning_rate": 7.887125220458554e-06, + "loss": 0.1188, + "step": 1500 + }, + { + "epoch": 2.5125628140703515, + "eval_loss": 0.13635103404521942, + "eval_runtime": 1806.1099, + "eval_samples_per_second": 2.644, + "eval_steps_per_second": 0.331, + "eval_wer": 11.713753937195762, + "step": 1500 + }, + { + "epoch": 2.5460636515912896, + "grad_norm": 3.608773946762085, + "learning_rate": 7.851851851851853e-06, + "loss": 0.1222, + "step": 1520 + }, + { + "epoch": 2.5795644891122276, + "grad_norm": 3.4494569301605225, + "learning_rate": 7.816578483245151e-06, + "loss": 0.1087, + "step": 1540 + }, + { + "epoch": 2.6130653266331656, + "grad_norm": 3.6112048625946045, + "learning_rate": 7.781305114638449e-06, + "loss": 0.1097, + "step": 1560 + }, + { + "epoch": 2.6465661641541036, + "grad_norm": 3.1978707313537598, + "learning_rate": 7.746031746031747e-06, + "loss": 0.1101, + "step": 1580 + }, + { + "epoch": 2.6800670016750416, + "grad_norm": 3.6376733779907227, + "learning_rate": 7.710758377425045e-06, + "loss": 0.1056, + "step": 1600 + }, + { + "epoch": 2.7135678391959797, + "grad_norm": 3.754915237426758, + "learning_rate": 7.675485008818343e-06, + "loss": 0.1103, + "step": 1620 + }, + { + "epoch": 2.7470686767169177, + "grad_norm": 3.329240322113037, + "learning_rate": 7.64021164021164e-06, + "loss": 0.1193, + "step": 1640 + }, + { + "epoch": 2.7805695142378557, + "grad_norm": 3.9300169944763184, + "learning_rate": 7.604938271604939e-06, + "loss": 0.1232, + "step": 1660 + }, + { + "epoch": 2.8140703517587937, + "grad_norm": 4.290626049041748, + "learning_rate": 7.569664902998237e-06, + "loss": 0.1374, + "step": 1680 + }, + { + "epoch": 2.8475711892797317, + "grad_norm": 3.8987998962402344, + "learning_rate": 7.534391534391535e-06, + "loss": 0.1062, + "step": 1700 + }, + { + "epoch": 2.8810720268006698, + "grad_norm": 4.234960079193115, + "learning_rate": 7.499118165784833e-06, + "loss": 0.111, + "step": 1720 + }, + { + "epoch": 2.914572864321608, + "grad_norm": 3.103458881378174, + "learning_rate": 7.463844797178131e-06, + "loss": 0.1076, + "step": 1740 + }, + { + "epoch": 2.948073701842546, + "grad_norm": 3.270204782485962, + "learning_rate": 7.428571428571429e-06, + "loss": 0.1224, + "step": 1760 + }, + { + "epoch": 2.981574539363484, + "grad_norm": 4.261337757110596, + "learning_rate": 7.393298059964727e-06, + "loss": 0.1224, + "step": 1780 + }, + { + "epoch": 3.0150753768844223, + "grad_norm": 1.9531301259994507, + "learning_rate": 7.358024691358025e-06, + "loss": 0.0913, + "step": 1800 + }, + { + "epoch": 3.0485762144053603, + "grad_norm": 1.866215467453003, + "learning_rate": 7.322751322751324e-06, + "loss": 0.0466, + "step": 1820 + }, + { + "epoch": 3.0820770519262983, + "grad_norm": 3.9456610679626465, + "learning_rate": 7.287477954144622e-06, + "loss": 0.0451, + "step": 1840 + }, + { + "epoch": 3.1155778894472363, + "grad_norm": 1.9518849849700928, + "learning_rate": 7.25220458553792e-06, + "loss": 0.0414, + "step": 1860 + }, + { + "epoch": 3.1490787269681744, + "grad_norm": 2.232792854309082, + "learning_rate": 7.216931216931218e-06, + "loss": 0.0502, + "step": 1880 + }, + { + "epoch": 3.1825795644891124, + "grad_norm": 2.074127674102783, + "learning_rate": 7.181657848324516e-06, + "loss": 0.057, + "step": 1900 + }, + { + "epoch": 3.2160804020100504, + "grad_norm": 3.5036733150482178, + "learning_rate": 7.146384479717814e-06, + "loss": 0.0515, + "step": 1920 + }, + { + "epoch": 3.2495812395309884, + "grad_norm": 2.324014186859131, + "learning_rate": 7.111111111111112e-06, + "loss": 0.0454, + "step": 1940 + }, + { + "epoch": 3.2830820770519265, + "grad_norm": 2.710326671600342, + "learning_rate": 7.07583774250441e-06, + "loss": 0.0473, + "step": 1960 + }, + { + "epoch": 3.3165829145728645, + "grad_norm": 2.8943896293640137, + "learning_rate": 7.040564373897708e-06, + "loss": 0.046, + "step": 1980 + }, + { + "epoch": 3.3500837520938025, + "grad_norm": 2.319986581802368, + "learning_rate": 7.005291005291006e-06, + "loss": 0.0449, + "step": 2000 + }, + { + "epoch": 3.3500837520938025, + "eval_loss": 0.10934468358755112, + "eval_runtime": 1762.4506, + "eval_samples_per_second": 2.71, + "eval_steps_per_second": 0.339, + "eval_wer": 7.172854824854443, + "step": 2000 + }, + { + "epoch": 3.3835845896147405, + "grad_norm": 3.317129135131836, + "learning_rate": 6.9700176366843046e-06, + "loss": 0.0645, + "step": 2020 + }, + { + "epoch": 3.4170854271356785, + "grad_norm": 1.9533768892288208, + "learning_rate": 6.9347442680776025e-06, + "loss": 0.0432, + "step": 2040 + }, + { + "epoch": 3.4505862646566166, + "grad_norm": 2.0508453845977783, + "learning_rate": 6.8994708994709005e-06, + "loss": 0.0521, + "step": 2060 + }, + { + "epoch": 3.4840871021775546, + "grad_norm": 2.163236141204834, + "learning_rate": 6.8641975308641985e-06, + "loss": 0.0529, + "step": 2080 + }, + { + "epoch": 3.5175879396984926, + "grad_norm": 2.7154581546783447, + "learning_rate": 6.8289241622574965e-06, + "loss": 0.0452, + "step": 2100 + }, + { + "epoch": 3.5510887772194306, + "grad_norm": 3.0822432041168213, + "learning_rate": 6.7936507936507944e-06, + "loss": 0.0546, + "step": 2120 + }, + { + "epoch": 3.5845896147403686, + "grad_norm": 4.19010591506958, + "learning_rate": 6.758377425044092e-06, + "loss": 0.0529, + "step": 2140 + }, + { + "epoch": 3.6180904522613067, + "grad_norm": 2.9883594512939453, + "learning_rate": 6.72310405643739e-06, + "loss": 0.0503, + "step": 2160 + }, + { + "epoch": 3.6515912897822447, + "grad_norm": 2.3664371967315674, + "learning_rate": 6.687830687830688e-06, + "loss": 0.0498, + "step": 2180 + }, + { + "epoch": 3.6850921273031827, + "grad_norm": 2.0549991130828857, + "learning_rate": 6.652557319223986e-06, + "loss": 0.051, + "step": 2200 + }, + { + "epoch": 3.7185929648241207, + "grad_norm": 2.5339038372039795, + "learning_rate": 6.617283950617285e-06, + "loss": 0.0568, + "step": 2220 + }, + { + "epoch": 3.7520938023450587, + "grad_norm": 1.9988099336624146, + "learning_rate": 6.582010582010583e-06, + "loss": 0.051, + "step": 2240 + }, + { + "epoch": 3.7855946398659968, + "grad_norm": 2.5243782997131348, + "learning_rate": 6.546737213403881e-06, + "loss": 0.056, + "step": 2260 + }, + { + "epoch": 3.819095477386935, + "grad_norm": 3.157158136367798, + "learning_rate": 6.511463844797179e-06, + "loss": 0.0497, + "step": 2280 + }, + { + "epoch": 3.852596314907873, + "grad_norm": 1.9286202192306519, + "learning_rate": 6.476190476190477e-06, + "loss": 0.0426, + "step": 2300 + }, + { + "epoch": 3.886097152428811, + "grad_norm": 3.808802604675293, + "learning_rate": 6.440917107583775e-06, + "loss": 0.0499, + "step": 2320 + }, + { + "epoch": 3.919597989949749, + "grad_norm": 2.506671667098999, + "learning_rate": 6.405643738977073e-06, + "loss": 0.052, + "step": 2340 + }, + { + "epoch": 3.953098827470687, + "grad_norm": 2.9451920986175537, + "learning_rate": 6.370370370370371e-06, + "loss": 0.0552, + "step": 2360 + }, + { + "epoch": 3.986599664991625, + "grad_norm": 2.592744827270508, + "learning_rate": 6.335097001763669e-06, + "loss": 0.0527, + "step": 2380 + }, + { + "epoch": 4.0201005025125625, + "grad_norm": 1.8891575336456299, + "learning_rate": 6.299823633156967e-06, + "loss": 0.0289, + "step": 2400 + }, + { + "epoch": 4.0536013400335005, + "grad_norm": 1.8053243160247803, + "learning_rate": 6.264550264550266e-06, + "loss": 0.0192, + "step": 2420 + }, + { + "epoch": 4.0871021775544385, + "grad_norm": 2.0084407329559326, + "learning_rate": 6.229276895943564e-06, + "loss": 0.0242, + "step": 2440 + }, + { + "epoch": 4.1206030150753765, + "grad_norm": 1.5919119119644165, + "learning_rate": 6.194003527336862e-06, + "loss": 0.0211, + "step": 2460 + }, + { + "epoch": 4.1541038525963145, + "grad_norm": 1.9214613437652588, + "learning_rate": 6.15873015873016e-06, + "loss": 0.0233, + "step": 2480 + }, + { + "epoch": 4.187604690117253, + "grad_norm": 1.2652311325073242, + "learning_rate": 6.123456790123458e-06, + "loss": 0.0199, + "step": 2500 + }, + { + "epoch": 4.187604690117253, + "eval_loss": 0.0981329157948494, + "eval_runtime": 1779.0213, + "eval_samples_per_second": 2.685, + "eval_steps_per_second": 0.336, + "eval_wer": 6.707549871146321, + "step": 2500 + }, + { + "epoch": 4.221105527638191, + "grad_norm": 2.026528835296631, + "learning_rate": 6.088183421516756e-06, + "loss": 0.0217, + "step": 2520 + }, + { + "epoch": 4.254606365159129, + "grad_norm": 1.596919059753418, + "learning_rate": 6.052910052910054e-06, + "loss": 0.0167, + "step": 2540 + }, + { + "epoch": 4.288107202680067, + "grad_norm": 2.9445090293884277, + "learning_rate": 6.017636684303352e-06, + "loss": 0.0225, + "step": 2560 + }, + { + "epoch": 4.321608040201005, + "grad_norm": 2.4160282611846924, + "learning_rate": 5.9823633156966496e-06, + "loss": 0.0253, + "step": 2580 + }, + { + "epoch": 4.355108877721943, + "grad_norm": 1.461127758026123, + "learning_rate": 5.9470899470899475e-06, + "loss": 0.0197, + "step": 2600 + }, + { + "epoch": 4.388609715242881, + "grad_norm": 2.7892863750457764, + "learning_rate": 5.911816578483246e-06, + "loss": 0.022, + "step": 2620 + }, + { + "epoch": 4.422110552763819, + "grad_norm": 1.651208758354187, + "learning_rate": 5.876543209876544e-06, + "loss": 0.0215, + "step": 2640 + }, + { + "epoch": 4.455611390284757, + "grad_norm": 2.2500391006469727, + "learning_rate": 5.841269841269842e-06, + "loss": 0.0247, + "step": 2660 + }, + { + "epoch": 4.489112227805695, + "grad_norm": 4.447635173797607, + "learning_rate": 5.80599647266314e-06, + "loss": 0.0263, + "step": 2680 + }, + { + "epoch": 4.522613065326633, + "grad_norm": 0.8300407528877258, + "learning_rate": 5.770723104056438e-06, + "loss": 0.0209, + "step": 2700 + }, + { + "epoch": 4.556113902847571, + "grad_norm": 1.6874111890792847, + "learning_rate": 5.735449735449736e-06, + "loss": 0.0195, + "step": 2720 + }, + { + "epoch": 4.589614740368509, + "grad_norm": 2.4045815467834473, + "learning_rate": 5.700176366843034e-06, + "loss": 0.0224, + "step": 2740 + }, + { + "epoch": 4.623115577889447, + "grad_norm": 2.3160908222198486, + "learning_rate": 5.664902998236332e-06, + "loss": 0.0179, + "step": 2760 + }, + { + "epoch": 4.656616415410385, + "grad_norm": 1.6684287786483765, + "learning_rate": 5.62962962962963e-06, + "loss": 0.0238, + "step": 2780 + }, + { + "epoch": 4.690117252931323, + "grad_norm": 1.973906397819519, + "learning_rate": 5.594356261022928e-06, + "loss": 0.0226, + "step": 2800 + }, + { + "epoch": 4.723618090452261, + "grad_norm": 2.270906686782837, + "learning_rate": 5.559082892416227e-06, + "loss": 0.0232, + "step": 2820 + }, + { + "epoch": 4.757118927973199, + "grad_norm": 1.8875011205673218, + "learning_rate": 5.523809523809525e-06, + "loss": 0.0218, + "step": 2840 + }, + { + "epoch": 4.790619765494137, + "grad_norm": 1.1312583684921265, + "learning_rate": 5.488536155202823e-06, + "loss": 0.0212, + "step": 2860 + }, + { + "epoch": 4.824120603015075, + "grad_norm": 0.864783525466919, + "learning_rate": 5.453262786596121e-06, + "loss": 0.023, + "step": 2880 + }, + { + "epoch": 4.857621440536013, + "grad_norm": 1.2935965061187744, + "learning_rate": 5.417989417989419e-06, + "loss": 0.019, + "step": 2900 + }, + { + "epoch": 4.891122278056951, + "grad_norm": 2.4576382637023926, + "learning_rate": 5.382716049382717e-06, + "loss": 0.0193, + "step": 2920 + }, + { + "epoch": 4.924623115577889, + "grad_norm": 2.71472430229187, + "learning_rate": 5.347442680776015e-06, + "loss": 0.0253, + "step": 2940 + }, + { + "epoch": 4.958123953098827, + "grad_norm": 2.84940505027771, + "learning_rate": 5.312169312169313e-06, + "loss": 0.0218, + "step": 2960 + }, + { + "epoch": 4.991624790619765, + "grad_norm": 1.8483999967575073, + "learning_rate": 5.276895943562611e-06, + "loss": 0.0226, + "step": 2980 + }, + { + "epoch": 5.025125628140704, + "grad_norm": 0.6126876473426819, + "learning_rate": 5.241622574955909e-06, + "loss": 0.0101, + "step": 3000 + }, + { + "epoch": 5.025125628140704, + "eval_loss": 0.09390027821063995, + "eval_runtime": 1767.4512, + "eval_samples_per_second": 2.702, + "eval_steps_per_second": 0.338, + "eval_wer": 5.500143170754987, + "step": 3000 + } + ], + "logging_steps": 20, + "max_steps": 5970, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "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": 2.76998696497152e+19, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-3000/training_args.bin b/checkpoint-3000/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6ffec31ca17b32691fb4cd1278f73919f3259cf4 --- /dev/null +++ b/checkpoint-3000/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41a3c645624f7463029286b7d5e1ad6924cfc44a0d153880431b77c40e5deb73 +size 5432 diff --git a/checkpoint-3500/config.json b/checkpoint-3500/config.json new file mode 100644 index 0000000000000000000000000000000000000000..337642e18341c758364c32c6b6159a090615329b --- /dev/null +++ b/checkpoint-3500/config.json @@ -0,0 +1,47 @@ +{ + "activation_dropout": 0.0, + "activation_function": "gelu", + "apply_spec_augment": false, + "architectures": [ + "WhisperForConditionalGeneration" + ], + "attention_dropout": 0.0, + "begin_suppress_tokens": null, + "bos_token_id": 50257, + "classifier_proj_size": 256, + "d_model": 768, + "decoder_attention_heads": 12, + "decoder_ffn_dim": 3072, + "decoder_layerdrop": 0.0, + "decoder_layers": 12, + "decoder_start_token_id": 50258, + "dropout": 0.0, + "encoder_attention_heads": 12, + "encoder_ffn_dim": 3072, + "encoder_layerdrop": 0.0, + "encoder_layers": 12, + "eos_token_id": 50257, + "forced_decoder_ids": null, + "init_std": 0.02, + "is_encoder_decoder": true, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "max_length": null, + "max_source_positions": 1500, + "max_target_positions": 448, + "median_filter_width": 7, + "model_type": "whisper", + "num_hidden_layers": 12, + "num_mel_bins": 80, + "pad_token_id": 50257, + "scale_embedding": false, + "torch_dtype": "float32", + "transformers_version": "4.51.3", + "use_cache": false, + "use_weighted_layer_sum": false, + "vocab_size": 51865 +} diff --git a/checkpoint-3500/generation_config.json b/checkpoint-3500/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..e1f4120825a303544ca23708d9dc52bbaa1dce16 --- /dev/null +++ b/checkpoint-3500/generation_config.json @@ -0,0 +1,9 @@ +{ + "bos_token_id": 50257, + "decoder_start_token_id": 50258, + "eos_token_id": 50257, + "pad_token_id": 50257, + "suppress_tokens": [], + "transformers_version": "4.51.3", + "use_cache": false +} diff --git a/checkpoint-3500/model.safetensors b/checkpoint-3500/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b25a8d49f7f67b25def854d3d35285ba3112b89d --- /dev/null +++ b/checkpoint-3500/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4c9d5eeebd381f71e8d3521e7bab3b695c785d3985afdf3339eec7ae4d35714e +size 966995080 diff --git a/checkpoint-3500/optimizer.pt b/checkpoint-3500/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..38ac405ab8004d9cded46725fd8ed4af34a420a6 --- /dev/null +++ b/checkpoint-3500/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a263ae7805290868d960a7cc34303bad00070582aa3802a225d1e95944e804c +size 1925064044 diff --git a/checkpoint-3500/preprocessor_config.json b/checkpoint-3500/preprocessor_config.json new file mode 100644 index 0000000000000000000000000000000000000000..d7c5b1dfd466c815664b16139e8f605620d22e4e --- /dev/null +++ b/checkpoint-3500/preprocessor_config.json @@ -0,0 +1,15 @@ +{ + "chunk_length": 30, + "dither": 0.0, + "feature_extractor_type": "WhisperFeatureExtractor", + "feature_size": 80, + "hop_length": 160, + "n_fft": 400, + "n_samples": 480000, + "nb_max_frames": 3000, + "padding_side": "right", + "padding_value": 0.0, + "processor_class": "WhisperProcessor", + "return_attention_mask": false, + "sampling_rate": 16000 +} diff --git a/checkpoint-3500/rng_state.pth b/checkpoint-3500/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..a38df0e52c8bcebc47289236d1ce711e63d11cbe --- /dev/null +++ b/checkpoint-3500/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:704796f5dbce8b1747ecaf7ae0a2ac72f59656dab1929ab508f429a11178eb49 +size 14244 diff --git a/checkpoint-3500/scaler.pt b/checkpoint-3500/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..f10e34580ac1de0872c5e3ec408a223a004964ed --- /dev/null +++ b/checkpoint-3500/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bac5e087c8c89dafdaa3996dd5354333a4ba800c145b65e8d368e26d313f16ae +size 988 diff --git a/checkpoint-3500/scheduler.pt b/checkpoint-3500/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..b629a2bffeb3d37e7408c15acee760efc0db0114 --- /dev/null +++ b/checkpoint-3500/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0e9f98d340308b1c027b262eb10f07c8fa960ab0c11a8efaba725078f548d61a +size 1064 diff --git a/checkpoint-3500/trainer_state.json b/checkpoint-3500/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..15ea9da684b1ae5c636e4f76c6db5994eaaa2e25 --- /dev/null +++ b/checkpoint-3500/trainer_state.json @@ -0,0 +1,1322 @@ +{ + "best_global_step": 3500, + "best_metric": 5.010976424549012, + "best_model_checkpoint": "./SALAMA_C3/checkpoint-3500", + "epoch": 5.8626465661641545, + "eval_steps": 500, + "global_step": 3500, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.03350083752093802, + "grad_norm": 11.757207870483398, + "learning_rate": 6.333333333333334e-07, + "loss": 0.5112, + "step": 20 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 10.197953224182129, + "learning_rate": 1.3e-06, + "loss": 0.5283, + "step": 40 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 12.811074256896973, + "learning_rate": 1.9666666666666668e-06, + "loss": 0.474, + "step": 60 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 8.559342384338379, + "learning_rate": 2.6e-06, + "loss": 0.424, + "step": 80 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 6.9629225730896, + "learning_rate": 3.266666666666667e-06, + "loss": 0.4465, + "step": 100 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 7.833058834075928, + "learning_rate": 3.9333333333333335e-06, + "loss": 0.3764, + "step": 120 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 6.882424831390381, + "learning_rate": 4.600000000000001e-06, + "loss": 0.3705, + "step": 140 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 5.4183244705200195, + "learning_rate": 5.2666666666666665e-06, + "loss": 0.4116, + "step": 160 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 6.103787899017334, + "learning_rate": 5.933333333333335e-06, + "loss": 0.3903, + "step": 180 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 6.084160804748535, + "learning_rate": 6.600000000000001e-06, + "loss": 0.3794, + "step": 200 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 4.531406879425049, + "learning_rate": 7.266666666666668e-06, + "loss": 0.3904, + "step": 220 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 6.702854156494141, + "learning_rate": 7.933333333333334e-06, + "loss": 0.3672, + "step": 240 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 6.386377811431885, + "learning_rate": 8.6e-06, + "loss": 0.3379, + "step": 260 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 5.33281135559082, + "learning_rate": 9.266666666666667e-06, + "loss": 0.4102, + "step": 280 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 6.031435012817383, + "learning_rate": 9.933333333333334e-06, + "loss": 0.3815, + "step": 300 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 4.8638176918029785, + "learning_rate": 9.968253968253969e-06, + "loss": 0.358, + "step": 320 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 5.352113723754883, + "learning_rate": 9.932980599647268e-06, + "loss": 0.3578, + "step": 340 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 5.560739040374756, + "learning_rate": 9.897707231040565e-06, + "loss": 0.3795, + "step": 360 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 6.186940670013428, + "learning_rate": 9.862433862433864e-06, + "loss": 0.3852, + "step": 380 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 5.37507438659668, + "learning_rate": 9.827160493827161e-06, + "loss": 0.3761, + "step": 400 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 6.373449802398682, + "learning_rate": 9.79188712522046e-06, + "loss": 0.3674, + "step": 420 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 5.702625274658203, + "learning_rate": 9.756613756613757e-06, + "loss": 0.4013, + "step": 440 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 5.5675153732299805, + "learning_rate": 9.721340388007056e-06, + "loss": 0.3495, + "step": 460 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 6.296374320983887, + "learning_rate": 9.686067019400353e-06, + "loss": 0.3704, + "step": 480 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 4.830463886260986, + "learning_rate": 9.650793650793652e-06, + "loss": 0.36, + "step": 500 + }, + { + "epoch": 0.8375209380234506, + "eval_loss": 0.27764827013015747, + "eval_runtime": 1752.3826, + "eval_samples_per_second": 2.725, + "eval_steps_per_second": 0.341, + "eval_wer": 19.59291781998664, + "step": 500 + }, + { + "epoch": 0.8710217755443886, + "grad_norm": 4.863058090209961, + "learning_rate": 9.61552028218695e-06, + "loss": 0.3874, + "step": 520 + }, + { + "epoch": 0.9045226130653267, + "grad_norm": 6.121027946472168, + "learning_rate": 9.580246913580248e-06, + "loss": 0.382, + "step": 540 + }, + { + "epoch": 0.9380234505862647, + "grad_norm": 4.538082599639893, + "learning_rate": 9.544973544973546e-06, + "loss": 0.3611, + "step": 560 + }, + { + "epoch": 0.9715242881072027, + "grad_norm": 4.8772382736206055, + "learning_rate": 9.509700176366844e-06, + "loss": 0.3584, + "step": 580 + }, + { + "epoch": 1.0050251256281406, + "grad_norm": 4.039212703704834, + "learning_rate": 9.474426807760142e-06, + "loss": 0.3701, + "step": 600 + }, + { + "epoch": 1.0385259631490786, + "grad_norm": 4.787687301635742, + "learning_rate": 9.43915343915344e-06, + "loss": 0.2058, + "step": 620 + }, + { + "epoch": 1.0720268006700167, + "grad_norm": 4.503021717071533, + "learning_rate": 9.403880070546738e-06, + "loss": 0.2455, + "step": 640 + }, + { + "epoch": 1.1055276381909547, + "grad_norm": 6.663857936859131, + "learning_rate": 9.368606701940036e-06, + "loss": 0.2179, + "step": 660 + }, + { + "epoch": 1.1390284757118927, + "grad_norm": 4.486196041107178, + "learning_rate": 9.333333333333334e-06, + "loss": 0.2196, + "step": 680 + }, + { + "epoch": 1.1725293132328307, + "grad_norm": 4.667060852050781, + "learning_rate": 9.298059964726633e-06, + "loss": 0.2183, + "step": 700 + }, + { + "epoch": 1.2060301507537687, + "grad_norm": 5.608316898345947, + "learning_rate": 9.26278659611993e-06, + "loss": 0.2161, + "step": 720 + }, + { + "epoch": 1.2395309882747068, + "grad_norm": 4.2184271812438965, + "learning_rate": 9.227513227513229e-06, + "loss": 0.2382, + "step": 740 + }, + { + "epoch": 1.2730318257956448, + "grad_norm": 3.871945381164551, + "learning_rate": 9.192239858906526e-06, + "loss": 0.2214, + "step": 760 + }, + { + "epoch": 1.3065326633165828, + "grad_norm": 3.730222225189209, + "learning_rate": 9.156966490299825e-06, + "loss": 0.2213, + "step": 780 + }, + { + "epoch": 1.3400335008375208, + "grad_norm": 4.740777015686035, + "learning_rate": 9.121693121693122e-06, + "loss": 0.2503, + "step": 800 + }, + { + "epoch": 1.3735343383584588, + "grad_norm": 4.123469352722168, + "learning_rate": 9.086419753086421e-06, + "loss": 0.2443, + "step": 820 + }, + { + "epoch": 1.4070351758793969, + "grad_norm": 3.988917827606201, + "learning_rate": 9.051146384479718e-06, + "loss": 0.241, + "step": 840 + }, + { + "epoch": 1.4405360134003349, + "grad_norm": 5.4730305671691895, + "learning_rate": 9.015873015873017e-06, + "loss": 0.2163, + "step": 860 + }, + { + "epoch": 1.474036850921273, + "grad_norm": 5.020652770996094, + "learning_rate": 8.980599647266314e-06, + "loss": 0.2387, + "step": 880 + }, + { + "epoch": 1.507537688442211, + "grad_norm": 4.231486797332764, + "learning_rate": 8.945326278659613e-06, + "loss": 0.2523, + "step": 900 + }, + { + "epoch": 1.541038525963149, + "grad_norm": 6.197975158691406, + "learning_rate": 8.910052910052912e-06, + "loss": 0.2255, + "step": 920 + }, + { + "epoch": 1.574539363484087, + "grad_norm": 5.489210605621338, + "learning_rate": 8.874779541446209e-06, + "loss": 0.2397, + "step": 940 + }, + { + "epoch": 1.608040201005025, + "grad_norm": 4.32737398147583, + "learning_rate": 8.839506172839508e-06, + "loss": 0.2298, + "step": 960 + }, + { + "epoch": 1.641541038525963, + "grad_norm": 4.504214763641357, + "learning_rate": 8.804232804232805e-06, + "loss": 0.2301, + "step": 980 + }, + { + "epoch": 1.675041876046901, + "grad_norm": 3.9694759845733643, + "learning_rate": 8.768959435626104e-06, + "loss": 0.2338, + "step": 1000 + }, + { + "epoch": 1.675041876046901, + "eval_loss": 0.18838582932949066, + "eval_runtime": 1770.3449, + "eval_samples_per_second": 2.698, + "eval_steps_per_second": 0.337, + "eval_wer": 14.071299035983584, + "step": 1000 + }, + { + "epoch": 1.708542713567839, + "grad_norm": 4.972934722900391, + "learning_rate": 8.7336860670194e-06, + "loss": 0.2388, + "step": 1020 + }, + { + "epoch": 1.742043551088777, + "grad_norm": 5.178994655609131, + "learning_rate": 8.6984126984127e-06, + "loss": 0.2531, + "step": 1040 + }, + { + "epoch": 1.775544388609715, + "grad_norm": 3.560372829437256, + "learning_rate": 8.663139329805997e-06, + "loss": 0.2187, + "step": 1060 + }, + { + "epoch": 1.809045226130653, + "grad_norm": 4.647324562072754, + "learning_rate": 8.627865961199296e-06, + "loss": 0.2413, + "step": 1080 + }, + { + "epoch": 1.8425460636515911, + "grad_norm": 3.6784422397613525, + "learning_rate": 8.592592592592593e-06, + "loss": 0.2242, + "step": 1100 + }, + { + "epoch": 1.8760469011725294, + "grad_norm": 5.928018093109131, + "learning_rate": 8.557319223985891e-06, + "loss": 0.2406, + "step": 1120 + }, + { + "epoch": 1.9095477386934674, + "grad_norm": 6.358487606048584, + "learning_rate": 8.52204585537919e-06, + "loss": 0.2482, + "step": 1140 + }, + { + "epoch": 1.9430485762144054, + "grad_norm": 4.409506797790527, + "learning_rate": 8.486772486772487e-06, + "loss": 0.2415, + "step": 1160 + }, + { + "epoch": 1.9765494137353434, + "grad_norm": 5.169639587402344, + "learning_rate": 8.451499118165786e-06, + "loss": 0.222, + "step": 1180 + }, + { + "epoch": 2.0100502512562812, + "grad_norm": 2.8841328620910645, + "learning_rate": 8.416225749559083e-06, + "loss": 0.2109, + "step": 1200 + }, + { + "epoch": 2.0435510887772192, + "grad_norm": 3.835498332977295, + "learning_rate": 8.380952380952382e-06, + "loss": 0.1284, + "step": 1220 + }, + { + "epoch": 2.0770519262981573, + "grad_norm": 2.584859848022461, + "learning_rate": 8.34567901234568e-06, + "loss": 0.1059, + "step": 1240 + }, + { + "epoch": 2.1105527638190953, + "grad_norm": 2.797682285308838, + "learning_rate": 8.310405643738978e-06, + "loss": 0.0967, + "step": 1260 + }, + { + "epoch": 2.1440536013400333, + "grad_norm": 3.2937309741973877, + "learning_rate": 8.275132275132275e-06, + "loss": 0.1267, + "step": 1280 + }, + { + "epoch": 2.1775544388609713, + "grad_norm": 3.344325065612793, + "learning_rate": 8.239858906525574e-06, + "loss": 0.1354, + "step": 1300 + }, + { + "epoch": 2.2110552763819094, + "grad_norm": 2.955522060394287, + "learning_rate": 8.204585537918873e-06, + "loss": 0.104, + "step": 1320 + }, + { + "epoch": 2.2445561139028474, + "grad_norm": 5.180908203125, + "learning_rate": 8.16931216931217e-06, + "loss": 0.1229, + "step": 1340 + }, + { + "epoch": 2.2780569514237854, + "grad_norm": 3.639417886734009, + "learning_rate": 8.134038800705469e-06, + "loss": 0.104, + "step": 1360 + }, + { + "epoch": 2.3115577889447234, + "grad_norm": 3.7380902767181396, + "learning_rate": 8.098765432098766e-06, + "loss": 0.1141, + "step": 1380 + }, + { + "epoch": 2.3450586264656614, + "grad_norm": 4.100318908691406, + "learning_rate": 8.063492063492065e-06, + "loss": 0.1123, + "step": 1400 + }, + { + "epoch": 2.3785594639865995, + "grad_norm": 2.8147048950195312, + "learning_rate": 8.028218694885362e-06, + "loss": 0.1039, + "step": 1420 + }, + { + "epoch": 2.4120603015075375, + "grad_norm": 3.7376410961151123, + "learning_rate": 7.99294532627866e-06, + "loss": 0.1179, + "step": 1440 + }, + { + "epoch": 2.4455611390284755, + "grad_norm": 4.320065975189209, + "learning_rate": 7.957671957671958e-06, + "loss": 0.1167, + "step": 1460 + }, + { + "epoch": 2.4790619765494135, + "grad_norm": 4.415127277374268, + "learning_rate": 7.922398589065257e-06, + "loss": 0.125, + "step": 1480 + }, + { + "epoch": 2.5125628140703515, + "grad_norm": 2.631763219833374, + "learning_rate": 7.887125220458554e-06, + "loss": 0.1188, + "step": 1500 + }, + { + "epoch": 2.5125628140703515, + "eval_loss": 0.13635103404521942, + "eval_runtime": 1806.1099, + "eval_samples_per_second": 2.644, + "eval_steps_per_second": 0.331, + "eval_wer": 11.713753937195762, + "step": 1500 + }, + { + "epoch": 2.5460636515912896, + "grad_norm": 3.608773946762085, + "learning_rate": 7.851851851851853e-06, + "loss": 0.1222, + "step": 1520 + }, + { + "epoch": 2.5795644891122276, + "grad_norm": 3.4494569301605225, + "learning_rate": 7.816578483245151e-06, + "loss": 0.1087, + "step": 1540 + }, + { + "epoch": 2.6130653266331656, + "grad_norm": 3.6112048625946045, + "learning_rate": 7.781305114638449e-06, + "loss": 0.1097, + "step": 1560 + }, + { + "epoch": 2.6465661641541036, + "grad_norm": 3.1978707313537598, + "learning_rate": 7.746031746031747e-06, + "loss": 0.1101, + "step": 1580 + }, + { + "epoch": 2.6800670016750416, + "grad_norm": 3.6376733779907227, + "learning_rate": 7.710758377425045e-06, + "loss": 0.1056, + "step": 1600 + }, + { + "epoch": 2.7135678391959797, + "grad_norm": 3.754915237426758, + "learning_rate": 7.675485008818343e-06, + "loss": 0.1103, + "step": 1620 + }, + { + "epoch": 2.7470686767169177, + "grad_norm": 3.329240322113037, + "learning_rate": 7.64021164021164e-06, + "loss": 0.1193, + "step": 1640 + }, + { + "epoch": 2.7805695142378557, + "grad_norm": 3.9300169944763184, + "learning_rate": 7.604938271604939e-06, + "loss": 0.1232, + "step": 1660 + }, + { + "epoch": 2.8140703517587937, + "grad_norm": 4.290626049041748, + "learning_rate": 7.569664902998237e-06, + "loss": 0.1374, + "step": 1680 + }, + { + "epoch": 2.8475711892797317, + "grad_norm": 3.8987998962402344, + "learning_rate": 7.534391534391535e-06, + "loss": 0.1062, + "step": 1700 + }, + { + "epoch": 2.8810720268006698, + "grad_norm": 4.234960079193115, + "learning_rate": 7.499118165784833e-06, + "loss": 0.111, + "step": 1720 + }, + { + "epoch": 2.914572864321608, + "grad_norm": 3.103458881378174, + "learning_rate": 7.463844797178131e-06, + "loss": 0.1076, + "step": 1740 + }, + { + "epoch": 2.948073701842546, + "grad_norm": 3.270204782485962, + "learning_rate": 7.428571428571429e-06, + "loss": 0.1224, + "step": 1760 + }, + { + "epoch": 2.981574539363484, + "grad_norm": 4.261337757110596, + "learning_rate": 7.393298059964727e-06, + "loss": 0.1224, + "step": 1780 + }, + { + "epoch": 3.0150753768844223, + "grad_norm": 1.9531301259994507, + "learning_rate": 7.358024691358025e-06, + "loss": 0.0913, + "step": 1800 + }, + { + "epoch": 3.0485762144053603, + "grad_norm": 1.866215467453003, + "learning_rate": 7.322751322751324e-06, + "loss": 0.0466, + "step": 1820 + }, + { + "epoch": 3.0820770519262983, + "grad_norm": 3.9456610679626465, + "learning_rate": 7.287477954144622e-06, + "loss": 0.0451, + "step": 1840 + }, + { + "epoch": 3.1155778894472363, + "grad_norm": 1.9518849849700928, + "learning_rate": 7.25220458553792e-06, + "loss": 0.0414, + "step": 1860 + }, + { + "epoch": 3.1490787269681744, + "grad_norm": 2.232792854309082, + "learning_rate": 7.216931216931218e-06, + "loss": 0.0502, + "step": 1880 + }, + { + "epoch": 3.1825795644891124, + "grad_norm": 2.074127674102783, + "learning_rate": 7.181657848324516e-06, + "loss": 0.057, + "step": 1900 + }, + { + "epoch": 3.2160804020100504, + "grad_norm": 3.5036733150482178, + "learning_rate": 7.146384479717814e-06, + "loss": 0.0515, + "step": 1920 + }, + { + "epoch": 3.2495812395309884, + "grad_norm": 2.324014186859131, + "learning_rate": 7.111111111111112e-06, + "loss": 0.0454, + "step": 1940 + }, + { + "epoch": 3.2830820770519265, + "grad_norm": 2.710326671600342, + "learning_rate": 7.07583774250441e-06, + "loss": 0.0473, + "step": 1960 + }, + { + "epoch": 3.3165829145728645, + "grad_norm": 2.8943896293640137, + "learning_rate": 7.040564373897708e-06, + "loss": 0.046, + "step": 1980 + }, + { + "epoch": 3.3500837520938025, + "grad_norm": 2.319986581802368, + "learning_rate": 7.005291005291006e-06, + "loss": 0.0449, + "step": 2000 + }, + { + "epoch": 3.3500837520938025, + "eval_loss": 0.10934468358755112, + "eval_runtime": 1762.4506, + "eval_samples_per_second": 2.71, + "eval_steps_per_second": 0.339, + "eval_wer": 7.172854824854443, + "step": 2000 + }, + { + "epoch": 3.3835845896147405, + "grad_norm": 3.317129135131836, + "learning_rate": 6.9700176366843046e-06, + "loss": 0.0645, + "step": 2020 + }, + { + "epoch": 3.4170854271356785, + "grad_norm": 1.9533768892288208, + "learning_rate": 6.9347442680776025e-06, + "loss": 0.0432, + "step": 2040 + }, + { + "epoch": 3.4505862646566166, + "grad_norm": 2.0508453845977783, + "learning_rate": 6.8994708994709005e-06, + "loss": 0.0521, + "step": 2060 + }, + { + "epoch": 3.4840871021775546, + "grad_norm": 2.163236141204834, + "learning_rate": 6.8641975308641985e-06, + "loss": 0.0529, + "step": 2080 + }, + { + "epoch": 3.5175879396984926, + "grad_norm": 2.7154581546783447, + "learning_rate": 6.8289241622574965e-06, + "loss": 0.0452, + "step": 2100 + }, + { + "epoch": 3.5510887772194306, + "grad_norm": 3.0822432041168213, + "learning_rate": 6.7936507936507944e-06, + "loss": 0.0546, + "step": 2120 + }, + { + "epoch": 3.5845896147403686, + "grad_norm": 4.19010591506958, + "learning_rate": 6.758377425044092e-06, + "loss": 0.0529, + "step": 2140 + }, + { + "epoch": 3.6180904522613067, + "grad_norm": 2.9883594512939453, + "learning_rate": 6.72310405643739e-06, + "loss": 0.0503, + "step": 2160 + }, + { + "epoch": 3.6515912897822447, + "grad_norm": 2.3664371967315674, + "learning_rate": 6.687830687830688e-06, + "loss": 0.0498, + "step": 2180 + }, + { + "epoch": 3.6850921273031827, + "grad_norm": 2.0549991130828857, + "learning_rate": 6.652557319223986e-06, + "loss": 0.051, + "step": 2200 + }, + { + "epoch": 3.7185929648241207, + "grad_norm": 2.5339038372039795, + "learning_rate": 6.617283950617285e-06, + "loss": 0.0568, + "step": 2220 + }, + { + "epoch": 3.7520938023450587, + "grad_norm": 1.9988099336624146, + "learning_rate": 6.582010582010583e-06, + "loss": 0.051, + "step": 2240 + }, + { + "epoch": 3.7855946398659968, + "grad_norm": 2.5243782997131348, + "learning_rate": 6.546737213403881e-06, + "loss": 0.056, + "step": 2260 + }, + { + "epoch": 3.819095477386935, + "grad_norm": 3.157158136367798, + "learning_rate": 6.511463844797179e-06, + "loss": 0.0497, + "step": 2280 + }, + { + "epoch": 3.852596314907873, + "grad_norm": 1.9286202192306519, + "learning_rate": 6.476190476190477e-06, + "loss": 0.0426, + "step": 2300 + }, + { + "epoch": 3.886097152428811, + "grad_norm": 3.808802604675293, + "learning_rate": 6.440917107583775e-06, + "loss": 0.0499, + "step": 2320 + }, + { + "epoch": 3.919597989949749, + "grad_norm": 2.506671667098999, + "learning_rate": 6.405643738977073e-06, + "loss": 0.052, + "step": 2340 + }, + { + "epoch": 3.953098827470687, + "grad_norm": 2.9451920986175537, + "learning_rate": 6.370370370370371e-06, + "loss": 0.0552, + "step": 2360 + }, + { + "epoch": 3.986599664991625, + "grad_norm": 2.592744827270508, + "learning_rate": 6.335097001763669e-06, + "loss": 0.0527, + "step": 2380 + }, + { + "epoch": 4.0201005025125625, + "grad_norm": 1.8891575336456299, + "learning_rate": 6.299823633156967e-06, + "loss": 0.0289, + "step": 2400 + }, + { + "epoch": 4.0536013400335005, + "grad_norm": 1.8053243160247803, + "learning_rate": 6.264550264550266e-06, + "loss": 0.0192, + "step": 2420 + }, + { + "epoch": 4.0871021775544385, + "grad_norm": 2.0084407329559326, + "learning_rate": 6.229276895943564e-06, + "loss": 0.0242, + "step": 2440 + }, + { + "epoch": 4.1206030150753765, + "grad_norm": 1.5919119119644165, + "learning_rate": 6.194003527336862e-06, + "loss": 0.0211, + "step": 2460 + }, + { + "epoch": 4.1541038525963145, + "grad_norm": 1.9214613437652588, + "learning_rate": 6.15873015873016e-06, + "loss": 0.0233, + "step": 2480 + }, + { + "epoch": 4.187604690117253, + "grad_norm": 1.2652311325073242, + "learning_rate": 6.123456790123458e-06, + "loss": 0.0199, + "step": 2500 + }, + { + "epoch": 4.187604690117253, + "eval_loss": 0.0981329157948494, + "eval_runtime": 1779.0213, + "eval_samples_per_second": 2.685, + "eval_steps_per_second": 0.336, + "eval_wer": 6.707549871146321, + "step": 2500 + }, + { + "epoch": 4.221105527638191, + "grad_norm": 2.026528835296631, + "learning_rate": 6.088183421516756e-06, + "loss": 0.0217, + "step": 2520 + }, + { + "epoch": 4.254606365159129, + "grad_norm": 1.596919059753418, + "learning_rate": 6.052910052910054e-06, + "loss": 0.0167, + "step": 2540 + }, + { + "epoch": 4.288107202680067, + "grad_norm": 2.9445090293884277, + "learning_rate": 6.017636684303352e-06, + "loss": 0.0225, + "step": 2560 + }, + { + "epoch": 4.321608040201005, + "grad_norm": 2.4160282611846924, + "learning_rate": 5.9823633156966496e-06, + "loss": 0.0253, + "step": 2580 + }, + { + "epoch": 4.355108877721943, + "grad_norm": 1.461127758026123, + "learning_rate": 5.9470899470899475e-06, + "loss": 0.0197, + "step": 2600 + }, + { + "epoch": 4.388609715242881, + "grad_norm": 2.7892863750457764, + "learning_rate": 5.911816578483246e-06, + "loss": 0.022, + "step": 2620 + }, + { + "epoch": 4.422110552763819, + "grad_norm": 1.651208758354187, + "learning_rate": 5.876543209876544e-06, + "loss": 0.0215, + "step": 2640 + }, + { + "epoch": 4.455611390284757, + "grad_norm": 2.2500391006469727, + "learning_rate": 5.841269841269842e-06, + "loss": 0.0247, + "step": 2660 + }, + { + "epoch": 4.489112227805695, + "grad_norm": 4.447635173797607, + "learning_rate": 5.80599647266314e-06, + "loss": 0.0263, + "step": 2680 + }, + { + "epoch": 4.522613065326633, + "grad_norm": 0.8300407528877258, + "learning_rate": 5.770723104056438e-06, + "loss": 0.0209, + "step": 2700 + }, + { + "epoch": 4.556113902847571, + "grad_norm": 1.6874111890792847, + "learning_rate": 5.735449735449736e-06, + "loss": 0.0195, + "step": 2720 + }, + { + "epoch": 4.589614740368509, + "grad_norm": 2.4045815467834473, + "learning_rate": 5.700176366843034e-06, + "loss": 0.0224, + "step": 2740 + }, + { + "epoch": 4.623115577889447, + "grad_norm": 2.3160908222198486, + "learning_rate": 5.664902998236332e-06, + "loss": 0.0179, + "step": 2760 + }, + { + "epoch": 4.656616415410385, + "grad_norm": 1.6684287786483765, + "learning_rate": 5.62962962962963e-06, + "loss": 0.0238, + "step": 2780 + }, + { + "epoch": 4.690117252931323, + "grad_norm": 1.973906397819519, + "learning_rate": 5.594356261022928e-06, + "loss": 0.0226, + "step": 2800 + }, + { + "epoch": 4.723618090452261, + "grad_norm": 2.270906686782837, + "learning_rate": 5.559082892416227e-06, + "loss": 0.0232, + "step": 2820 + }, + { + "epoch": 4.757118927973199, + "grad_norm": 1.8875011205673218, + "learning_rate": 5.523809523809525e-06, + "loss": 0.0218, + "step": 2840 + }, + { + "epoch": 4.790619765494137, + "grad_norm": 1.1312583684921265, + "learning_rate": 5.488536155202823e-06, + "loss": 0.0212, + "step": 2860 + }, + { + "epoch": 4.824120603015075, + "grad_norm": 0.864783525466919, + "learning_rate": 5.453262786596121e-06, + "loss": 0.023, + "step": 2880 + }, + { + "epoch": 4.857621440536013, + "grad_norm": 1.2935965061187744, + "learning_rate": 5.417989417989419e-06, + "loss": 0.019, + "step": 2900 + }, + { + "epoch": 4.891122278056951, + "grad_norm": 2.4576382637023926, + "learning_rate": 5.382716049382717e-06, + "loss": 0.0193, + "step": 2920 + }, + { + "epoch": 4.924623115577889, + "grad_norm": 2.71472430229187, + "learning_rate": 5.347442680776015e-06, + "loss": 0.0253, + "step": 2940 + }, + { + "epoch": 4.958123953098827, + "grad_norm": 2.84940505027771, + "learning_rate": 5.312169312169313e-06, + "loss": 0.0218, + "step": 2960 + }, + { + "epoch": 4.991624790619765, + "grad_norm": 1.8483999967575073, + "learning_rate": 5.276895943562611e-06, + "loss": 0.0226, + "step": 2980 + }, + { + "epoch": 5.025125628140704, + "grad_norm": 0.6126876473426819, + "learning_rate": 5.241622574955909e-06, + "loss": 0.0101, + "step": 3000 + }, + { + "epoch": 5.025125628140704, + "eval_loss": 0.09390027821063995, + "eval_runtime": 1767.4512, + "eval_samples_per_second": 2.702, + "eval_steps_per_second": 0.338, + "eval_wer": 5.500143170754987, + "step": 3000 + }, + { + "epoch": 5.058626465661642, + "grad_norm": 0.3711394965648651, + "learning_rate": 5.2063492063492076e-06, + "loss": 0.0119, + "step": 3020 + }, + { + "epoch": 5.09212730318258, + "grad_norm": 0.5738839507102966, + "learning_rate": 5.1710758377425055e-06, + "loss": 0.0086, + "step": 3040 + }, + { + "epoch": 5.125628140703517, + "grad_norm": 0.7609245777130127, + "learning_rate": 5.1358024691358035e-06, + "loss": 0.0093, + "step": 3060 + }, + { + "epoch": 5.159128978224456, + "grad_norm": 1.2764722108840942, + "learning_rate": 5.1005291005291015e-06, + "loss": 0.0111, + "step": 3080 + }, + { + "epoch": 5.192629815745394, + "grad_norm": 2.1169776916503906, + "learning_rate": 5.0652557319223995e-06, + "loss": 0.0107, + "step": 3100 + }, + { + "epoch": 5.226130653266332, + "grad_norm": 2.1893081665039062, + "learning_rate": 5.0299823633156974e-06, + "loss": 0.0097, + "step": 3120 + }, + { + "epoch": 5.259631490787269, + "grad_norm": 2.2419638633728027, + "learning_rate": 4.9947089947089946e-06, + "loss": 0.0098, + "step": 3140 + }, + { + "epoch": 5.293132328308207, + "grad_norm": 0.6479611992835999, + "learning_rate": 4.959435626102293e-06, + "loss": 0.0082, + "step": 3160 + }, + { + "epoch": 5.326633165829146, + "grad_norm": 0.4799642860889435, + "learning_rate": 4.924162257495591e-06, + "loss": 0.0119, + "step": 3180 + }, + { + "epoch": 5.360134003350084, + "grad_norm": 0.7716453075408936, + "learning_rate": 4.888888888888889e-06, + "loss": 0.0086, + "step": 3200 + }, + { + "epoch": 5.393634840871022, + "grad_norm": 1.2303547859191895, + "learning_rate": 4.853615520282187e-06, + "loss": 0.009, + "step": 3220 + }, + { + "epoch": 5.42713567839196, + "grad_norm": 0.6345349550247192, + "learning_rate": 4.818342151675485e-06, + "loss": 0.0085, + "step": 3240 + }, + { + "epoch": 5.460636515912898, + "grad_norm": 0.9741530418395996, + "learning_rate": 4.783068783068783e-06, + "loss": 0.0082, + "step": 3260 + }, + { + "epoch": 5.494137353433836, + "grad_norm": 1.1631624698638916, + "learning_rate": 4.747795414462081e-06, + "loss": 0.0086, + "step": 3280 + }, + { + "epoch": 5.527638190954773, + "grad_norm": 0.6502953767776489, + "learning_rate": 4.712522045855379e-06, + "loss": 0.0093, + "step": 3300 + }, + { + "epoch": 5.561139028475711, + "grad_norm": 0.7464337348937988, + "learning_rate": 4.677248677248677e-06, + "loss": 0.0086, + "step": 3320 + }, + { + "epoch": 5.594639865996649, + "grad_norm": 1.017751693725586, + "learning_rate": 4.641975308641975e-06, + "loss": 0.0091, + "step": 3340 + }, + { + "epoch": 5.628140703517588, + "grad_norm": 0.4273395836353302, + "learning_rate": 4.606701940035274e-06, + "loss": 0.0078, + "step": 3360 + }, + { + "epoch": 5.661641541038526, + "grad_norm": 0.6737497448921204, + "learning_rate": 4.571428571428572e-06, + "loss": 0.0106, + "step": 3380 + }, + { + "epoch": 5.695142378559464, + "grad_norm": 1.0791343450546265, + "learning_rate": 4.53615520282187e-06, + "loss": 0.0097, + "step": 3400 + }, + { + "epoch": 5.728643216080402, + "grad_norm": 1.0891772508621216, + "learning_rate": 4.500881834215168e-06, + "loss": 0.0109, + "step": 3420 + }, + { + "epoch": 5.76214405360134, + "grad_norm": 0.7465157508850098, + "learning_rate": 4.465608465608466e-06, + "loss": 0.0078, + "step": 3440 + }, + { + "epoch": 5.795644891122278, + "grad_norm": 0.7693866491317749, + "learning_rate": 4.430335097001764e-06, + "loss": 0.009, + "step": 3460 + }, + { + "epoch": 5.8291457286432165, + "grad_norm": 1.3295698165893555, + "learning_rate": 4.395061728395062e-06, + "loss": 0.016, + "step": 3480 + }, + { + "epoch": 5.8626465661641545, + "grad_norm": 2.4605352878570557, + "learning_rate": 4.35978835978836e-06, + "loss": 0.0081, + "step": 3500 + }, + { + "epoch": 5.8626465661641545, + "eval_loss": 0.09377142041921616, + "eval_runtime": 1773.01, + "eval_samples_per_second": 2.694, + "eval_steps_per_second": 0.337, + "eval_wer": 5.010976424549012, + "step": 3500 + } + ], + "logging_steps": 20, + "max_steps": 5970, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "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": 3.23172360548352e+19, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-3500/training_args.bin b/checkpoint-3500/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6ffec31ca17b32691fb4cd1278f73919f3259cf4 --- /dev/null +++ b/checkpoint-3500/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41a3c645624f7463029286b7d5e1ad6924cfc44a0d153880431b77c40e5deb73 +size 5432 diff --git a/checkpoint-4000/config.json b/checkpoint-4000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..337642e18341c758364c32c6b6159a090615329b --- /dev/null +++ b/checkpoint-4000/config.json @@ -0,0 +1,47 @@ +{ + "activation_dropout": 0.0, + "activation_function": "gelu", + "apply_spec_augment": false, + "architectures": [ + "WhisperForConditionalGeneration" + ], + "attention_dropout": 0.0, + "begin_suppress_tokens": null, + "bos_token_id": 50257, + "classifier_proj_size": 256, + "d_model": 768, + "decoder_attention_heads": 12, + "decoder_ffn_dim": 3072, + "decoder_layerdrop": 0.0, + "decoder_layers": 12, + "decoder_start_token_id": 50258, + "dropout": 0.0, + "encoder_attention_heads": 12, + "encoder_ffn_dim": 3072, + "encoder_layerdrop": 0.0, + "encoder_layers": 12, + "eos_token_id": 50257, + "forced_decoder_ids": null, + "init_std": 0.02, + "is_encoder_decoder": true, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "max_length": null, + "max_source_positions": 1500, + "max_target_positions": 448, + "median_filter_width": 7, + "model_type": "whisper", + "num_hidden_layers": 12, + "num_mel_bins": 80, + "pad_token_id": 50257, + "scale_embedding": false, + "torch_dtype": "float32", + "transformers_version": "4.51.3", + "use_cache": false, + "use_weighted_layer_sum": false, + "vocab_size": 51865 +} diff --git a/checkpoint-4000/generation_config.json b/checkpoint-4000/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..e1f4120825a303544ca23708d9dc52bbaa1dce16 --- /dev/null +++ b/checkpoint-4000/generation_config.json @@ -0,0 +1,9 @@ +{ + "bos_token_id": 50257, + "decoder_start_token_id": 50258, + "eos_token_id": 50257, + "pad_token_id": 50257, + "suppress_tokens": [], + "transformers_version": "4.51.3", + "use_cache": false +} diff --git a/checkpoint-4000/model.safetensors b/checkpoint-4000/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..3f0c7aa887d3bce9cd2cc5eaccae58553d1348fd --- /dev/null +++ b/checkpoint-4000/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:91acc97120660c95a65a2b60d76486ac64d39536e98f53518052435e27911e34 +size 966995080 diff --git a/checkpoint-4000/optimizer.pt b/checkpoint-4000/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..cc29b75dad3a51db4ba30a1f6783c4b6374ef1c0 --- /dev/null +++ b/checkpoint-4000/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6320c0154169b61d2e901ffdea9eecef62c8ba3439357ad4212021392e0cb78b +size 1925064044 diff --git a/checkpoint-4000/preprocessor_config.json b/checkpoint-4000/preprocessor_config.json new file mode 100644 index 0000000000000000000000000000000000000000..d7c5b1dfd466c815664b16139e8f605620d22e4e --- /dev/null +++ b/checkpoint-4000/preprocessor_config.json @@ -0,0 +1,15 @@ +{ + "chunk_length": 30, + "dither": 0.0, + "feature_extractor_type": "WhisperFeatureExtractor", + "feature_size": 80, + "hop_length": 160, + "n_fft": 400, + "n_samples": 480000, + "nb_max_frames": 3000, + "padding_side": "right", + "padding_value": 0.0, + "processor_class": "WhisperProcessor", + "return_attention_mask": false, + "sampling_rate": 16000 +} diff --git a/checkpoint-4000/rng_state.pth b/checkpoint-4000/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..f0b60772ee8ee326dcbb04fae8f625f8d651aa9e --- /dev/null +++ b/checkpoint-4000/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5cd4ed764dbf7b86dcf50e3fceadc0673152ad003e766806dfb8d6615b39e6a0 +size 14244 diff --git a/checkpoint-4000/scaler.pt b/checkpoint-4000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..de752266246eb104f98c67d40caaf4d48a8acb38 --- /dev/null +++ b/checkpoint-4000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2088bf7a1757be905776a5a478c342182ece5ad2b8f127ed6e453d8c07616d2e +size 988 diff --git a/checkpoint-4000/scheduler.pt b/checkpoint-4000/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..3e077cb91ed1d017b00fd5844bda7068088c2a43 --- /dev/null +++ b/checkpoint-4000/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6607e05b3b7479ae4e17a862379eeb725e14b09750bd36b15086aa6da4ef497 +size 1064 diff --git a/checkpoint-4000/trainer_state.json b/checkpoint-4000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..d1eff0238ee518c3206ba9087012e1a755a8720b --- /dev/null +++ b/checkpoint-4000/trainer_state.json @@ -0,0 +1,1506 @@ +{ + "best_global_step": 4000, + "best_metric": 4.89166746205975, + "best_model_checkpoint": "./SALAMA_C3/checkpoint-4000", + "epoch": 6.700167504187605, + "eval_steps": 500, + "global_step": 4000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.03350083752093802, + "grad_norm": 11.757207870483398, + "learning_rate": 6.333333333333334e-07, + "loss": 0.5112, + "step": 20 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 10.197953224182129, + "learning_rate": 1.3e-06, + "loss": 0.5283, + "step": 40 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 12.811074256896973, + "learning_rate": 1.9666666666666668e-06, + "loss": 0.474, + "step": 60 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 8.559342384338379, + "learning_rate": 2.6e-06, + "loss": 0.424, + "step": 80 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 6.9629225730896, + "learning_rate": 3.266666666666667e-06, + "loss": 0.4465, + "step": 100 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 7.833058834075928, + "learning_rate": 3.9333333333333335e-06, + "loss": 0.3764, + "step": 120 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 6.882424831390381, + "learning_rate": 4.600000000000001e-06, + "loss": 0.3705, + "step": 140 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 5.4183244705200195, + "learning_rate": 5.2666666666666665e-06, + "loss": 0.4116, + "step": 160 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 6.103787899017334, + "learning_rate": 5.933333333333335e-06, + "loss": 0.3903, + "step": 180 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 6.084160804748535, + "learning_rate": 6.600000000000001e-06, + "loss": 0.3794, + "step": 200 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 4.531406879425049, + "learning_rate": 7.266666666666668e-06, + "loss": 0.3904, + "step": 220 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 6.702854156494141, + "learning_rate": 7.933333333333334e-06, + "loss": 0.3672, + "step": 240 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 6.386377811431885, + "learning_rate": 8.6e-06, + "loss": 0.3379, + "step": 260 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 5.33281135559082, + "learning_rate": 9.266666666666667e-06, + "loss": 0.4102, + "step": 280 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 6.031435012817383, + "learning_rate": 9.933333333333334e-06, + "loss": 0.3815, + "step": 300 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 4.8638176918029785, + "learning_rate": 9.968253968253969e-06, + "loss": 0.358, + "step": 320 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 5.352113723754883, + "learning_rate": 9.932980599647268e-06, + "loss": 0.3578, + "step": 340 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 5.560739040374756, + "learning_rate": 9.897707231040565e-06, + "loss": 0.3795, + "step": 360 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 6.186940670013428, + "learning_rate": 9.862433862433864e-06, + "loss": 0.3852, + "step": 380 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 5.37507438659668, + "learning_rate": 9.827160493827161e-06, + "loss": 0.3761, + "step": 400 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 6.373449802398682, + "learning_rate": 9.79188712522046e-06, + "loss": 0.3674, + "step": 420 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 5.702625274658203, + "learning_rate": 9.756613756613757e-06, + "loss": 0.4013, + "step": 440 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 5.5675153732299805, + "learning_rate": 9.721340388007056e-06, + "loss": 0.3495, + "step": 460 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 6.296374320983887, + "learning_rate": 9.686067019400353e-06, + "loss": 0.3704, + "step": 480 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 4.830463886260986, + "learning_rate": 9.650793650793652e-06, + "loss": 0.36, + "step": 500 + }, + { + "epoch": 0.8375209380234506, + "eval_loss": 0.27764827013015747, + "eval_runtime": 1752.3826, + "eval_samples_per_second": 2.725, + "eval_steps_per_second": 0.341, + "eval_wer": 19.59291781998664, + "step": 500 + }, + { + "epoch": 0.8710217755443886, + "grad_norm": 4.863058090209961, + "learning_rate": 9.61552028218695e-06, + "loss": 0.3874, + "step": 520 + }, + { + "epoch": 0.9045226130653267, + "grad_norm": 6.121027946472168, + "learning_rate": 9.580246913580248e-06, + "loss": 0.382, + "step": 540 + }, + { + "epoch": 0.9380234505862647, + "grad_norm": 4.538082599639893, + "learning_rate": 9.544973544973546e-06, + "loss": 0.3611, + "step": 560 + }, + { + "epoch": 0.9715242881072027, + "grad_norm": 4.8772382736206055, + "learning_rate": 9.509700176366844e-06, + "loss": 0.3584, + "step": 580 + }, + { + "epoch": 1.0050251256281406, + "grad_norm": 4.039212703704834, + "learning_rate": 9.474426807760142e-06, + "loss": 0.3701, + "step": 600 + }, + { + "epoch": 1.0385259631490786, + "grad_norm": 4.787687301635742, + "learning_rate": 9.43915343915344e-06, + "loss": 0.2058, + "step": 620 + }, + { + "epoch": 1.0720268006700167, + "grad_norm": 4.503021717071533, + "learning_rate": 9.403880070546738e-06, + "loss": 0.2455, + "step": 640 + }, + { + "epoch": 1.1055276381909547, + "grad_norm": 6.663857936859131, + "learning_rate": 9.368606701940036e-06, + "loss": 0.2179, + "step": 660 + }, + { + "epoch": 1.1390284757118927, + "grad_norm": 4.486196041107178, + "learning_rate": 9.333333333333334e-06, + "loss": 0.2196, + "step": 680 + }, + { + "epoch": 1.1725293132328307, + "grad_norm": 4.667060852050781, + "learning_rate": 9.298059964726633e-06, + "loss": 0.2183, + "step": 700 + }, + { + "epoch": 1.2060301507537687, + "grad_norm": 5.608316898345947, + "learning_rate": 9.26278659611993e-06, + "loss": 0.2161, + "step": 720 + }, + { + "epoch": 1.2395309882747068, + "grad_norm": 4.2184271812438965, + "learning_rate": 9.227513227513229e-06, + "loss": 0.2382, + "step": 740 + }, + { + "epoch": 1.2730318257956448, + "grad_norm": 3.871945381164551, + "learning_rate": 9.192239858906526e-06, + "loss": 0.2214, + "step": 760 + }, + { + "epoch": 1.3065326633165828, + "grad_norm": 3.730222225189209, + "learning_rate": 9.156966490299825e-06, + "loss": 0.2213, + "step": 780 + }, + { + "epoch": 1.3400335008375208, + "grad_norm": 4.740777015686035, + "learning_rate": 9.121693121693122e-06, + "loss": 0.2503, + "step": 800 + }, + { + "epoch": 1.3735343383584588, + "grad_norm": 4.123469352722168, + "learning_rate": 9.086419753086421e-06, + "loss": 0.2443, + "step": 820 + }, + { + "epoch": 1.4070351758793969, + "grad_norm": 3.988917827606201, + "learning_rate": 9.051146384479718e-06, + "loss": 0.241, + "step": 840 + }, + { + "epoch": 1.4405360134003349, + "grad_norm": 5.4730305671691895, + "learning_rate": 9.015873015873017e-06, + "loss": 0.2163, + "step": 860 + }, + { + "epoch": 1.474036850921273, + "grad_norm": 5.020652770996094, + "learning_rate": 8.980599647266314e-06, + "loss": 0.2387, + "step": 880 + }, + { + "epoch": 1.507537688442211, + "grad_norm": 4.231486797332764, + "learning_rate": 8.945326278659613e-06, + "loss": 0.2523, + "step": 900 + }, + { + "epoch": 1.541038525963149, + "grad_norm": 6.197975158691406, + "learning_rate": 8.910052910052912e-06, + "loss": 0.2255, + "step": 920 + }, + { + "epoch": 1.574539363484087, + "grad_norm": 5.489210605621338, + "learning_rate": 8.874779541446209e-06, + "loss": 0.2397, + "step": 940 + }, + { + "epoch": 1.608040201005025, + "grad_norm": 4.32737398147583, + "learning_rate": 8.839506172839508e-06, + "loss": 0.2298, + "step": 960 + }, + { + "epoch": 1.641541038525963, + "grad_norm": 4.504214763641357, + "learning_rate": 8.804232804232805e-06, + "loss": 0.2301, + "step": 980 + }, + { + "epoch": 1.675041876046901, + "grad_norm": 3.9694759845733643, + "learning_rate": 8.768959435626104e-06, + "loss": 0.2338, + "step": 1000 + }, + { + "epoch": 1.675041876046901, + "eval_loss": 0.18838582932949066, + "eval_runtime": 1770.3449, + "eval_samples_per_second": 2.698, + "eval_steps_per_second": 0.337, + "eval_wer": 14.071299035983584, + "step": 1000 + }, + { + "epoch": 1.708542713567839, + "grad_norm": 4.972934722900391, + "learning_rate": 8.7336860670194e-06, + "loss": 0.2388, + "step": 1020 + }, + { + "epoch": 1.742043551088777, + "grad_norm": 5.178994655609131, + "learning_rate": 8.6984126984127e-06, + "loss": 0.2531, + "step": 1040 + }, + { + "epoch": 1.775544388609715, + "grad_norm": 3.560372829437256, + "learning_rate": 8.663139329805997e-06, + "loss": 0.2187, + "step": 1060 + }, + { + "epoch": 1.809045226130653, + "grad_norm": 4.647324562072754, + "learning_rate": 8.627865961199296e-06, + "loss": 0.2413, + "step": 1080 + }, + { + "epoch": 1.8425460636515911, + "grad_norm": 3.6784422397613525, + "learning_rate": 8.592592592592593e-06, + "loss": 0.2242, + "step": 1100 + }, + { + "epoch": 1.8760469011725294, + "grad_norm": 5.928018093109131, + "learning_rate": 8.557319223985891e-06, + "loss": 0.2406, + "step": 1120 + }, + { + "epoch": 1.9095477386934674, + "grad_norm": 6.358487606048584, + "learning_rate": 8.52204585537919e-06, + "loss": 0.2482, + "step": 1140 + }, + { + "epoch": 1.9430485762144054, + "grad_norm": 4.409506797790527, + "learning_rate": 8.486772486772487e-06, + "loss": 0.2415, + "step": 1160 + }, + { + "epoch": 1.9765494137353434, + "grad_norm": 5.169639587402344, + "learning_rate": 8.451499118165786e-06, + "loss": 0.222, + "step": 1180 + }, + { + "epoch": 2.0100502512562812, + "grad_norm": 2.8841328620910645, + "learning_rate": 8.416225749559083e-06, + "loss": 0.2109, + "step": 1200 + }, + { + "epoch": 2.0435510887772192, + "grad_norm": 3.835498332977295, + "learning_rate": 8.380952380952382e-06, + "loss": 0.1284, + "step": 1220 + }, + { + "epoch": 2.0770519262981573, + "grad_norm": 2.584859848022461, + "learning_rate": 8.34567901234568e-06, + "loss": 0.1059, + "step": 1240 + }, + { + "epoch": 2.1105527638190953, + "grad_norm": 2.797682285308838, + "learning_rate": 8.310405643738978e-06, + "loss": 0.0967, + "step": 1260 + }, + { + "epoch": 2.1440536013400333, + "grad_norm": 3.2937309741973877, + "learning_rate": 8.275132275132275e-06, + "loss": 0.1267, + "step": 1280 + }, + { + "epoch": 2.1775544388609713, + "grad_norm": 3.344325065612793, + "learning_rate": 8.239858906525574e-06, + "loss": 0.1354, + "step": 1300 + }, + { + "epoch": 2.2110552763819094, + "grad_norm": 2.955522060394287, + "learning_rate": 8.204585537918873e-06, + "loss": 0.104, + "step": 1320 + }, + { + "epoch": 2.2445561139028474, + "grad_norm": 5.180908203125, + "learning_rate": 8.16931216931217e-06, + "loss": 0.1229, + "step": 1340 + }, + { + "epoch": 2.2780569514237854, + "grad_norm": 3.639417886734009, + "learning_rate": 8.134038800705469e-06, + "loss": 0.104, + "step": 1360 + }, + { + "epoch": 2.3115577889447234, + "grad_norm": 3.7380902767181396, + "learning_rate": 8.098765432098766e-06, + "loss": 0.1141, + "step": 1380 + }, + { + "epoch": 2.3450586264656614, + "grad_norm": 4.100318908691406, + "learning_rate": 8.063492063492065e-06, + "loss": 0.1123, + "step": 1400 + }, + { + "epoch": 2.3785594639865995, + "grad_norm": 2.8147048950195312, + "learning_rate": 8.028218694885362e-06, + "loss": 0.1039, + "step": 1420 + }, + { + "epoch": 2.4120603015075375, + "grad_norm": 3.7376410961151123, + "learning_rate": 7.99294532627866e-06, + "loss": 0.1179, + "step": 1440 + }, + { + "epoch": 2.4455611390284755, + "grad_norm": 4.320065975189209, + "learning_rate": 7.957671957671958e-06, + "loss": 0.1167, + "step": 1460 + }, + { + "epoch": 2.4790619765494135, + "grad_norm": 4.415127277374268, + "learning_rate": 7.922398589065257e-06, + "loss": 0.125, + "step": 1480 + }, + { + "epoch": 2.5125628140703515, + "grad_norm": 2.631763219833374, + "learning_rate": 7.887125220458554e-06, + "loss": 0.1188, + "step": 1500 + }, + { + "epoch": 2.5125628140703515, + "eval_loss": 0.13635103404521942, + "eval_runtime": 1806.1099, + "eval_samples_per_second": 2.644, + "eval_steps_per_second": 0.331, + "eval_wer": 11.713753937195762, + "step": 1500 + }, + { + "epoch": 2.5460636515912896, + "grad_norm": 3.608773946762085, + "learning_rate": 7.851851851851853e-06, + "loss": 0.1222, + "step": 1520 + }, + { + "epoch": 2.5795644891122276, + "grad_norm": 3.4494569301605225, + "learning_rate": 7.816578483245151e-06, + "loss": 0.1087, + "step": 1540 + }, + { + "epoch": 2.6130653266331656, + "grad_norm": 3.6112048625946045, + "learning_rate": 7.781305114638449e-06, + "loss": 0.1097, + "step": 1560 + }, + { + "epoch": 2.6465661641541036, + "grad_norm": 3.1978707313537598, + "learning_rate": 7.746031746031747e-06, + "loss": 0.1101, + "step": 1580 + }, + { + "epoch": 2.6800670016750416, + "grad_norm": 3.6376733779907227, + "learning_rate": 7.710758377425045e-06, + "loss": 0.1056, + "step": 1600 + }, + { + "epoch": 2.7135678391959797, + "grad_norm": 3.754915237426758, + "learning_rate": 7.675485008818343e-06, + "loss": 0.1103, + "step": 1620 + }, + { + "epoch": 2.7470686767169177, + "grad_norm": 3.329240322113037, + "learning_rate": 7.64021164021164e-06, + "loss": 0.1193, + "step": 1640 + }, + { + "epoch": 2.7805695142378557, + "grad_norm": 3.9300169944763184, + "learning_rate": 7.604938271604939e-06, + "loss": 0.1232, + "step": 1660 + }, + { + "epoch": 2.8140703517587937, + "grad_norm": 4.290626049041748, + "learning_rate": 7.569664902998237e-06, + "loss": 0.1374, + "step": 1680 + }, + { + "epoch": 2.8475711892797317, + "grad_norm": 3.8987998962402344, + "learning_rate": 7.534391534391535e-06, + "loss": 0.1062, + "step": 1700 + }, + { + "epoch": 2.8810720268006698, + "grad_norm": 4.234960079193115, + "learning_rate": 7.499118165784833e-06, + "loss": 0.111, + "step": 1720 + }, + { + "epoch": 2.914572864321608, + "grad_norm": 3.103458881378174, + "learning_rate": 7.463844797178131e-06, + "loss": 0.1076, + "step": 1740 + }, + { + "epoch": 2.948073701842546, + "grad_norm": 3.270204782485962, + "learning_rate": 7.428571428571429e-06, + "loss": 0.1224, + "step": 1760 + }, + { + "epoch": 2.981574539363484, + "grad_norm": 4.261337757110596, + "learning_rate": 7.393298059964727e-06, + "loss": 0.1224, + "step": 1780 + }, + { + "epoch": 3.0150753768844223, + "grad_norm": 1.9531301259994507, + "learning_rate": 7.358024691358025e-06, + "loss": 0.0913, + "step": 1800 + }, + { + "epoch": 3.0485762144053603, + "grad_norm": 1.866215467453003, + "learning_rate": 7.322751322751324e-06, + "loss": 0.0466, + "step": 1820 + }, + { + "epoch": 3.0820770519262983, + "grad_norm": 3.9456610679626465, + "learning_rate": 7.287477954144622e-06, + "loss": 0.0451, + "step": 1840 + }, + { + "epoch": 3.1155778894472363, + "grad_norm": 1.9518849849700928, + "learning_rate": 7.25220458553792e-06, + "loss": 0.0414, + "step": 1860 + }, + { + "epoch": 3.1490787269681744, + "grad_norm": 2.232792854309082, + "learning_rate": 7.216931216931218e-06, + "loss": 0.0502, + "step": 1880 + }, + { + "epoch": 3.1825795644891124, + "grad_norm": 2.074127674102783, + "learning_rate": 7.181657848324516e-06, + "loss": 0.057, + "step": 1900 + }, + { + "epoch": 3.2160804020100504, + "grad_norm": 3.5036733150482178, + "learning_rate": 7.146384479717814e-06, + "loss": 0.0515, + "step": 1920 + }, + { + "epoch": 3.2495812395309884, + "grad_norm": 2.324014186859131, + "learning_rate": 7.111111111111112e-06, + "loss": 0.0454, + "step": 1940 + }, + { + "epoch": 3.2830820770519265, + "grad_norm": 2.710326671600342, + "learning_rate": 7.07583774250441e-06, + "loss": 0.0473, + "step": 1960 + }, + { + "epoch": 3.3165829145728645, + "grad_norm": 2.8943896293640137, + "learning_rate": 7.040564373897708e-06, + "loss": 0.046, + "step": 1980 + }, + { + "epoch": 3.3500837520938025, + "grad_norm": 2.319986581802368, + "learning_rate": 7.005291005291006e-06, + "loss": 0.0449, + "step": 2000 + }, + { + "epoch": 3.3500837520938025, + "eval_loss": 0.10934468358755112, + "eval_runtime": 1762.4506, + "eval_samples_per_second": 2.71, + "eval_steps_per_second": 0.339, + "eval_wer": 7.172854824854443, + "step": 2000 + }, + { + "epoch": 3.3835845896147405, + "grad_norm": 3.317129135131836, + "learning_rate": 6.9700176366843046e-06, + "loss": 0.0645, + "step": 2020 + }, + { + "epoch": 3.4170854271356785, + "grad_norm": 1.9533768892288208, + "learning_rate": 6.9347442680776025e-06, + "loss": 0.0432, + "step": 2040 + }, + { + "epoch": 3.4505862646566166, + "grad_norm": 2.0508453845977783, + "learning_rate": 6.8994708994709005e-06, + "loss": 0.0521, + "step": 2060 + }, + { + "epoch": 3.4840871021775546, + "grad_norm": 2.163236141204834, + "learning_rate": 6.8641975308641985e-06, + "loss": 0.0529, + "step": 2080 + }, + { + "epoch": 3.5175879396984926, + "grad_norm": 2.7154581546783447, + "learning_rate": 6.8289241622574965e-06, + "loss": 0.0452, + "step": 2100 + }, + { + "epoch": 3.5510887772194306, + "grad_norm": 3.0822432041168213, + "learning_rate": 6.7936507936507944e-06, + "loss": 0.0546, + "step": 2120 + }, + { + "epoch": 3.5845896147403686, + "grad_norm": 4.19010591506958, + "learning_rate": 6.758377425044092e-06, + "loss": 0.0529, + "step": 2140 + }, + { + "epoch": 3.6180904522613067, + "grad_norm": 2.9883594512939453, + "learning_rate": 6.72310405643739e-06, + "loss": 0.0503, + "step": 2160 + }, + { + "epoch": 3.6515912897822447, + "grad_norm": 2.3664371967315674, + "learning_rate": 6.687830687830688e-06, + "loss": 0.0498, + "step": 2180 + }, + { + "epoch": 3.6850921273031827, + "grad_norm": 2.0549991130828857, + "learning_rate": 6.652557319223986e-06, + "loss": 0.051, + "step": 2200 + }, + { + "epoch": 3.7185929648241207, + "grad_norm": 2.5339038372039795, + "learning_rate": 6.617283950617285e-06, + "loss": 0.0568, + "step": 2220 + }, + { + "epoch": 3.7520938023450587, + "grad_norm": 1.9988099336624146, + "learning_rate": 6.582010582010583e-06, + "loss": 0.051, + "step": 2240 + }, + { + "epoch": 3.7855946398659968, + "grad_norm": 2.5243782997131348, + "learning_rate": 6.546737213403881e-06, + "loss": 0.056, + "step": 2260 + }, + { + "epoch": 3.819095477386935, + "grad_norm": 3.157158136367798, + "learning_rate": 6.511463844797179e-06, + "loss": 0.0497, + "step": 2280 + }, + { + "epoch": 3.852596314907873, + "grad_norm": 1.9286202192306519, + "learning_rate": 6.476190476190477e-06, + "loss": 0.0426, + "step": 2300 + }, + { + "epoch": 3.886097152428811, + "grad_norm": 3.808802604675293, + "learning_rate": 6.440917107583775e-06, + "loss": 0.0499, + "step": 2320 + }, + { + "epoch": 3.919597989949749, + "grad_norm": 2.506671667098999, + "learning_rate": 6.405643738977073e-06, + "loss": 0.052, + "step": 2340 + }, + { + "epoch": 3.953098827470687, + "grad_norm": 2.9451920986175537, + "learning_rate": 6.370370370370371e-06, + "loss": 0.0552, + "step": 2360 + }, + { + "epoch": 3.986599664991625, + "grad_norm": 2.592744827270508, + "learning_rate": 6.335097001763669e-06, + "loss": 0.0527, + "step": 2380 + }, + { + "epoch": 4.0201005025125625, + "grad_norm": 1.8891575336456299, + "learning_rate": 6.299823633156967e-06, + "loss": 0.0289, + "step": 2400 + }, + { + "epoch": 4.0536013400335005, + "grad_norm": 1.8053243160247803, + "learning_rate": 6.264550264550266e-06, + "loss": 0.0192, + "step": 2420 + }, + { + "epoch": 4.0871021775544385, + "grad_norm": 2.0084407329559326, + "learning_rate": 6.229276895943564e-06, + "loss": 0.0242, + "step": 2440 + }, + { + "epoch": 4.1206030150753765, + "grad_norm": 1.5919119119644165, + "learning_rate": 6.194003527336862e-06, + "loss": 0.0211, + "step": 2460 + }, + { + "epoch": 4.1541038525963145, + "grad_norm": 1.9214613437652588, + "learning_rate": 6.15873015873016e-06, + "loss": 0.0233, + "step": 2480 + }, + { + "epoch": 4.187604690117253, + "grad_norm": 1.2652311325073242, + "learning_rate": 6.123456790123458e-06, + "loss": 0.0199, + "step": 2500 + }, + { + "epoch": 4.187604690117253, + "eval_loss": 0.0981329157948494, + "eval_runtime": 1779.0213, + "eval_samples_per_second": 2.685, + "eval_steps_per_second": 0.336, + "eval_wer": 6.707549871146321, + "step": 2500 + }, + { + "epoch": 4.221105527638191, + "grad_norm": 2.026528835296631, + "learning_rate": 6.088183421516756e-06, + "loss": 0.0217, + "step": 2520 + }, + { + "epoch": 4.254606365159129, + "grad_norm": 1.596919059753418, + "learning_rate": 6.052910052910054e-06, + "loss": 0.0167, + "step": 2540 + }, + { + "epoch": 4.288107202680067, + "grad_norm": 2.9445090293884277, + "learning_rate": 6.017636684303352e-06, + "loss": 0.0225, + "step": 2560 + }, + { + "epoch": 4.321608040201005, + "grad_norm": 2.4160282611846924, + "learning_rate": 5.9823633156966496e-06, + "loss": 0.0253, + "step": 2580 + }, + { + "epoch": 4.355108877721943, + "grad_norm": 1.461127758026123, + "learning_rate": 5.9470899470899475e-06, + "loss": 0.0197, + "step": 2600 + }, + { + "epoch": 4.388609715242881, + "grad_norm": 2.7892863750457764, + "learning_rate": 5.911816578483246e-06, + "loss": 0.022, + "step": 2620 + }, + { + "epoch": 4.422110552763819, + "grad_norm": 1.651208758354187, + "learning_rate": 5.876543209876544e-06, + "loss": 0.0215, + "step": 2640 + }, + { + "epoch": 4.455611390284757, + "grad_norm": 2.2500391006469727, + "learning_rate": 5.841269841269842e-06, + "loss": 0.0247, + "step": 2660 + }, + { + "epoch": 4.489112227805695, + "grad_norm": 4.447635173797607, + "learning_rate": 5.80599647266314e-06, + "loss": 0.0263, + "step": 2680 + }, + { + "epoch": 4.522613065326633, + "grad_norm": 0.8300407528877258, + "learning_rate": 5.770723104056438e-06, + "loss": 0.0209, + "step": 2700 + }, + { + "epoch": 4.556113902847571, + "grad_norm": 1.6874111890792847, + "learning_rate": 5.735449735449736e-06, + "loss": 0.0195, + "step": 2720 + }, + { + "epoch": 4.589614740368509, + "grad_norm": 2.4045815467834473, + "learning_rate": 5.700176366843034e-06, + "loss": 0.0224, + "step": 2740 + }, + { + "epoch": 4.623115577889447, + "grad_norm": 2.3160908222198486, + "learning_rate": 5.664902998236332e-06, + "loss": 0.0179, + "step": 2760 + }, + { + "epoch": 4.656616415410385, + "grad_norm": 1.6684287786483765, + "learning_rate": 5.62962962962963e-06, + "loss": 0.0238, + "step": 2780 + }, + { + "epoch": 4.690117252931323, + "grad_norm": 1.973906397819519, + "learning_rate": 5.594356261022928e-06, + "loss": 0.0226, + "step": 2800 + }, + { + "epoch": 4.723618090452261, + "grad_norm": 2.270906686782837, + "learning_rate": 5.559082892416227e-06, + "loss": 0.0232, + "step": 2820 + }, + { + "epoch": 4.757118927973199, + "grad_norm": 1.8875011205673218, + "learning_rate": 5.523809523809525e-06, + "loss": 0.0218, + "step": 2840 + }, + { + "epoch": 4.790619765494137, + "grad_norm": 1.1312583684921265, + "learning_rate": 5.488536155202823e-06, + "loss": 0.0212, + "step": 2860 + }, + { + "epoch": 4.824120603015075, + "grad_norm": 0.864783525466919, + "learning_rate": 5.453262786596121e-06, + "loss": 0.023, + "step": 2880 + }, + { + "epoch": 4.857621440536013, + "grad_norm": 1.2935965061187744, + "learning_rate": 5.417989417989419e-06, + "loss": 0.019, + "step": 2900 + }, + { + "epoch": 4.891122278056951, + "grad_norm": 2.4576382637023926, + "learning_rate": 5.382716049382717e-06, + "loss": 0.0193, + "step": 2920 + }, + { + "epoch": 4.924623115577889, + "grad_norm": 2.71472430229187, + "learning_rate": 5.347442680776015e-06, + "loss": 0.0253, + "step": 2940 + }, + { + "epoch": 4.958123953098827, + "grad_norm": 2.84940505027771, + "learning_rate": 5.312169312169313e-06, + "loss": 0.0218, + "step": 2960 + }, + { + "epoch": 4.991624790619765, + "grad_norm": 1.8483999967575073, + "learning_rate": 5.276895943562611e-06, + "loss": 0.0226, + "step": 2980 + }, + { + "epoch": 5.025125628140704, + "grad_norm": 0.6126876473426819, + "learning_rate": 5.241622574955909e-06, + "loss": 0.0101, + "step": 3000 + }, + { + "epoch": 5.025125628140704, + "eval_loss": 0.09390027821063995, + "eval_runtime": 1767.4512, + "eval_samples_per_second": 2.702, + "eval_steps_per_second": 0.338, + "eval_wer": 5.500143170754987, + "step": 3000 + }, + { + "epoch": 5.058626465661642, + "grad_norm": 0.3711394965648651, + "learning_rate": 5.2063492063492076e-06, + "loss": 0.0119, + "step": 3020 + }, + { + "epoch": 5.09212730318258, + "grad_norm": 0.5738839507102966, + "learning_rate": 5.1710758377425055e-06, + "loss": 0.0086, + "step": 3040 + }, + { + "epoch": 5.125628140703517, + "grad_norm": 0.7609245777130127, + "learning_rate": 5.1358024691358035e-06, + "loss": 0.0093, + "step": 3060 + }, + { + "epoch": 5.159128978224456, + "grad_norm": 1.2764722108840942, + "learning_rate": 5.1005291005291015e-06, + "loss": 0.0111, + "step": 3080 + }, + { + "epoch": 5.192629815745394, + "grad_norm": 2.1169776916503906, + "learning_rate": 5.0652557319223995e-06, + "loss": 0.0107, + "step": 3100 + }, + { + "epoch": 5.226130653266332, + "grad_norm": 2.1893081665039062, + "learning_rate": 5.0299823633156974e-06, + "loss": 0.0097, + "step": 3120 + }, + { + "epoch": 5.259631490787269, + "grad_norm": 2.2419638633728027, + "learning_rate": 4.9947089947089946e-06, + "loss": 0.0098, + "step": 3140 + }, + { + "epoch": 5.293132328308207, + "grad_norm": 0.6479611992835999, + "learning_rate": 4.959435626102293e-06, + "loss": 0.0082, + "step": 3160 + }, + { + "epoch": 5.326633165829146, + "grad_norm": 0.4799642860889435, + "learning_rate": 4.924162257495591e-06, + "loss": 0.0119, + "step": 3180 + }, + { + "epoch": 5.360134003350084, + "grad_norm": 0.7716453075408936, + "learning_rate": 4.888888888888889e-06, + "loss": 0.0086, + "step": 3200 + }, + { + "epoch": 5.393634840871022, + "grad_norm": 1.2303547859191895, + "learning_rate": 4.853615520282187e-06, + "loss": 0.009, + "step": 3220 + }, + { + "epoch": 5.42713567839196, + "grad_norm": 0.6345349550247192, + "learning_rate": 4.818342151675485e-06, + "loss": 0.0085, + "step": 3240 + }, + { + "epoch": 5.460636515912898, + "grad_norm": 0.9741530418395996, + "learning_rate": 4.783068783068783e-06, + "loss": 0.0082, + "step": 3260 + }, + { + "epoch": 5.494137353433836, + "grad_norm": 1.1631624698638916, + "learning_rate": 4.747795414462081e-06, + "loss": 0.0086, + "step": 3280 + }, + { + "epoch": 5.527638190954773, + "grad_norm": 0.6502953767776489, + "learning_rate": 4.712522045855379e-06, + "loss": 0.0093, + "step": 3300 + }, + { + "epoch": 5.561139028475711, + "grad_norm": 0.7464337348937988, + "learning_rate": 4.677248677248677e-06, + "loss": 0.0086, + "step": 3320 + }, + { + "epoch": 5.594639865996649, + "grad_norm": 1.017751693725586, + "learning_rate": 4.641975308641975e-06, + "loss": 0.0091, + "step": 3340 + }, + { + "epoch": 5.628140703517588, + "grad_norm": 0.4273395836353302, + "learning_rate": 4.606701940035274e-06, + "loss": 0.0078, + "step": 3360 + }, + { + "epoch": 5.661641541038526, + "grad_norm": 0.6737497448921204, + "learning_rate": 4.571428571428572e-06, + "loss": 0.0106, + "step": 3380 + }, + { + "epoch": 5.695142378559464, + "grad_norm": 1.0791343450546265, + "learning_rate": 4.53615520282187e-06, + "loss": 0.0097, + "step": 3400 + }, + { + "epoch": 5.728643216080402, + "grad_norm": 1.0891772508621216, + "learning_rate": 4.500881834215168e-06, + "loss": 0.0109, + "step": 3420 + }, + { + "epoch": 5.76214405360134, + "grad_norm": 0.7465157508850098, + "learning_rate": 4.465608465608466e-06, + "loss": 0.0078, + "step": 3440 + }, + { + "epoch": 5.795644891122278, + "grad_norm": 0.7693866491317749, + "learning_rate": 4.430335097001764e-06, + "loss": 0.009, + "step": 3460 + }, + { + "epoch": 5.8291457286432165, + "grad_norm": 1.3295698165893555, + "learning_rate": 4.395061728395062e-06, + "loss": 0.016, + "step": 3480 + }, + { + "epoch": 5.8626465661641545, + "grad_norm": 2.4605352878570557, + "learning_rate": 4.35978835978836e-06, + "loss": 0.0081, + "step": 3500 + }, + { + "epoch": 5.8626465661641545, + "eval_loss": 0.09377142041921616, + "eval_runtime": 1773.01, + "eval_samples_per_second": 2.694, + "eval_steps_per_second": 0.337, + "eval_wer": 5.010976424549012, + "step": 3500 + }, + { + "epoch": 5.8961474036850925, + "grad_norm": 0.8250058889389038, + "learning_rate": 4.324514991181658e-06, + "loss": 0.0108, + "step": 3520 + }, + { + "epoch": 5.9296482412060305, + "grad_norm": 1.3606537580490112, + "learning_rate": 4.289241622574956e-06, + "loss": 0.0125, + "step": 3540 + }, + { + "epoch": 5.9631490787269685, + "grad_norm": 0.6893450021743774, + "learning_rate": 4.2539682539682546e-06, + "loss": 0.0101, + "step": 3560 + }, + { + "epoch": 5.9966499162479066, + "grad_norm": 0.8129726052284241, + "learning_rate": 4.2186948853615525e-06, + "loss": 0.0095, + "step": 3580 + }, + { + "epoch": 6.030150753768845, + "grad_norm": 0.5319514274597168, + "learning_rate": 4.1834215167548505e-06, + "loss": 0.0051, + "step": 3600 + }, + { + "epoch": 6.063651591289783, + "grad_norm": 0.20368462800979614, + "learning_rate": 4.1481481481481485e-06, + "loss": 0.0049, + "step": 3620 + }, + { + "epoch": 6.097152428810721, + "grad_norm": 1.1721038818359375, + "learning_rate": 4.1128747795414465e-06, + "loss": 0.0041, + "step": 3640 + }, + { + "epoch": 6.130653266331659, + "grad_norm": 0.17129285633563995, + "learning_rate": 4.0776014109347444e-06, + "loss": 0.0055, + "step": 3660 + }, + { + "epoch": 6.164154103852597, + "grad_norm": 0.31987234950065613, + "learning_rate": 4.042328042328042e-06, + "loss": 0.0037, + "step": 3680 + }, + { + "epoch": 6.197654941373535, + "grad_norm": 0.3214021921157837, + "learning_rate": 4.00705467372134e-06, + "loss": 0.0052, + "step": 3700 + }, + { + "epoch": 6.231155778894473, + "grad_norm": 1.3790876865386963, + "learning_rate": 3.971781305114638e-06, + "loss": 0.0053, + "step": 3720 + }, + { + "epoch": 6.264656616415411, + "grad_norm": 0.549566924571991, + "learning_rate": 3.936507936507936e-06, + "loss": 0.0048, + "step": 3740 + }, + { + "epoch": 6.298157453936349, + "grad_norm": 0.2458494007587433, + "learning_rate": 3.901234567901235e-06, + "loss": 0.006, + "step": 3760 + }, + { + "epoch": 6.331658291457287, + "grad_norm": 2.3661324977874756, + "learning_rate": 3.865961199294533e-06, + "loss": 0.0046, + "step": 3780 + }, + { + "epoch": 6.365159128978225, + "grad_norm": 0.7839031219482422, + "learning_rate": 3.830687830687831e-06, + "loss": 0.0059, + "step": 3800 + }, + { + "epoch": 6.398659966499163, + "grad_norm": 0.2562466263771057, + "learning_rate": 3.795414462081129e-06, + "loss": 0.0038, + "step": 3820 + }, + { + "epoch": 6.432160804020101, + "grad_norm": 0.9680606126785278, + "learning_rate": 3.760141093474427e-06, + "loss": 0.0053, + "step": 3840 + }, + { + "epoch": 6.465661641541039, + "grad_norm": 0.5647270083427429, + "learning_rate": 3.724867724867725e-06, + "loss": 0.0049, + "step": 3860 + }, + { + "epoch": 6.499162479061977, + "grad_norm": 0.5850030183792114, + "learning_rate": 3.689594356261023e-06, + "loss": 0.0039, + "step": 3880 + }, + { + "epoch": 6.532663316582915, + "grad_norm": 0.1745942085981369, + "learning_rate": 3.654320987654321e-06, + "loss": 0.0035, + "step": 3900 + }, + { + "epoch": 6.566164154103853, + "grad_norm": 0.2170235961675644, + "learning_rate": 3.6190476190476194e-06, + "loss": 0.0045, + "step": 3920 + }, + { + "epoch": 6.599664991624791, + "grad_norm": 0.30363383889198303, + "learning_rate": 3.5837742504409174e-06, + "loss": 0.0031, + "step": 3940 + }, + { + "epoch": 6.633165829145729, + "grad_norm": 0.22252851724624634, + "learning_rate": 3.5485008818342153e-06, + "loss": 0.004, + "step": 3960 + }, + { + "epoch": 6.666666666666667, + "grad_norm": 0.9289886951446533, + "learning_rate": 3.5132275132275133e-06, + "loss": 0.0055, + "step": 3980 + }, + { + "epoch": 6.700167504187605, + "grad_norm": 1.094429612159729, + "learning_rate": 3.4779541446208113e-06, + "loss": 0.0041, + "step": 4000 + }, + { + "epoch": 6.700167504187605, + "eval_loss": 0.09379494935274124, + "eval_runtime": 1761.7753, + "eval_samples_per_second": 2.711, + "eval_steps_per_second": 0.339, + "eval_wer": 4.89166746205975, + "step": 4000 + } + ], + "logging_steps": 20, + "max_steps": 5970, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "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": 3.693373670375424e+19, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-4000/training_args.bin b/checkpoint-4000/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6ffec31ca17b32691fb4cd1278f73919f3259cf4 --- /dev/null +++ b/checkpoint-4000/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41a3c645624f7463029286b7d5e1ad6924cfc44a0d153880431b77c40e5deb73 +size 5432 diff --git a/checkpoint-4500/config.json b/checkpoint-4500/config.json new file mode 100644 index 0000000000000000000000000000000000000000..337642e18341c758364c32c6b6159a090615329b --- /dev/null +++ b/checkpoint-4500/config.json @@ -0,0 +1,47 @@ +{ + "activation_dropout": 0.0, + "activation_function": "gelu", + "apply_spec_augment": false, + "architectures": [ + "WhisperForConditionalGeneration" + ], + "attention_dropout": 0.0, + "begin_suppress_tokens": null, + "bos_token_id": 50257, + "classifier_proj_size": 256, + "d_model": 768, + "decoder_attention_heads": 12, + "decoder_ffn_dim": 3072, + "decoder_layerdrop": 0.0, + "decoder_layers": 12, + "decoder_start_token_id": 50258, + "dropout": 0.0, + "encoder_attention_heads": 12, + "encoder_ffn_dim": 3072, + "encoder_layerdrop": 0.0, + "encoder_layers": 12, + "eos_token_id": 50257, + "forced_decoder_ids": null, + "init_std": 0.02, + "is_encoder_decoder": true, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "max_length": null, + "max_source_positions": 1500, + "max_target_positions": 448, + "median_filter_width": 7, + "model_type": "whisper", + "num_hidden_layers": 12, + "num_mel_bins": 80, + "pad_token_id": 50257, + "scale_embedding": false, + "torch_dtype": "float32", + "transformers_version": "4.51.3", + "use_cache": false, + "use_weighted_layer_sum": false, + "vocab_size": 51865 +} diff --git a/checkpoint-4500/generation_config.json b/checkpoint-4500/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..e1f4120825a303544ca23708d9dc52bbaa1dce16 --- /dev/null +++ b/checkpoint-4500/generation_config.json @@ -0,0 +1,9 @@ +{ + "bos_token_id": 50257, + "decoder_start_token_id": 50258, + "eos_token_id": 50257, + "pad_token_id": 50257, + "suppress_tokens": [], + "transformers_version": "4.51.3", + "use_cache": false +} diff --git a/checkpoint-4500/model.safetensors b/checkpoint-4500/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..1bdfdbaca5c711fe1cd226e87ae1fc61634133b2 --- /dev/null +++ b/checkpoint-4500/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:67f3f81214997d78ff05f34da285df9535a668fd45705f2c1b80276325e040e6 +size 966995080 diff --git a/checkpoint-4500/optimizer.pt b/checkpoint-4500/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..34cd273f986adafbbdc4279a2410da794dc2470d --- /dev/null +++ b/checkpoint-4500/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b4847f29a2d0289da07a565245a3d1e3c82d4cb03ade9af85154b07f0f805cb7 +size 1925064044 diff --git a/checkpoint-4500/preprocessor_config.json b/checkpoint-4500/preprocessor_config.json new file mode 100644 index 0000000000000000000000000000000000000000..d7c5b1dfd466c815664b16139e8f605620d22e4e --- /dev/null +++ b/checkpoint-4500/preprocessor_config.json @@ -0,0 +1,15 @@ +{ + "chunk_length": 30, + "dither": 0.0, + "feature_extractor_type": "WhisperFeatureExtractor", + "feature_size": 80, + "hop_length": 160, + "n_fft": 400, + "n_samples": 480000, + "nb_max_frames": 3000, + "padding_side": "right", + "padding_value": 0.0, + "processor_class": "WhisperProcessor", + "return_attention_mask": false, + "sampling_rate": 16000 +} diff --git a/checkpoint-4500/rng_state.pth b/checkpoint-4500/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..02c20e14c6fa53d2fa57418d552c4fec84c16ec8 --- /dev/null +++ b/checkpoint-4500/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a59e76708c7e36c5b29b697ed7a4e3c74a382ef5ea34a10751ae97e2ff0d1eb4 +size 14244 diff --git a/checkpoint-4500/scaler.pt b/checkpoint-4500/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..ed5ac75d0af0b1b08edf701983b3d9135f660cdc --- /dev/null +++ b/checkpoint-4500/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e11e2adff07dcf7983c1d81a368ff3e679eebe81f68e8b1686b82e864e67d00a +size 988 diff --git a/checkpoint-4500/scheduler.pt b/checkpoint-4500/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..d125135f2ab0e119a1d0c1920570280363d50550 --- /dev/null +++ b/checkpoint-4500/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fe5e63f171df35b08dd589864833a301bba214cd63c6f22056db23d6a70fc83f +size 1064 diff --git a/checkpoint-4500/trainer_state.json b/checkpoint-4500/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..3dc4b691b4924fdea4db36754314cef700655095 --- /dev/null +++ b/checkpoint-4500/trainer_state.json @@ -0,0 +1,1690 @@ +{ + "best_global_step": 4500, + "best_metric": 4.50510642359454, + "best_model_checkpoint": "./SALAMA_C3/checkpoint-4500", + "epoch": 7.5376884422110555, + "eval_steps": 500, + "global_step": 4500, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.03350083752093802, + "grad_norm": 11.757207870483398, + "learning_rate": 6.333333333333334e-07, + "loss": 0.5112, + "step": 20 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 10.197953224182129, + "learning_rate": 1.3e-06, + "loss": 0.5283, + "step": 40 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 12.811074256896973, + "learning_rate": 1.9666666666666668e-06, + "loss": 0.474, + "step": 60 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 8.559342384338379, + "learning_rate": 2.6e-06, + "loss": 0.424, + "step": 80 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 6.9629225730896, + "learning_rate": 3.266666666666667e-06, + "loss": 0.4465, + "step": 100 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 7.833058834075928, + "learning_rate": 3.9333333333333335e-06, + "loss": 0.3764, + "step": 120 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 6.882424831390381, + "learning_rate": 4.600000000000001e-06, + "loss": 0.3705, + "step": 140 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 5.4183244705200195, + "learning_rate": 5.2666666666666665e-06, + "loss": 0.4116, + "step": 160 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 6.103787899017334, + "learning_rate": 5.933333333333335e-06, + "loss": 0.3903, + "step": 180 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 6.084160804748535, + "learning_rate": 6.600000000000001e-06, + "loss": 0.3794, + "step": 200 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 4.531406879425049, + "learning_rate": 7.266666666666668e-06, + "loss": 0.3904, + "step": 220 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 6.702854156494141, + "learning_rate": 7.933333333333334e-06, + "loss": 0.3672, + "step": 240 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 6.386377811431885, + "learning_rate": 8.6e-06, + "loss": 0.3379, + "step": 260 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 5.33281135559082, + "learning_rate": 9.266666666666667e-06, + "loss": 0.4102, + "step": 280 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 6.031435012817383, + "learning_rate": 9.933333333333334e-06, + "loss": 0.3815, + "step": 300 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 4.8638176918029785, + "learning_rate": 9.968253968253969e-06, + "loss": 0.358, + "step": 320 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 5.352113723754883, + "learning_rate": 9.932980599647268e-06, + "loss": 0.3578, + "step": 340 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 5.560739040374756, + "learning_rate": 9.897707231040565e-06, + "loss": 0.3795, + "step": 360 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 6.186940670013428, + "learning_rate": 9.862433862433864e-06, + "loss": 0.3852, + "step": 380 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 5.37507438659668, + "learning_rate": 9.827160493827161e-06, + "loss": 0.3761, + "step": 400 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 6.373449802398682, + "learning_rate": 9.79188712522046e-06, + "loss": 0.3674, + "step": 420 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 5.702625274658203, + "learning_rate": 9.756613756613757e-06, + "loss": 0.4013, + "step": 440 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 5.5675153732299805, + "learning_rate": 9.721340388007056e-06, + "loss": 0.3495, + "step": 460 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 6.296374320983887, + "learning_rate": 9.686067019400353e-06, + "loss": 0.3704, + "step": 480 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 4.830463886260986, + "learning_rate": 9.650793650793652e-06, + "loss": 0.36, + "step": 500 + }, + { + "epoch": 0.8375209380234506, + "eval_loss": 0.27764827013015747, + "eval_runtime": 1752.3826, + "eval_samples_per_second": 2.725, + "eval_steps_per_second": 0.341, + "eval_wer": 19.59291781998664, + "step": 500 + }, + { + "epoch": 0.8710217755443886, + "grad_norm": 4.863058090209961, + "learning_rate": 9.61552028218695e-06, + "loss": 0.3874, + "step": 520 + }, + { + "epoch": 0.9045226130653267, + "grad_norm": 6.121027946472168, + "learning_rate": 9.580246913580248e-06, + "loss": 0.382, + "step": 540 + }, + { + "epoch": 0.9380234505862647, + "grad_norm": 4.538082599639893, + "learning_rate": 9.544973544973546e-06, + "loss": 0.3611, + "step": 560 + }, + { + "epoch": 0.9715242881072027, + "grad_norm": 4.8772382736206055, + "learning_rate": 9.509700176366844e-06, + "loss": 0.3584, + "step": 580 + }, + { + "epoch": 1.0050251256281406, + "grad_norm": 4.039212703704834, + "learning_rate": 9.474426807760142e-06, + "loss": 0.3701, + "step": 600 + }, + { + "epoch": 1.0385259631490786, + "grad_norm": 4.787687301635742, + "learning_rate": 9.43915343915344e-06, + "loss": 0.2058, + "step": 620 + }, + { + "epoch": 1.0720268006700167, + "grad_norm": 4.503021717071533, + "learning_rate": 9.403880070546738e-06, + "loss": 0.2455, + "step": 640 + }, + { + "epoch": 1.1055276381909547, + "grad_norm": 6.663857936859131, + "learning_rate": 9.368606701940036e-06, + "loss": 0.2179, + "step": 660 + }, + { + "epoch": 1.1390284757118927, + "grad_norm": 4.486196041107178, + "learning_rate": 9.333333333333334e-06, + "loss": 0.2196, + "step": 680 + }, + { + "epoch": 1.1725293132328307, + "grad_norm": 4.667060852050781, + "learning_rate": 9.298059964726633e-06, + "loss": 0.2183, + "step": 700 + }, + { + "epoch": 1.2060301507537687, + "grad_norm": 5.608316898345947, + "learning_rate": 9.26278659611993e-06, + "loss": 0.2161, + "step": 720 + }, + { + "epoch": 1.2395309882747068, + "grad_norm": 4.2184271812438965, + "learning_rate": 9.227513227513229e-06, + "loss": 0.2382, + "step": 740 + }, + { + "epoch": 1.2730318257956448, + "grad_norm": 3.871945381164551, + "learning_rate": 9.192239858906526e-06, + "loss": 0.2214, + "step": 760 + }, + { + "epoch": 1.3065326633165828, + "grad_norm": 3.730222225189209, + "learning_rate": 9.156966490299825e-06, + "loss": 0.2213, + "step": 780 + }, + { + "epoch": 1.3400335008375208, + "grad_norm": 4.740777015686035, + "learning_rate": 9.121693121693122e-06, + "loss": 0.2503, + "step": 800 + }, + { + "epoch": 1.3735343383584588, + "grad_norm": 4.123469352722168, + "learning_rate": 9.086419753086421e-06, + "loss": 0.2443, + "step": 820 + }, + { + "epoch": 1.4070351758793969, + "grad_norm": 3.988917827606201, + "learning_rate": 9.051146384479718e-06, + "loss": 0.241, + "step": 840 + }, + { + "epoch": 1.4405360134003349, + "grad_norm": 5.4730305671691895, + "learning_rate": 9.015873015873017e-06, + "loss": 0.2163, + "step": 860 + }, + { + "epoch": 1.474036850921273, + "grad_norm": 5.020652770996094, + "learning_rate": 8.980599647266314e-06, + "loss": 0.2387, + "step": 880 + }, + { + "epoch": 1.507537688442211, + "grad_norm": 4.231486797332764, + "learning_rate": 8.945326278659613e-06, + "loss": 0.2523, + "step": 900 + }, + { + "epoch": 1.541038525963149, + "grad_norm": 6.197975158691406, + "learning_rate": 8.910052910052912e-06, + "loss": 0.2255, + "step": 920 + }, + { + "epoch": 1.574539363484087, + "grad_norm": 5.489210605621338, + "learning_rate": 8.874779541446209e-06, + "loss": 0.2397, + "step": 940 + }, + { + "epoch": 1.608040201005025, + "grad_norm": 4.32737398147583, + "learning_rate": 8.839506172839508e-06, + "loss": 0.2298, + "step": 960 + }, + { + "epoch": 1.641541038525963, + "grad_norm": 4.504214763641357, + "learning_rate": 8.804232804232805e-06, + "loss": 0.2301, + "step": 980 + }, + { + "epoch": 1.675041876046901, + "grad_norm": 3.9694759845733643, + "learning_rate": 8.768959435626104e-06, + "loss": 0.2338, + "step": 1000 + }, + { + "epoch": 1.675041876046901, + "eval_loss": 0.18838582932949066, + "eval_runtime": 1770.3449, + "eval_samples_per_second": 2.698, + "eval_steps_per_second": 0.337, + "eval_wer": 14.071299035983584, + "step": 1000 + }, + { + "epoch": 1.708542713567839, + "grad_norm": 4.972934722900391, + "learning_rate": 8.7336860670194e-06, + "loss": 0.2388, + "step": 1020 + }, + { + "epoch": 1.742043551088777, + "grad_norm": 5.178994655609131, + "learning_rate": 8.6984126984127e-06, + "loss": 0.2531, + "step": 1040 + }, + { + "epoch": 1.775544388609715, + "grad_norm": 3.560372829437256, + "learning_rate": 8.663139329805997e-06, + "loss": 0.2187, + "step": 1060 + }, + { + "epoch": 1.809045226130653, + "grad_norm": 4.647324562072754, + "learning_rate": 8.627865961199296e-06, + "loss": 0.2413, + "step": 1080 + }, + { + "epoch": 1.8425460636515911, + "grad_norm": 3.6784422397613525, + "learning_rate": 8.592592592592593e-06, + "loss": 0.2242, + "step": 1100 + }, + { + "epoch": 1.8760469011725294, + "grad_norm": 5.928018093109131, + "learning_rate": 8.557319223985891e-06, + "loss": 0.2406, + "step": 1120 + }, + { + "epoch": 1.9095477386934674, + "grad_norm": 6.358487606048584, + "learning_rate": 8.52204585537919e-06, + "loss": 0.2482, + "step": 1140 + }, + { + "epoch": 1.9430485762144054, + "grad_norm": 4.409506797790527, + "learning_rate": 8.486772486772487e-06, + "loss": 0.2415, + "step": 1160 + }, + { + "epoch": 1.9765494137353434, + "grad_norm": 5.169639587402344, + "learning_rate": 8.451499118165786e-06, + "loss": 0.222, + "step": 1180 + }, + { + "epoch": 2.0100502512562812, + "grad_norm": 2.8841328620910645, + "learning_rate": 8.416225749559083e-06, + "loss": 0.2109, + "step": 1200 + }, + { + "epoch": 2.0435510887772192, + "grad_norm": 3.835498332977295, + "learning_rate": 8.380952380952382e-06, + "loss": 0.1284, + "step": 1220 + }, + { + "epoch": 2.0770519262981573, + "grad_norm": 2.584859848022461, + "learning_rate": 8.34567901234568e-06, + "loss": 0.1059, + "step": 1240 + }, + { + "epoch": 2.1105527638190953, + "grad_norm": 2.797682285308838, + "learning_rate": 8.310405643738978e-06, + "loss": 0.0967, + "step": 1260 + }, + { + "epoch": 2.1440536013400333, + "grad_norm": 3.2937309741973877, + "learning_rate": 8.275132275132275e-06, + "loss": 0.1267, + "step": 1280 + }, + { + "epoch": 2.1775544388609713, + "grad_norm": 3.344325065612793, + "learning_rate": 8.239858906525574e-06, + "loss": 0.1354, + "step": 1300 + }, + { + "epoch": 2.2110552763819094, + "grad_norm": 2.955522060394287, + "learning_rate": 8.204585537918873e-06, + "loss": 0.104, + "step": 1320 + }, + { + "epoch": 2.2445561139028474, + "grad_norm": 5.180908203125, + "learning_rate": 8.16931216931217e-06, + "loss": 0.1229, + "step": 1340 + }, + { + "epoch": 2.2780569514237854, + "grad_norm": 3.639417886734009, + "learning_rate": 8.134038800705469e-06, + "loss": 0.104, + "step": 1360 + }, + { + "epoch": 2.3115577889447234, + "grad_norm": 3.7380902767181396, + "learning_rate": 8.098765432098766e-06, + "loss": 0.1141, + "step": 1380 + }, + { + "epoch": 2.3450586264656614, + "grad_norm": 4.100318908691406, + "learning_rate": 8.063492063492065e-06, + "loss": 0.1123, + "step": 1400 + }, + { + "epoch": 2.3785594639865995, + "grad_norm": 2.8147048950195312, + "learning_rate": 8.028218694885362e-06, + "loss": 0.1039, + "step": 1420 + }, + { + "epoch": 2.4120603015075375, + "grad_norm": 3.7376410961151123, + "learning_rate": 7.99294532627866e-06, + "loss": 0.1179, + "step": 1440 + }, + { + "epoch": 2.4455611390284755, + "grad_norm": 4.320065975189209, + "learning_rate": 7.957671957671958e-06, + "loss": 0.1167, + "step": 1460 + }, + { + "epoch": 2.4790619765494135, + "grad_norm": 4.415127277374268, + "learning_rate": 7.922398589065257e-06, + "loss": 0.125, + "step": 1480 + }, + { + "epoch": 2.5125628140703515, + "grad_norm": 2.631763219833374, + "learning_rate": 7.887125220458554e-06, + "loss": 0.1188, + "step": 1500 + }, + { + "epoch": 2.5125628140703515, + "eval_loss": 0.13635103404521942, + "eval_runtime": 1806.1099, + "eval_samples_per_second": 2.644, + "eval_steps_per_second": 0.331, + "eval_wer": 11.713753937195762, + "step": 1500 + }, + { + "epoch": 2.5460636515912896, + "grad_norm": 3.608773946762085, + "learning_rate": 7.851851851851853e-06, + "loss": 0.1222, + "step": 1520 + }, + { + "epoch": 2.5795644891122276, + "grad_norm": 3.4494569301605225, + "learning_rate": 7.816578483245151e-06, + "loss": 0.1087, + "step": 1540 + }, + { + "epoch": 2.6130653266331656, + "grad_norm": 3.6112048625946045, + "learning_rate": 7.781305114638449e-06, + "loss": 0.1097, + "step": 1560 + }, + { + "epoch": 2.6465661641541036, + "grad_norm": 3.1978707313537598, + "learning_rate": 7.746031746031747e-06, + "loss": 0.1101, + "step": 1580 + }, + { + "epoch": 2.6800670016750416, + "grad_norm": 3.6376733779907227, + "learning_rate": 7.710758377425045e-06, + "loss": 0.1056, + "step": 1600 + }, + { + "epoch": 2.7135678391959797, + "grad_norm": 3.754915237426758, + "learning_rate": 7.675485008818343e-06, + "loss": 0.1103, + "step": 1620 + }, + { + "epoch": 2.7470686767169177, + "grad_norm": 3.329240322113037, + "learning_rate": 7.64021164021164e-06, + "loss": 0.1193, + "step": 1640 + }, + { + "epoch": 2.7805695142378557, + "grad_norm": 3.9300169944763184, + "learning_rate": 7.604938271604939e-06, + "loss": 0.1232, + "step": 1660 + }, + { + "epoch": 2.8140703517587937, + "grad_norm": 4.290626049041748, + "learning_rate": 7.569664902998237e-06, + "loss": 0.1374, + "step": 1680 + }, + { + "epoch": 2.8475711892797317, + "grad_norm": 3.8987998962402344, + "learning_rate": 7.534391534391535e-06, + "loss": 0.1062, + "step": 1700 + }, + { + "epoch": 2.8810720268006698, + "grad_norm": 4.234960079193115, + "learning_rate": 7.499118165784833e-06, + "loss": 0.111, + "step": 1720 + }, + { + "epoch": 2.914572864321608, + "grad_norm": 3.103458881378174, + "learning_rate": 7.463844797178131e-06, + "loss": 0.1076, + "step": 1740 + }, + { + "epoch": 2.948073701842546, + "grad_norm": 3.270204782485962, + "learning_rate": 7.428571428571429e-06, + "loss": 0.1224, + "step": 1760 + }, + { + "epoch": 2.981574539363484, + "grad_norm": 4.261337757110596, + "learning_rate": 7.393298059964727e-06, + "loss": 0.1224, + "step": 1780 + }, + { + "epoch": 3.0150753768844223, + "grad_norm": 1.9531301259994507, + "learning_rate": 7.358024691358025e-06, + "loss": 0.0913, + "step": 1800 + }, + { + "epoch": 3.0485762144053603, + "grad_norm": 1.866215467453003, + "learning_rate": 7.322751322751324e-06, + "loss": 0.0466, + "step": 1820 + }, + { + "epoch": 3.0820770519262983, + "grad_norm": 3.9456610679626465, + "learning_rate": 7.287477954144622e-06, + "loss": 0.0451, + "step": 1840 + }, + { + "epoch": 3.1155778894472363, + "grad_norm": 1.9518849849700928, + "learning_rate": 7.25220458553792e-06, + "loss": 0.0414, + "step": 1860 + }, + { + "epoch": 3.1490787269681744, + "grad_norm": 2.232792854309082, + "learning_rate": 7.216931216931218e-06, + "loss": 0.0502, + "step": 1880 + }, + { + "epoch": 3.1825795644891124, + "grad_norm": 2.074127674102783, + "learning_rate": 7.181657848324516e-06, + "loss": 0.057, + "step": 1900 + }, + { + "epoch": 3.2160804020100504, + "grad_norm": 3.5036733150482178, + "learning_rate": 7.146384479717814e-06, + "loss": 0.0515, + "step": 1920 + }, + { + "epoch": 3.2495812395309884, + "grad_norm": 2.324014186859131, + "learning_rate": 7.111111111111112e-06, + "loss": 0.0454, + "step": 1940 + }, + { + "epoch": 3.2830820770519265, + "grad_norm": 2.710326671600342, + "learning_rate": 7.07583774250441e-06, + "loss": 0.0473, + "step": 1960 + }, + { + "epoch": 3.3165829145728645, + "grad_norm": 2.8943896293640137, + "learning_rate": 7.040564373897708e-06, + "loss": 0.046, + "step": 1980 + }, + { + "epoch": 3.3500837520938025, + "grad_norm": 2.319986581802368, + "learning_rate": 7.005291005291006e-06, + "loss": 0.0449, + "step": 2000 + }, + { + "epoch": 3.3500837520938025, + "eval_loss": 0.10934468358755112, + "eval_runtime": 1762.4506, + "eval_samples_per_second": 2.71, + "eval_steps_per_second": 0.339, + "eval_wer": 7.172854824854443, + "step": 2000 + }, + { + "epoch": 3.3835845896147405, + "grad_norm": 3.317129135131836, + "learning_rate": 6.9700176366843046e-06, + "loss": 0.0645, + "step": 2020 + }, + { + "epoch": 3.4170854271356785, + "grad_norm": 1.9533768892288208, + "learning_rate": 6.9347442680776025e-06, + "loss": 0.0432, + "step": 2040 + }, + { + "epoch": 3.4505862646566166, + "grad_norm": 2.0508453845977783, + "learning_rate": 6.8994708994709005e-06, + "loss": 0.0521, + "step": 2060 + }, + { + "epoch": 3.4840871021775546, + "grad_norm": 2.163236141204834, + "learning_rate": 6.8641975308641985e-06, + "loss": 0.0529, + "step": 2080 + }, + { + "epoch": 3.5175879396984926, + "grad_norm": 2.7154581546783447, + "learning_rate": 6.8289241622574965e-06, + "loss": 0.0452, + "step": 2100 + }, + { + "epoch": 3.5510887772194306, + "grad_norm": 3.0822432041168213, + "learning_rate": 6.7936507936507944e-06, + "loss": 0.0546, + "step": 2120 + }, + { + "epoch": 3.5845896147403686, + "grad_norm": 4.19010591506958, + "learning_rate": 6.758377425044092e-06, + "loss": 0.0529, + "step": 2140 + }, + { + "epoch": 3.6180904522613067, + "grad_norm": 2.9883594512939453, + "learning_rate": 6.72310405643739e-06, + "loss": 0.0503, + "step": 2160 + }, + { + "epoch": 3.6515912897822447, + "grad_norm": 2.3664371967315674, + "learning_rate": 6.687830687830688e-06, + "loss": 0.0498, + "step": 2180 + }, + { + "epoch": 3.6850921273031827, + "grad_norm": 2.0549991130828857, + "learning_rate": 6.652557319223986e-06, + "loss": 0.051, + "step": 2200 + }, + { + "epoch": 3.7185929648241207, + "grad_norm": 2.5339038372039795, + "learning_rate": 6.617283950617285e-06, + "loss": 0.0568, + "step": 2220 + }, + { + "epoch": 3.7520938023450587, + "grad_norm": 1.9988099336624146, + "learning_rate": 6.582010582010583e-06, + "loss": 0.051, + "step": 2240 + }, + { + "epoch": 3.7855946398659968, + "grad_norm": 2.5243782997131348, + "learning_rate": 6.546737213403881e-06, + "loss": 0.056, + "step": 2260 + }, + { + "epoch": 3.819095477386935, + "grad_norm": 3.157158136367798, + "learning_rate": 6.511463844797179e-06, + "loss": 0.0497, + "step": 2280 + }, + { + "epoch": 3.852596314907873, + "grad_norm": 1.9286202192306519, + "learning_rate": 6.476190476190477e-06, + "loss": 0.0426, + "step": 2300 + }, + { + "epoch": 3.886097152428811, + "grad_norm": 3.808802604675293, + "learning_rate": 6.440917107583775e-06, + "loss": 0.0499, + "step": 2320 + }, + { + "epoch": 3.919597989949749, + "grad_norm": 2.506671667098999, + "learning_rate": 6.405643738977073e-06, + "loss": 0.052, + "step": 2340 + }, + { + "epoch": 3.953098827470687, + "grad_norm": 2.9451920986175537, + "learning_rate": 6.370370370370371e-06, + "loss": 0.0552, + "step": 2360 + }, + { + "epoch": 3.986599664991625, + "grad_norm": 2.592744827270508, + "learning_rate": 6.335097001763669e-06, + "loss": 0.0527, + "step": 2380 + }, + { + "epoch": 4.0201005025125625, + "grad_norm": 1.8891575336456299, + "learning_rate": 6.299823633156967e-06, + "loss": 0.0289, + "step": 2400 + }, + { + "epoch": 4.0536013400335005, + "grad_norm": 1.8053243160247803, + "learning_rate": 6.264550264550266e-06, + "loss": 0.0192, + "step": 2420 + }, + { + "epoch": 4.0871021775544385, + "grad_norm": 2.0084407329559326, + "learning_rate": 6.229276895943564e-06, + "loss": 0.0242, + "step": 2440 + }, + { + "epoch": 4.1206030150753765, + "grad_norm": 1.5919119119644165, + "learning_rate": 6.194003527336862e-06, + "loss": 0.0211, + "step": 2460 + }, + { + "epoch": 4.1541038525963145, + "grad_norm": 1.9214613437652588, + "learning_rate": 6.15873015873016e-06, + "loss": 0.0233, + "step": 2480 + }, + { + "epoch": 4.187604690117253, + "grad_norm": 1.2652311325073242, + "learning_rate": 6.123456790123458e-06, + "loss": 0.0199, + "step": 2500 + }, + { + "epoch": 4.187604690117253, + "eval_loss": 0.0981329157948494, + "eval_runtime": 1779.0213, + "eval_samples_per_second": 2.685, + "eval_steps_per_second": 0.336, + "eval_wer": 6.707549871146321, + "step": 2500 + }, + { + "epoch": 4.221105527638191, + "grad_norm": 2.026528835296631, + "learning_rate": 6.088183421516756e-06, + "loss": 0.0217, + "step": 2520 + }, + { + "epoch": 4.254606365159129, + "grad_norm": 1.596919059753418, + "learning_rate": 6.052910052910054e-06, + "loss": 0.0167, + "step": 2540 + }, + { + "epoch": 4.288107202680067, + "grad_norm": 2.9445090293884277, + "learning_rate": 6.017636684303352e-06, + "loss": 0.0225, + "step": 2560 + }, + { + "epoch": 4.321608040201005, + "grad_norm": 2.4160282611846924, + "learning_rate": 5.9823633156966496e-06, + "loss": 0.0253, + "step": 2580 + }, + { + "epoch": 4.355108877721943, + "grad_norm": 1.461127758026123, + "learning_rate": 5.9470899470899475e-06, + "loss": 0.0197, + "step": 2600 + }, + { + "epoch": 4.388609715242881, + "grad_norm": 2.7892863750457764, + "learning_rate": 5.911816578483246e-06, + "loss": 0.022, + "step": 2620 + }, + { + "epoch": 4.422110552763819, + "grad_norm": 1.651208758354187, + "learning_rate": 5.876543209876544e-06, + "loss": 0.0215, + "step": 2640 + }, + { + "epoch": 4.455611390284757, + "grad_norm": 2.2500391006469727, + "learning_rate": 5.841269841269842e-06, + "loss": 0.0247, + "step": 2660 + }, + { + "epoch": 4.489112227805695, + "grad_norm": 4.447635173797607, + "learning_rate": 5.80599647266314e-06, + "loss": 0.0263, + "step": 2680 + }, + { + "epoch": 4.522613065326633, + "grad_norm": 0.8300407528877258, + "learning_rate": 5.770723104056438e-06, + "loss": 0.0209, + "step": 2700 + }, + { + "epoch": 4.556113902847571, + "grad_norm": 1.6874111890792847, + "learning_rate": 5.735449735449736e-06, + "loss": 0.0195, + "step": 2720 + }, + { + "epoch": 4.589614740368509, + "grad_norm": 2.4045815467834473, + "learning_rate": 5.700176366843034e-06, + "loss": 0.0224, + "step": 2740 + }, + { + "epoch": 4.623115577889447, + "grad_norm": 2.3160908222198486, + "learning_rate": 5.664902998236332e-06, + "loss": 0.0179, + "step": 2760 + }, + { + "epoch": 4.656616415410385, + "grad_norm": 1.6684287786483765, + "learning_rate": 5.62962962962963e-06, + "loss": 0.0238, + "step": 2780 + }, + { + "epoch": 4.690117252931323, + "grad_norm": 1.973906397819519, + "learning_rate": 5.594356261022928e-06, + "loss": 0.0226, + "step": 2800 + }, + { + "epoch": 4.723618090452261, + "grad_norm": 2.270906686782837, + "learning_rate": 5.559082892416227e-06, + "loss": 0.0232, + "step": 2820 + }, + { + "epoch": 4.757118927973199, + "grad_norm": 1.8875011205673218, + "learning_rate": 5.523809523809525e-06, + "loss": 0.0218, + "step": 2840 + }, + { + "epoch": 4.790619765494137, + "grad_norm": 1.1312583684921265, + "learning_rate": 5.488536155202823e-06, + "loss": 0.0212, + "step": 2860 + }, + { + "epoch": 4.824120603015075, + "grad_norm": 0.864783525466919, + "learning_rate": 5.453262786596121e-06, + "loss": 0.023, + "step": 2880 + }, + { + "epoch": 4.857621440536013, + "grad_norm": 1.2935965061187744, + "learning_rate": 5.417989417989419e-06, + "loss": 0.019, + "step": 2900 + }, + { + "epoch": 4.891122278056951, + "grad_norm": 2.4576382637023926, + "learning_rate": 5.382716049382717e-06, + "loss": 0.0193, + "step": 2920 + }, + { + "epoch": 4.924623115577889, + "grad_norm": 2.71472430229187, + "learning_rate": 5.347442680776015e-06, + "loss": 0.0253, + "step": 2940 + }, + { + "epoch": 4.958123953098827, + "grad_norm": 2.84940505027771, + "learning_rate": 5.312169312169313e-06, + "loss": 0.0218, + "step": 2960 + }, + { + "epoch": 4.991624790619765, + "grad_norm": 1.8483999967575073, + "learning_rate": 5.276895943562611e-06, + "loss": 0.0226, + "step": 2980 + }, + { + "epoch": 5.025125628140704, + "grad_norm": 0.6126876473426819, + "learning_rate": 5.241622574955909e-06, + "loss": 0.0101, + "step": 3000 + }, + { + "epoch": 5.025125628140704, + "eval_loss": 0.09390027821063995, + "eval_runtime": 1767.4512, + "eval_samples_per_second": 2.702, + "eval_steps_per_second": 0.338, + "eval_wer": 5.500143170754987, + "step": 3000 + }, + { + "epoch": 5.058626465661642, + "grad_norm": 0.3711394965648651, + "learning_rate": 5.2063492063492076e-06, + "loss": 0.0119, + "step": 3020 + }, + { + "epoch": 5.09212730318258, + "grad_norm": 0.5738839507102966, + "learning_rate": 5.1710758377425055e-06, + "loss": 0.0086, + "step": 3040 + }, + { + "epoch": 5.125628140703517, + "grad_norm": 0.7609245777130127, + "learning_rate": 5.1358024691358035e-06, + "loss": 0.0093, + "step": 3060 + }, + { + "epoch": 5.159128978224456, + "grad_norm": 1.2764722108840942, + "learning_rate": 5.1005291005291015e-06, + "loss": 0.0111, + "step": 3080 + }, + { + "epoch": 5.192629815745394, + "grad_norm": 2.1169776916503906, + "learning_rate": 5.0652557319223995e-06, + "loss": 0.0107, + "step": 3100 + }, + { + "epoch": 5.226130653266332, + "grad_norm": 2.1893081665039062, + "learning_rate": 5.0299823633156974e-06, + "loss": 0.0097, + "step": 3120 + }, + { + "epoch": 5.259631490787269, + "grad_norm": 2.2419638633728027, + "learning_rate": 4.9947089947089946e-06, + "loss": 0.0098, + "step": 3140 + }, + { + "epoch": 5.293132328308207, + "grad_norm": 0.6479611992835999, + "learning_rate": 4.959435626102293e-06, + "loss": 0.0082, + "step": 3160 + }, + { + "epoch": 5.326633165829146, + "grad_norm": 0.4799642860889435, + "learning_rate": 4.924162257495591e-06, + "loss": 0.0119, + "step": 3180 + }, + { + "epoch": 5.360134003350084, + "grad_norm": 0.7716453075408936, + "learning_rate": 4.888888888888889e-06, + "loss": 0.0086, + "step": 3200 + }, + { + "epoch": 5.393634840871022, + "grad_norm": 1.2303547859191895, + "learning_rate": 4.853615520282187e-06, + "loss": 0.009, + "step": 3220 + }, + { + "epoch": 5.42713567839196, + "grad_norm": 0.6345349550247192, + "learning_rate": 4.818342151675485e-06, + "loss": 0.0085, + "step": 3240 + }, + { + "epoch": 5.460636515912898, + "grad_norm": 0.9741530418395996, + "learning_rate": 4.783068783068783e-06, + "loss": 0.0082, + "step": 3260 + }, + { + "epoch": 5.494137353433836, + "grad_norm": 1.1631624698638916, + "learning_rate": 4.747795414462081e-06, + "loss": 0.0086, + "step": 3280 + }, + { + "epoch": 5.527638190954773, + "grad_norm": 0.6502953767776489, + "learning_rate": 4.712522045855379e-06, + "loss": 0.0093, + "step": 3300 + }, + { + "epoch": 5.561139028475711, + "grad_norm": 0.7464337348937988, + "learning_rate": 4.677248677248677e-06, + "loss": 0.0086, + "step": 3320 + }, + { + "epoch": 5.594639865996649, + "grad_norm": 1.017751693725586, + "learning_rate": 4.641975308641975e-06, + "loss": 0.0091, + "step": 3340 + }, + { + "epoch": 5.628140703517588, + "grad_norm": 0.4273395836353302, + "learning_rate": 4.606701940035274e-06, + "loss": 0.0078, + "step": 3360 + }, + { + "epoch": 5.661641541038526, + "grad_norm": 0.6737497448921204, + "learning_rate": 4.571428571428572e-06, + "loss": 0.0106, + "step": 3380 + }, + { + "epoch": 5.695142378559464, + "grad_norm": 1.0791343450546265, + "learning_rate": 4.53615520282187e-06, + "loss": 0.0097, + "step": 3400 + }, + { + "epoch": 5.728643216080402, + "grad_norm": 1.0891772508621216, + "learning_rate": 4.500881834215168e-06, + "loss": 0.0109, + "step": 3420 + }, + { + "epoch": 5.76214405360134, + "grad_norm": 0.7465157508850098, + "learning_rate": 4.465608465608466e-06, + "loss": 0.0078, + "step": 3440 + }, + { + "epoch": 5.795644891122278, + "grad_norm": 0.7693866491317749, + "learning_rate": 4.430335097001764e-06, + "loss": 0.009, + "step": 3460 + }, + { + "epoch": 5.8291457286432165, + "grad_norm": 1.3295698165893555, + "learning_rate": 4.395061728395062e-06, + "loss": 0.016, + "step": 3480 + }, + { + "epoch": 5.8626465661641545, + "grad_norm": 2.4605352878570557, + "learning_rate": 4.35978835978836e-06, + "loss": 0.0081, + "step": 3500 + }, + { + "epoch": 5.8626465661641545, + "eval_loss": 0.09377142041921616, + "eval_runtime": 1773.01, + "eval_samples_per_second": 2.694, + "eval_steps_per_second": 0.337, + "eval_wer": 5.010976424549012, + "step": 3500 + }, + { + "epoch": 5.8961474036850925, + "grad_norm": 0.8250058889389038, + "learning_rate": 4.324514991181658e-06, + "loss": 0.0108, + "step": 3520 + }, + { + "epoch": 5.9296482412060305, + "grad_norm": 1.3606537580490112, + "learning_rate": 4.289241622574956e-06, + "loss": 0.0125, + "step": 3540 + }, + { + "epoch": 5.9631490787269685, + "grad_norm": 0.6893450021743774, + "learning_rate": 4.2539682539682546e-06, + "loss": 0.0101, + "step": 3560 + }, + { + "epoch": 5.9966499162479066, + "grad_norm": 0.8129726052284241, + "learning_rate": 4.2186948853615525e-06, + "loss": 0.0095, + "step": 3580 + }, + { + "epoch": 6.030150753768845, + "grad_norm": 0.5319514274597168, + "learning_rate": 4.1834215167548505e-06, + "loss": 0.0051, + "step": 3600 + }, + { + "epoch": 6.063651591289783, + "grad_norm": 0.20368462800979614, + "learning_rate": 4.1481481481481485e-06, + "loss": 0.0049, + "step": 3620 + }, + { + "epoch": 6.097152428810721, + "grad_norm": 1.1721038818359375, + "learning_rate": 4.1128747795414465e-06, + "loss": 0.0041, + "step": 3640 + }, + { + "epoch": 6.130653266331659, + "grad_norm": 0.17129285633563995, + "learning_rate": 4.0776014109347444e-06, + "loss": 0.0055, + "step": 3660 + }, + { + "epoch": 6.164154103852597, + "grad_norm": 0.31987234950065613, + "learning_rate": 4.042328042328042e-06, + "loss": 0.0037, + "step": 3680 + }, + { + "epoch": 6.197654941373535, + "grad_norm": 0.3214021921157837, + "learning_rate": 4.00705467372134e-06, + "loss": 0.0052, + "step": 3700 + }, + { + "epoch": 6.231155778894473, + "grad_norm": 1.3790876865386963, + "learning_rate": 3.971781305114638e-06, + "loss": 0.0053, + "step": 3720 + }, + { + "epoch": 6.264656616415411, + "grad_norm": 0.549566924571991, + "learning_rate": 3.936507936507936e-06, + "loss": 0.0048, + "step": 3740 + }, + { + "epoch": 6.298157453936349, + "grad_norm": 0.2458494007587433, + "learning_rate": 3.901234567901235e-06, + "loss": 0.006, + "step": 3760 + }, + { + "epoch": 6.331658291457287, + "grad_norm": 2.3661324977874756, + "learning_rate": 3.865961199294533e-06, + "loss": 0.0046, + "step": 3780 + }, + { + "epoch": 6.365159128978225, + "grad_norm": 0.7839031219482422, + "learning_rate": 3.830687830687831e-06, + "loss": 0.0059, + "step": 3800 + }, + { + "epoch": 6.398659966499163, + "grad_norm": 0.2562466263771057, + "learning_rate": 3.795414462081129e-06, + "loss": 0.0038, + "step": 3820 + }, + { + "epoch": 6.432160804020101, + "grad_norm": 0.9680606126785278, + "learning_rate": 3.760141093474427e-06, + "loss": 0.0053, + "step": 3840 + }, + { + "epoch": 6.465661641541039, + "grad_norm": 0.5647270083427429, + "learning_rate": 3.724867724867725e-06, + "loss": 0.0049, + "step": 3860 + }, + { + "epoch": 6.499162479061977, + "grad_norm": 0.5850030183792114, + "learning_rate": 3.689594356261023e-06, + "loss": 0.0039, + "step": 3880 + }, + { + "epoch": 6.532663316582915, + "grad_norm": 0.1745942085981369, + "learning_rate": 3.654320987654321e-06, + "loss": 0.0035, + "step": 3900 + }, + { + "epoch": 6.566164154103853, + "grad_norm": 0.2170235961675644, + "learning_rate": 3.6190476190476194e-06, + "loss": 0.0045, + "step": 3920 + }, + { + "epoch": 6.599664991624791, + "grad_norm": 0.30363383889198303, + "learning_rate": 3.5837742504409174e-06, + "loss": 0.0031, + "step": 3940 + }, + { + "epoch": 6.633165829145729, + "grad_norm": 0.22252851724624634, + "learning_rate": 3.5485008818342153e-06, + "loss": 0.004, + "step": 3960 + }, + { + "epoch": 6.666666666666667, + "grad_norm": 0.9289886951446533, + "learning_rate": 3.5132275132275133e-06, + "loss": 0.0055, + "step": 3980 + }, + { + "epoch": 6.700167504187605, + "grad_norm": 1.094429612159729, + "learning_rate": 3.4779541446208113e-06, + "loss": 0.0041, + "step": 4000 + }, + { + "epoch": 6.700167504187605, + "eval_loss": 0.09379494935274124, + "eval_runtime": 1761.7753, + "eval_samples_per_second": 2.711, + "eval_steps_per_second": 0.339, + "eval_wer": 4.89166746205975, + "step": 4000 + }, + { + "epoch": 6.733668341708543, + "grad_norm": 0.8961842656135559, + "learning_rate": 3.4426807760141097e-06, + "loss": 0.0034, + "step": 4020 + }, + { + "epoch": 6.767169179229481, + "grad_norm": 0.45882654190063477, + "learning_rate": 3.4074074074074077e-06, + "loss": 0.0054, + "step": 4040 + }, + { + "epoch": 6.800670016750419, + "grad_norm": 0.27223679423332214, + "learning_rate": 3.3721340388007056e-06, + "loss": 0.0034, + "step": 4060 + }, + { + "epoch": 6.834170854271357, + "grad_norm": 1.3323800563812256, + "learning_rate": 3.3368606701940036e-06, + "loss": 0.0055, + "step": 4080 + }, + { + "epoch": 6.867671691792295, + "grad_norm": 0.24382267892360687, + "learning_rate": 3.3015873015873016e-06, + "loss": 0.0038, + "step": 4100 + }, + { + "epoch": 6.901172529313233, + "grad_norm": 0.2211693376302719, + "learning_rate": 3.2663139329806e-06, + "loss": 0.0043, + "step": 4120 + }, + { + "epoch": 6.934673366834171, + "grad_norm": 0.20270536839962006, + "learning_rate": 3.231040564373898e-06, + "loss": 0.0047, + "step": 4140 + }, + { + "epoch": 6.968174204355109, + "grad_norm": 2.0083487033843994, + "learning_rate": 3.195767195767196e-06, + "loss": 0.0054, + "step": 4160 + }, + { + "epoch": 7.001675041876047, + "grad_norm": 0.1442304253578186, + "learning_rate": 3.160493827160494e-06, + "loss": 0.005, + "step": 4180 + }, + { + "epoch": 7.035175879396985, + "grad_norm": 0.38336917757987976, + "learning_rate": 3.126984126984127e-06, + "loss": 0.0031, + "step": 4200 + }, + { + "epoch": 7.068676716917923, + "grad_norm": 0.13572201132774353, + "learning_rate": 3.0917107583774254e-06, + "loss": 0.0027, + "step": 4220 + }, + { + "epoch": 7.102177554438861, + "grad_norm": 1.8334097862243652, + "learning_rate": 3.0564373897707234e-06, + "loss": 0.0025, + "step": 4240 + }, + { + "epoch": 7.135678391959799, + "grad_norm": 1.0188051462173462, + "learning_rate": 3.0211640211640214e-06, + "loss": 0.0032, + "step": 4260 + }, + { + "epoch": 7.169179229480737, + "grad_norm": 0.13007956743240356, + "learning_rate": 2.9858906525573194e-06, + "loss": 0.0033, + "step": 4280 + }, + { + "epoch": 7.202680067001675, + "grad_norm": 0.07188330590724945, + "learning_rate": 2.9506172839506173e-06, + "loss": 0.0025, + "step": 4300 + }, + { + "epoch": 7.236180904522613, + "grad_norm": 0.2624744176864624, + "learning_rate": 2.9153439153439157e-06, + "loss": 0.0022, + "step": 4320 + }, + { + "epoch": 7.269681742043551, + "grad_norm": 0.1817062646150589, + "learning_rate": 2.8800705467372137e-06, + "loss": 0.0027, + "step": 4340 + }, + { + "epoch": 7.303182579564489, + "grad_norm": 0.23730872571468353, + "learning_rate": 2.8447971781305117e-06, + "loss": 0.0028, + "step": 4360 + }, + { + "epoch": 7.336683417085427, + "grad_norm": 0.44462695717811584, + "learning_rate": 2.8095238095238096e-06, + "loss": 0.0027, + "step": 4380 + }, + { + "epoch": 7.370184254606365, + "grad_norm": 0.16056309640407562, + "learning_rate": 2.7742504409171076e-06, + "loss": 0.0021, + "step": 4400 + }, + { + "epoch": 7.403685092127303, + "grad_norm": 0.1740381121635437, + "learning_rate": 2.738977072310406e-06, + "loss": 0.0021, + "step": 4420 + }, + { + "epoch": 7.437185929648241, + "grad_norm": 0.2567131519317627, + "learning_rate": 2.703703703703704e-06, + "loss": 0.0025, + "step": 4440 + }, + { + "epoch": 7.4706867671691795, + "grad_norm": 0.22708427906036377, + "learning_rate": 2.668430335097002e-06, + "loss": 0.0018, + "step": 4460 + }, + { + "epoch": 7.5041876046901175, + "grad_norm": 0.14704178273677826, + "learning_rate": 2.6331569664903e-06, + "loss": 0.0023, + "step": 4480 + }, + { + "epoch": 7.5376884422110555, + "grad_norm": 0.26313552260398865, + "learning_rate": 2.597883597883598e-06, + "loss": 0.0026, + "step": 4500 + }, + { + "epoch": 7.5376884422110555, + "eval_loss": 0.09553142637014389, + "eval_runtime": 1387.1198, + "eval_samples_per_second": 3.443, + "eval_steps_per_second": 0.43, + "eval_wer": 4.50510642359454, + "step": 4500 + } + ], + "logging_steps": 20, + "max_steps": 5970, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "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": 4.155023735267328e+19, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-4500/training_args.bin b/checkpoint-4500/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6ffec31ca17b32691fb4cd1278f73919f3259cf4 --- /dev/null +++ b/checkpoint-4500/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41a3c645624f7463029286b7d5e1ad6924cfc44a0d153880431b77c40e5deb73 +size 5432 diff --git a/checkpoint-500/config.json b/checkpoint-500/config.json new file mode 100644 index 0000000000000000000000000000000000000000..337642e18341c758364c32c6b6159a090615329b --- /dev/null +++ b/checkpoint-500/config.json @@ -0,0 +1,47 @@ +{ + "activation_dropout": 0.0, + "activation_function": "gelu", + "apply_spec_augment": false, + "architectures": [ + "WhisperForConditionalGeneration" + ], + "attention_dropout": 0.0, + "begin_suppress_tokens": null, + "bos_token_id": 50257, + "classifier_proj_size": 256, + "d_model": 768, + "decoder_attention_heads": 12, + "decoder_ffn_dim": 3072, + "decoder_layerdrop": 0.0, + "decoder_layers": 12, + "decoder_start_token_id": 50258, + "dropout": 0.0, + "encoder_attention_heads": 12, + "encoder_ffn_dim": 3072, + "encoder_layerdrop": 0.0, + "encoder_layers": 12, + "eos_token_id": 50257, + "forced_decoder_ids": null, + "init_std": 0.02, + "is_encoder_decoder": true, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "max_length": null, + "max_source_positions": 1500, + "max_target_positions": 448, + "median_filter_width": 7, + "model_type": "whisper", + "num_hidden_layers": 12, + "num_mel_bins": 80, + "pad_token_id": 50257, + "scale_embedding": false, + "torch_dtype": "float32", + "transformers_version": "4.51.3", + "use_cache": false, + "use_weighted_layer_sum": false, + "vocab_size": 51865 +} diff --git a/checkpoint-500/generation_config.json b/checkpoint-500/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..e1f4120825a303544ca23708d9dc52bbaa1dce16 --- /dev/null +++ b/checkpoint-500/generation_config.json @@ -0,0 +1,9 @@ +{ + "bos_token_id": 50257, + "decoder_start_token_id": 50258, + "eos_token_id": 50257, + "pad_token_id": 50257, + "suppress_tokens": [], + "transformers_version": "4.51.3", + "use_cache": false +} diff --git a/checkpoint-500/model.safetensors b/checkpoint-500/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..1e8db236c251967aa16515bec48076b99f8b8d64 --- /dev/null +++ b/checkpoint-500/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:999245c32c88f60935c0f92a6247a5dda8d6458efeb58378a807b28e368de7cd +size 966995080 diff --git a/checkpoint-500/optimizer.pt b/checkpoint-500/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..cd4104a18b67a407229d7c421625489ca737918f --- /dev/null +++ b/checkpoint-500/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:551bd4680b6865861681f2dba63b632a9d2b9c77d8c664ad9a2c92ab68d4ad01 +size 1925064044 diff --git a/checkpoint-500/preprocessor_config.json b/checkpoint-500/preprocessor_config.json new file mode 100644 index 0000000000000000000000000000000000000000..d7c5b1dfd466c815664b16139e8f605620d22e4e --- /dev/null +++ b/checkpoint-500/preprocessor_config.json @@ -0,0 +1,15 @@ +{ + "chunk_length": 30, + "dither": 0.0, + "feature_extractor_type": "WhisperFeatureExtractor", + "feature_size": 80, + "hop_length": 160, + "n_fft": 400, + "n_samples": 480000, + "nb_max_frames": 3000, + "padding_side": "right", + "padding_value": 0.0, + "processor_class": "WhisperProcessor", + "return_attention_mask": false, + "sampling_rate": 16000 +} diff --git a/checkpoint-500/rng_state.pth b/checkpoint-500/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..5ae35093e25f5605c6453c2c15f0421e1369a22f --- /dev/null +++ b/checkpoint-500/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77a953c1b48670da1bc7db6257f23ac91e47f20593c5406b86aa1a7c77f1bd47 +size 14244 diff --git a/checkpoint-500/scaler.pt b/checkpoint-500/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..098832f95b15985778c2e66d80ae4f9bb56abda6 --- /dev/null +++ b/checkpoint-500/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d0eedba10306fcce7794de00b74f743d1bfca325d3acb93013ff1f0ca6b6275 +size 988 diff --git a/checkpoint-500/scheduler.pt b/checkpoint-500/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..207c8c0c319e1be18451853a7b59e21667e5b4db --- /dev/null +++ b/checkpoint-500/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fa583b5ecf0d07cb085d49a5571a9b29c14481aa819499b0475c87fc8f6ebf68 +size 1064 diff --git a/checkpoint-500/trainer_state.json b/checkpoint-500/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..ee8170e7b32b3983b9ed1fac3fc6a21a15a9d356 --- /dev/null +++ b/checkpoint-500/trainer_state.json @@ -0,0 +1,218 @@ +{ + "best_global_step": 500, + "best_metric": 19.59291781998664, + "best_model_checkpoint": "./SALAMA_C3/checkpoint-500", + "epoch": 0.8375209380234506, + "eval_steps": 500, + "global_step": 500, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.03350083752093802, + "grad_norm": 11.757207870483398, + "learning_rate": 6.333333333333334e-07, + "loss": 0.5112, + "step": 20 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 10.197953224182129, + "learning_rate": 1.3e-06, + "loss": 0.5283, + "step": 40 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 12.811074256896973, + "learning_rate": 1.9666666666666668e-06, + "loss": 0.474, + "step": 60 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 8.559342384338379, + "learning_rate": 2.6e-06, + "loss": 0.424, + "step": 80 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 6.9629225730896, + "learning_rate": 3.266666666666667e-06, + "loss": 0.4465, + "step": 100 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 7.833058834075928, + "learning_rate": 3.9333333333333335e-06, + "loss": 0.3764, + "step": 120 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 6.882424831390381, + "learning_rate": 4.600000000000001e-06, + "loss": 0.3705, + "step": 140 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 5.4183244705200195, + "learning_rate": 5.2666666666666665e-06, + "loss": 0.4116, + "step": 160 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 6.103787899017334, + "learning_rate": 5.933333333333335e-06, + "loss": 0.3903, + "step": 180 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 6.084160804748535, + "learning_rate": 6.600000000000001e-06, + "loss": 0.3794, + "step": 200 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 4.531406879425049, + "learning_rate": 7.266666666666668e-06, + "loss": 0.3904, + "step": 220 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 6.702854156494141, + "learning_rate": 7.933333333333334e-06, + "loss": 0.3672, + "step": 240 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 6.386377811431885, + "learning_rate": 8.6e-06, + "loss": 0.3379, + "step": 260 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 5.33281135559082, + "learning_rate": 9.266666666666667e-06, + "loss": 0.4102, + "step": 280 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 6.031435012817383, + "learning_rate": 9.933333333333334e-06, + "loss": 0.3815, + "step": 300 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 4.8638176918029785, + "learning_rate": 9.968253968253969e-06, + "loss": 0.358, + "step": 320 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 5.352113723754883, + "learning_rate": 9.932980599647268e-06, + "loss": 0.3578, + "step": 340 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 5.560739040374756, + "learning_rate": 9.897707231040565e-06, + "loss": 0.3795, + "step": 360 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 6.186940670013428, + "learning_rate": 9.862433862433864e-06, + "loss": 0.3852, + "step": 380 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 5.37507438659668, + "learning_rate": 9.827160493827161e-06, + "loss": 0.3761, + "step": 400 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 6.373449802398682, + "learning_rate": 9.79188712522046e-06, + "loss": 0.3674, + "step": 420 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 5.702625274658203, + "learning_rate": 9.756613756613757e-06, + "loss": 0.4013, + "step": 440 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 5.5675153732299805, + "learning_rate": 9.721340388007056e-06, + "loss": 0.3495, + "step": 460 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 6.296374320983887, + "learning_rate": 9.686067019400353e-06, + "loss": 0.3704, + "step": 480 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 4.830463886260986, + "learning_rate": 9.650793650793652e-06, + "loss": 0.36, + "step": 500 + }, + { + "epoch": 0.8375209380234506, + "eval_loss": 0.27764827013015747, + "eval_runtime": 1752.3826, + "eval_samples_per_second": 2.725, + "eval_steps_per_second": 0.341, + "eval_wer": 19.59291781998664, + "step": 500 + } + ], + "logging_steps": 20, + "max_steps": 5970, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "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": 4.61736640512e+18, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-500/training_args.bin b/checkpoint-500/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6ffec31ca17b32691fb4cd1278f73919f3259cf4 --- /dev/null +++ b/checkpoint-500/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41a3c645624f7463029286b7d5e1ad6924cfc44a0d153880431b77c40e5deb73 +size 5432 diff --git a/checkpoint-5000/config.json b/checkpoint-5000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..337642e18341c758364c32c6b6159a090615329b --- /dev/null +++ b/checkpoint-5000/config.json @@ -0,0 +1,47 @@ +{ + "activation_dropout": 0.0, + "activation_function": "gelu", + "apply_spec_augment": false, + "architectures": [ + "WhisperForConditionalGeneration" + ], + "attention_dropout": 0.0, + "begin_suppress_tokens": null, + "bos_token_id": 50257, + "classifier_proj_size": 256, + "d_model": 768, + "decoder_attention_heads": 12, + "decoder_ffn_dim": 3072, + "decoder_layerdrop": 0.0, + "decoder_layers": 12, + "decoder_start_token_id": 50258, + "dropout": 0.0, + "encoder_attention_heads": 12, + "encoder_ffn_dim": 3072, + "encoder_layerdrop": 0.0, + "encoder_layers": 12, + "eos_token_id": 50257, + "forced_decoder_ids": null, + "init_std": 0.02, + "is_encoder_decoder": true, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "max_length": null, + "max_source_positions": 1500, + "max_target_positions": 448, + "median_filter_width": 7, + "model_type": "whisper", + "num_hidden_layers": 12, + "num_mel_bins": 80, + "pad_token_id": 50257, + "scale_embedding": false, + "torch_dtype": "float32", + "transformers_version": "4.51.3", + "use_cache": false, + "use_weighted_layer_sum": false, + "vocab_size": 51865 +} diff --git a/checkpoint-5000/generation_config.json b/checkpoint-5000/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..e1f4120825a303544ca23708d9dc52bbaa1dce16 --- /dev/null +++ b/checkpoint-5000/generation_config.json @@ -0,0 +1,9 @@ +{ + "bos_token_id": 50257, + "decoder_start_token_id": 50258, + "eos_token_id": 50257, + "pad_token_id": 50257, + "suppress_tokens": [], + "transformers_version": "4.51.3", + "use_cache": false +} diff --git a/checkpoint-5000/model.safetensors b/checkpoint-5000/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..a1fe88ae18af27445bc0bc786aeb71e4770b532d --- /dev/null +++ b/checkpoint-5000/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7be9865bb28c24a1db84e0269d5db00b1a0daeff83a9c6c5829dcb3456046890 +size 966995080 diff --git a/checkpoint-5000/optimizer.pt b/checkpoint-5000/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..f21cdf060c0086a34012731eb3b5b634fabe8719 --- /dev/null +++ b/checkpoint-5000/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:994a78b5eed7073b08c39cb2b15ea273bf2df9317de70e94f4cbb996655c4826 +size 1925064044 diff --git a/checkpoint-5000/preprocessor_config.json b/checkpoint-5000/preprocessor_config.json new file mode 100644 index 0000000000000000000000000000000000000000..d7c5b1dfd466c815664b16139e8f605620d22e4e --- /dev/null +++ b/checkpoint-5000/preprocessor_config.json @@ -0,0 +1,15 @@ +{ + "chunk_length": 30, + "dither": 0.0, + "feature_extractor_type": "WhisperFeatureExtractor", + "feature_size": 80, + "hop_length": 160, + "n_fft": 400, + "n_samples": 480000, + "nb_max_frames": 3000, + "padding_side": "right", + "padding_value": 0.0, + "processor_class": "WhisperProcessor", + "return_attention_mask": false, + "sampling_rate": 16000 +} diff --git a/checkpoint-5000/rng_state.pth b/checkpoint-5000/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..5d8b3366322320f7a4a4bdfb9e6aa809799b8fd6 --- /dev/null +++ b/checkpoint-5000/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:baf8c29032e4b3df5a311b9c5b06a929756e3b8f3b03f1e28116cd3a951b1fbf +size 14244 diff --git a/checkpoint-5000/scaler.pt b/checkpoint-5000/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..ee6e5ba226a5f9bb62dfc957d513a343c35968a9 --- /dev/null +++ b/checkpoint-5000/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:902a9a5afd4d79b4c161df90caed486dc5d0979929a322391f1ec6b2527e2697 +size 988 diff --git a/checkpoint-5000/scheduler.pt b/checkpoint-5000/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..d3e189586c59d0aa38161cedb883ee2677cfa25b --- /dev/null +++ b/checkpoint-5000/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c46ec5ae96bc2c3ff06c24ed20c506ae88bbe98194aa0ff5d382c9885d331871 +size 1064 diff --git a/checkpoint-5000/trainer_state.json b/checkpoint-5000/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..d0dee935cce4d8f66187804de2b1f0b7a2396a3a --- /dev/null +++ b/checkpoint-5000/trainer_state.json @@ -0,0 +1,1874 @@ +{ + "best_global_step": 5000, + "best_metric": 4.457382838598836, + "best_model_checkpoint": "./SALAMA_C3/checkpoint-5000", + "epoch": 8.375209380234505, + "eval_steps": 500, + "global_step": 5000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.03350083752093802, + "grad_norm": 11.757207870483398, + "learning_rate": 6.333333333333334e-07, + "loss": 0.5112, + "step": 20 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 10.197953224182129, + "learning_rate": 1.3e-06, + "loss": 0.5283, + "step": 40 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 12.811074256896973, + "learning_rate": 1.9666666666666668e-06, + "loss": 0.474, + "step": 60 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 8.559342384338379, + "learning_rate": 2.6e-06, + "loss": 0.424, + "step": 80 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 6.9629225730896, + "learning_rate": 3.266666666666667e-06, + "loss": 0.4465, + "step": 100 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 7.833058834075928, + "learning_rate": 3.9333333333333335e-06, + "loss": 0.3764, + "step": 120 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 6.882424831390381, + "learning_rate": 4.600000000000001e-06, + "loss": 0.3705, + "step": 140 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 5.4183244705200195, + "learning_rate": 5.2666666666666665e-06, + "loss": 0.4116, + "step": 160 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 6.103787899017334, + "learning_rate": 5.933333333333335e-06, + "loss": 0.3903, + "step": 180 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 6.084160804748535, + "learning_rate": 6.600000000000001e-06, + "loss": 0.3794, + "step": 200 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 4.531406879425049, + "learning_rate": 7.266666666666668e-06, + "loss": 0.3904, + "step": 220 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 6.702854156494141, + "learning_rate": 7.933333333333334e-06, + "loss": 0.3672, + "step": 240 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 6.386377811431885, + "learning_rate": 8.6e-06, + "loss": 0.3379, + "step": 260 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 5.33281135559082, + "learning_rate": 9.266666666666667e-06, + "loss": 0.4102, + "step": 280 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 6.031435012817383, + "learning_rate": 9.933333333333334e-06, + "loss": 0.3815, + "step": 300 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 4.8638176918029785, + "learning_rate": 9.968253968253969e-06, + "loss": 0.358, + "step": 320 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 5.352113723754883, + "learning_rate": 9.932980599647268e-06, + "loss": 0.3578, + "step": 340 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 5.560739040374756, + "learning_rate": 9.897707231040565e-06, + "loss": 0.3795, + "step": 360 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 6.186940670013428, + "learning_rate": 9.862433862433864e-06, + "loss": 0.3852, + "step": 380 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 5.37507438659668, + "learning_rate": 9.827160493827161e-06, + "loss": 0.3761, + "step": 400 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 6.373449802398682, + "learning_rate": 9.79188712522046e-06, + "loss": 0.3674, + "step": 420 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 5.702625274658203, + "learning_rate": 9.756613756613757e-06, + "loss": 0.4013, + "step": 440 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 5.5675153732299805, + "learning_rate": 9.721340388007056e-06, + "loss": 0.3495, + "step": 460 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 6.296374320983887, + "learning_rate": 9.686067019400353e-06, + "loss": 0.3704, + "step": 480 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 4.830463886260986, + "learning_rate": 9.650793650793652e-06, + "loss": 0.36, + "step": 500 + }, + { + "epoch": 0.8375209380234506, + "eval_loss": 0.27764827013015747, + "eval_runtime": 1752.3826, + "eval_samples_per_second": 2.725, + "eval_steps_per_second": 0.341, + "eval_wer": 19.59291781998664, + "step": 500 + }, + { + "epoch": 0.8710217755443886, + "grad_norm": 4.863058090209961, + "learning_rate": 9.61552028218695e-06, + "loss": 0.3874, + "step": 520 + }, + { + "epoch": 0.9045226130653267, + "grad_norm": 6.121027946472168, + "learning_rate": 9.580246913580248e-06, + "loss": 0.382, + "step": 540 + }, + { + "epoch": 0.9380234505862647, + "grad_norm": 4.538082599639893, + "learning_rate": 9.544973544973546e-06, + "loss": 0.3611, + "step": 560 + }, + { + "epoch": 0.9715242881072027, + "grad_norm": 4.8772382736206055, + "learning_rate": 9.509700176366844e-06, + "loss": 0.3584, + "step": 580 + }, + { + "epoch": 1.0050251256281406, + "grad_norm": 4.039212703704834, + "learning_rate": 9.474426807760142e-06, + "loss": 0.3701, + "step": 600 + }, + { + "epoch": 1.0385259631490786, + "grad_norm": 4.787687301635742, + "learning_rate": 9.43915343915344e-06, + "loss": 0.2058, + "step": 620 + }, + { + "epoch": 1.0720268006700167, + "grad_norm": 4.503021717071533, + "learning_rate": 9.403880070546738e-06, + "loss": 0.2455, + "step": 640 + }, + { + "epoch": 1.1055276381909547, + "grad_norm": 6.663857936859131, + "learning_rate": 9.368606701940036e-06, + "loss": 0.2179, + "step": 660 + }, + { + "epoch": 1.1390284757118927, + "grad_norm": 4.486196041107178, + "learning_rate": 9.333333333333334e-06, + "loss": 0.2196, + "step": 680 + }, + { + "epoch": 1.1725293132328307, + "grad_norm": 4.667060852050781, + "learning_rate": 9.298059964726633e-06, + "loss": 0.2183, + "step": 700 + }, + { + "epoch": 1.2060301507537687, + "grad_norm": 5.608316898345947, + "learning_rate": 9.26278659611993e-06, + "loss": 0.2161, + "step": 720 + }, + { + "epoch": 1.2395309882747068, + "grad_norm": 4.2184271812438965, + "learning_rate": 9.227513227513229e-06, + "loss": 0.2382, + "step": 740 + }, + { + "epoch": 1.2730318257956448, + "grad_norm": 3.871945381164551, + "learning_rate": 9.192239858906526e-06, + "loss": 0.2214, + "step": 760 + }, + { + "epoch": 1.3065326633165828, + "grad_norm": 3.730222225189209, + "learning_rate": 9.156966490299825e-06, + "loss": 0.2213, + "step": 780 + }, + { + "epoch": 1.3400335008375208, + "grad_norm": 4.740777015686035, + "learning_rate": 9.121693121693122e-06, + "loss": 0.2503, + "step": 800 + }, + { + "epoch": 1.3735343383584588, + "grad_norm": 4.123469352722168, + "learning_rate": 9.086419753086421e-06, + "loss": 0.2443, + "step": 820 + }, + { + "epoch": 1.4070351758793969, + "grad_norm": 3.988917827606201, + "learning_rate": 9.051146384479718e-06, + "loss": 0.241, + "step": 840 + }, + { + "epoch": 1.4405360134003349, + "grad_norm": 5.4730305671691895, + "learning_rate": 9.015873015873017e-06, + "loss": 0.2163, + "step": 860 + }, + { + "epoch": 1.474036850921273, + "grad_norm": 5.020652770996094, + "learning_rate": 8.980599647266314e-06, + "loss": 0.2387, + "step": 880 + }, + { + "epoch": 1.507537688442211, + "grad_norm": 4.231486797332764, + "learning_rate": 8.945326278659613e-06, + "loss": 0.2523, + "step": 900 + }, + { + "epoch": 1.541038525963149, + "grad_norm": 6.197975158691406, + "learning_rate": 8.910052910052912e-06, + "loss": 0.2255, + "step": 920 + }, + { + "epoch": 1.574539363484087, + "grad_norm": 5.489210605621338, + "learning_rate": 8.874779541446209e-06, + "loss": 0.2397, + "step": 940 + }, + { + "epoch": 1.608040201005025, + "grad_norm": 4.32737398147583, + "learning_rate": 8.839506172839508e-06, + "loss": 0.2298, + "step": 960 + }, + { + "epoch": 1.641541038525963, + "grad_norm": 4.504214763641357, + "learning_rate": 8.804232804232805e-06, + "loss": 0.2301, + "step": 980 + }, + { + "epoch": 1.675041876046901, + "grad_norm": 3.9694759845733643, + "learning_rate": 8.768959435626104e-06, + "loss": 0.2338, + "step": 1000 + }, + { + "epoch": 1.675041876046901, + "eval_loss": 0.18838582932949066, + "eval_runtime": 1770.3449, + "eval_samples_per_second": 2.698, + "eval_steps_per_second": 0.337, + "eval_wer": 14.071299035983584, + "step": 1000 + }, + { + "epoch": 1.708542713567839, + "grad_norm": 4.972934722900391, + "learning_rate": 8.7336860670194e-06, + "loss": 0.2388, + "step": 1020 + }, + { + "epoch": 1.742043551088777, + "grad_norm": 5.178994655609131, + "learning_rate": 8.6984126984127e-06, + "loss": 0.2531, + "step": 1040 + }, + { + "epoch": 1.775544388609715, + "grad_norm": 3.560372829437256, + "learning_rate": 8.663139329805997e-06, + "loss": 0.2187, + "step": 1060 + }, + { + "epoch": 1.809045226130653, + "grad_norm": 4.647324562072754, + "learning_rate": 8.627865961199296e-06, + "loss": 0.2413, + "step": 1080 + }, + { + "epoch": 1.8425460636515911, + "grad_norm": 3.6784422397613525, + "learning_rate": 8.592592592592593e-06, + "loss": 0.2242, + "step": 1100 + }, + { + "epoch": 1.8760469011725294, + "grad_norm": 5.928018093109131, + "learning_rate": 8.557319223985891e-06, + "loss": 0.2406, + "step": 1120 + }, + { + "epoch": 1.9095477386934674, + "grad_norm": 6.358487606048584, + "learning_rate": 8.52204585537919e-06, + "loss": 0.2482, + "step": 1140 + }, + { + "epoch": 1.9430485762144054, + "grad_norm": 4.409506797790527, + "learning_rate": 8.486772486772487e-06, + "loss": 0.2415, + "step": 1160 + }, + { + "epoch": 1.9765494137353434, + "grad_norm": 5.169639587402344, + "learning_rate": 8.451499118165786e-06, + "loss": 0.222, + "step": 1180 + }, + { + "epoch": 2.0100502512562812, + "grad_norm": 2.8841328620910645, + "learning_rate": 8.416225749559083e-06, + "loss": 0.2109, + "step": 1200 + }, + { + "epoch": 2.0435510887772192, + "grad_norm": 3.835498332977295, + "learning_rate": 8.380952380952382e-06, + "loss": 0.1284, + "step": 1220 + }, + { + "epoch": 2.0770519262981573, + "grad_norm": 2.584859848022461, + "learning_rate": 8.34567901234568e-06, + "loss": 0.1059, + "step": 1240 + }, + { + "epoch": 2.1105527638190953, + "grad_norm": 2.797682285308838, + "learning_rate": 8.310405643738978e-06, + "loss": 0.0967, + "step": 1260 + }, + { + "epoch": 2.1440536013400333, + "grad_norm": 3.2937309741973877, + "learning_rate": 8.275132275132275e-06, + "loss": 0.1267, + "step": 1280 + }, + { + "epoch": 2.1775544388609713, + "grad_norm": 3.344325065612793, + "learning_rate": 8.239858906525574e-06, + "loss": 0.1354, + "step": 1300 + }, + { + "epoch": 2.2110552763819094, + "grad_norm": 2.955522060394287, + "learning_rate": 8.204585537918873e-06, + "loss": 0.104, + "step": 1320 + }, + { + "epoch": 2.2445561139028474, + "grad_norm": 5.180908203125, + "learning_rate": 8.16931216931217e-06, + "loss": 0.1229, + "step": 1340 + }, + { + "epoch": 2.2780569514237854, + "grad_norm": 3.639417886734009, + "learning_rate": 8.134038800705469e-06, + "loss": 0.104, + "step": 1360 + }, + { + "epoch": 2.3115577889447234, + "grad_norm": 3.7380902767181396, + "learning_rate": 8.098765432098766e-06, + "loss": 0.1141, + "step": 1380 + }, + { + "epoch": 2.3450586264656614, + "grad_norm": 4.100318908691406, + "learning_rate": 8.063492063492065e-06, + "loss": 0.1123, + "step": 1400 + }, + { + "epoch": 2.3785594639865995, + "grad_norm": 2.8147048950195312, + "learning_rate": 8.028218694885362e-06, + "loss": 0.1039, + "step": 1420 + }, + { + "epoch": 2.4120603015075375, + "grad_norm": 3.7376410961151123, + "learning_rate": 7.99294532627866e-06, + "loss": 0.1179, + "step": 1440 + }, + { + "epoch": 2.4455611390284755, + "grad_norm": 4.320065975189209, + "learning_rate": 7.957671957671958e-06, + "loss": 0.1167, + "step": 1460 + }, + { + "epoch": 2.4790619765494135, + "grad_norm": 4.415127277374268, + "learning_rate": 7.922398589065257e-06, + "loss": 0.125, + "step": 1480 + }, + { + "epoch": 2.5125628140703515, + "grad_norm": 2.631763219833374, + "learning_rate": 7.887125220458554e-06, + "loss": 0.1188, + "step": 1500 + }, + { + "epoch": 2.5125628140703515, + "eval_loss": 0.13635103404521942, + "eval_runtime": 1806.1099, + "eval_samples_per_second": 2.644, + "eval_steps_per_second": 0.331, + "eval_wer": 11.713753937195762, + "step": 1500 + }, + { + "epoch": 2.5460636515912896, + "grad_norm": 3.608773946762085, + "learning_rate": 7.851851851851853e-06, + "loss": 0.1222, + "step": 1520 + }, + { + "epoch": 2.5795644891122276, + "grad_norm": 3.4494569301605225, + "learning_rate": 7.816578483245151e-06, + "loss": 0.1087, + "step": 1540 + }, + { + "epoch": 2.6130653266331656, + "grad_norm": 3.6112048625946045, + "learning_rate": 7.781305114638449e-06, + "loss": 0.1097, + "step": 1560 + }, + { + "epoch": 2.6465661641541036, + "grad_norm": 3.1978707313537598, + "learning_rate": 7.746031746031747e-06, + "loss": 0.1101, + "step": 1580 + }, + { + "epoch": 2.6800670016750416, + "grad_norm": 3.6376733779907227, + "learning_rate": 7.710758377425045e-06, + "loss": 0.1056, + "step": 1600 + }, + { + "epoch": 2.7135678391959797, + "grad_norm": 3.754915237426758, + "learning_rate": 7.675485008818343e-06, + "loss": 0.1103, + "step": 1620 + }, + { + "epoch": 2.7470686767169177, + "grad_norm": 3.329240322113037, + "learning_rate": 7.64021164021164e-06, + "loss": 0.1193, + "step": 1640 + }, + { + "epoch": 2.7805695142378557, + "grad_norm": 3.9300169944763184, + "learning_rate": 7.604938271604939e-06, + "loss": 0.1232, + "step": 1660 + }, + { + "epoch": 2.8140703517587937, + "grad_norm": 4.290626049041748, + "learning_rate": 7.569664902998237e-06, + "loss": 0.1374, + "step": 1680 + }, + { + "epoch": 2.8475711892797317, + "grad_norm": 3.8987998962402344, + "learning_rate": 7.534391534391535e-06, + "loss": 0.1062, + "step": 1700 + }, + { + "epoch": 2.8810720268006698, + "grad_norm": 4.234960079193115, + "learning_rate": 7.499118165784833e-06, + "loss": 0.111, + "step": 1720 + }, + { + "epoch": 2.914572864321608, + "grad_norm": 3.103458881378174, + "learning_rate": 7.463844797178131e-06, + "loss": 0.1076, + "step": 1740 + }, + { + "epoch": 2.948073701842546, + "grad_norm": 3.270204782485962, + "learning_rate": 7.428571428571429e-06, + "loss": 0.1224, + "step": 1760 + }, + { + "epoch": 2.981574539363484, + "grad_norm": 4.261337757110596, + "learning_rate": 7.393298059964727e-06, + "loss": 0.1224, + "step": 1780 + }, + { + "epoch": 3.0150753768844223, + "grad_norm": 1.9531301259994507, + "learning_rate": 7.358024691358025e-06, + "loss": 0.0913, + "step": 1800 + }, + { + "epoch": 3.0485762144053603, + "grad_norm": 1.866215467453003, + "learning_rate": 7.322751322751324e-06, + "loss": 0.0466, + "step": 1820 + }, + { + "epoch": 3.0820770519262983, + "grad_norm": 3.9456610679626465, + "learning_rate": 7.287477954144622e-06, + "loss": 0.0451, + "step": 1840 + }, + { + "epoch": 3.1155778894472363, + "grad_norm": 1.9518849849700928, + "learning_rate": 7.25220458553792e-06, + "loss": 0.0414, + "step": 1860 + }, + { + "epoch": 3.1490787269681744, + "grad_norm": 2.232792854309082, + "learning_rate": 7.216931216931218e-06, + "loss": 0.0502, + "step": 1880 + }, + { + "epoch": 3.1825795644891124, + "grad_norm": 2.074127674102783, + "learning_rate": 7.181657848324516e-06, + "loss": 0.057, + "step": 1900 + }, + { + "epoch": 3.2160804020100504, + "grad_norm": 3.5036733150482178, + "learning_rate": 7.146384479717814e-06, + "loss": 0.0515, + "step": 1920 + }, + { + "epoch": 3.2495812395309884, + "grad_norm": 2.324014186859131, + "learning_rate": 7.111111111111112e-06, + "loss": 0.0454, + "step": 1940 + }, + { + "epoch": 3.2830820770519265, + "grad_norm": 2.710326671600342, + "learning_rate": 7.07583774250441e-06, + "loss": 0.0473, + "step": 1960 + }, + { + "epoch": 3.3165829145728645, + "grad_norm": 2.8943896293640137, + "learning_rate": 7.040564373897708e-06, + "loss": 0.046, + "step": 1980 + }, + { + "epoch": 3.3500837520938025, + "grad_norm": 2.319986581802368, + "learning_rate": 7.005291005291006e-06, + "loss": 0.0449, + "step": 2000 + }, + { + "epoch": 3.3500837520938025, + "eval_loss": 0.10934468358755112, + "eval_runtime": 1762.4506, + "eval_samples_per_second": 2.71, + "eval_steps_per_second": 0.339, + "eval_wer": 7.172854824854443, + "step": 2000 + }, + { + "epoch": 3.3835845896147405, + "grad_norm": 3.317129135131836, + "learning_rate": 6.9700176366843046e-06, + "loss": 0.0645, + "step": 2020 + }, + { + "epoch": 3.4170854271356785, + "grad_norm": 1.9533768892288208, + "learning_rate": 6.9347442680776025e-06, + "loss": 0.0432, + "step": 2040 + }, + { + "epoch": 3.4505862646566166, + "grad_norm": 2.0508453845977783, + "learning_rate": 6.8994708994709005e-06, + "loss": 0.0521, + "step": 2060 + }, + { + "epoch": 3.4840871021775546, + "grad_norm": 2.163236141204834, + "learning_rate": 6.8641975308641985e-06, + "loss": 0.0529, + "step": 2080 + }, + { + "epoch": 3.5175879396984926, + "grad_norm": 2.7154581546783447, + "learning_rate": 6.8289241622574965e-06, + "loss": 0.0452, + "step": 2100 + }, + { + "epoch": 3.5510887772194306, + "grad_norm": 3.0822432041168213, + "learning_rate": 6.7936507936507944e-06, + "loss": 0.0546, + "step": 2120 + }, + { + "epoch": 3.5845896147403686, + "grad_norm": 4.19010591506958, + "learning_rate": 6.758377425044092e-06, + "loss": 0.0529, + "step": 2140 + }, + { + "epoch": 3.6180904522613067, + "grad_norm": 2.9883594512939453, + "learning_rate": 6.72310405643739e-06, + "loss": 0.0503, + "step": 2160 + }, + { + "epoch": 3.6515912897822447, + "grad_norm": 2.3664371967315674, + "learning_rate": 6.687830687830688e-06, + "loss": 0.0498, + "step": 2180 + }, + { + "epoch": 3.6850921273031827, + "grad_norm": 2.0549991130828857, + "learning_rate": 6.652557319223986e-06, + "loss": 0.051, + "step": 2200 + }, + { + "epoch": 3.7185929648241207, + "grad_norm": 2.5339038372039795, + "learning_rate": 6.617283950617285e-06, + "loss": 0.0568, + "step": 2220 + }, + { + "epoch": 3.7520938023450587, + "grad_norm": 1.9988099336624146, + "learning_rate": 6.582010582010583e-06, + "loss": 0.051, + "step": 2240 + }, + { + "epoch": 3.7855946398659968, + "grad_norm": 2.5243782997131348, + "learning_rate": 6.546737213403881e-06, + "loss": 0.056, + "step": 2260 + }, + { + "epoch": 3.819095477386935, + "grad_norm": 3.157158136367798, + "learning_rate": 6.511463844797179e-06, + "loss": 0.0497, + "step": 2280 + }, + { + "epoch": 3.852596314907873, + "grad_norm": 1.9286202192306519, + "learning_rate": 6.476190476190477e-06, + "loss": 0.0426, + "step": 2300 + }, + { + "epoch": 3.886097152428811, + "grad_norm": 3.808802604675293, + "learning_rate": 6.440917107583775e-06, + "loss": 0.0499, + "step": 2320 + }, + { + "epoch": 3.919597989949749, + "grad_norm": 2.506671667098999, + "learning_rate": 6.405643738977073e-06, + "loss": 0.052, + "step": 2340 + }, + { + "epoch": 3.953098827470687, + "grad_norm": 2.9451920986175537, + "learning_rate": 6.370370370370371e-06, + "loss": 0.0552, + "step": 2360 + }, + { + "epoch": 3.986599664991625, + "grad_norm": 2.592744827270508, + "learning_rate": 6.335097001763669e-06, + "loss": 0.0527, + "step": 2380 + }, + { + "epoch": 4.0201005025125625, + "grad_norm": 1.8891575336456299, + "learning_rate": 6.299823633156967e-06, + "loss": 0.0289, + "step": 2400 + }, + { + "epoch": 4.0536013400335005, + "grad_norm": 1.8053243160247803, + "learning_rate": 6.264550264550266e-06, + "loss": 0.0192, + "step": 2420 + }, + { + "epoch": 4.0871021775544385, + "grad_norm": 2.0084407329559326, + "learning_rate": 6.229276895943564e-06, + "loss": 0.0242, + "step": 2440 + }, + { + "epoch": 4.1206030150753765, + "grad_norm": 1.5919119119644165, + "learning_rate": 6.194003527336862e-06, + "loss": 0.0211, + "step": 2460 + }, + { + "epoch": 4.1541038525963145, + "grad_norm": 1.9214613437652588, + "learning_rate": 6.15873015873016e-06, + "loss": 0.0233, + "step": 2480 + }, + { + "epoch": 4.187604690117253, + "grad_norm": 1.2652311325073242, + "learning_rate": 6.123456790123458e-06, + "loss": 0.0199, + "step": 2500 + }, + { + "epoch": 4.187604690117253, + "eval_loss": 0.0981329157948494, + "eval_runtime": 1779.0213, + "eval_samples_per_second": 2.685, + "eval_steps_per_second": 0.336, + "eval_wer": 6.707549871146321, + "step": 2500 + }, + { + "epoch": 4.221105527638191, + "grad_norm": 2.026528835296631, + "learning_rate": 6.088183421516756e-06, + "loss": 0.0217, + "step": 2520 + }, + { + "epoch": 4.254606365159129, + "grad_norm": 1.596919059753418, + "learning_rate": 6.052910052910054e-06, + "loss": 0.0167, + "step": 2540 + }, + { + "epoch": 4.288107202680067, + "grad_norm": 2.9445090293884277, + "learning_rate": 6.017636684303352e-06, + "loss": 0.0225, + "step": 2560 + }, + { + "epoch": 4.321608040201005, + "grad_norm": 2.4160282611846924, + "learning_rate": 5.9823633156966496e-06, + "loss": 0.0253, + "step": 2580 + }, + { + "epoch": 4.355108877721943, + "grad_norm": 1.461127758026123, + "learning_rate": 5.9470899470899475e-06, + "loss": 0.0197, + "step": 2600 + }, + { + "epoch": 4.388609715242881, + "grad_norm": 2.7892863750457764, + "learning_rate": 5.911816578483246e-06, + "loss": 0.022, + "step": 2620 + }, + { + "epoch": 4.422110552763819, + "grad_norm": 1.651208758354187, + "learning_rate": 5.876543209876544e-06, + "loss": 0.0215, + "step": 2640 + }, + { + "epoch": 4.455611390284757, + "grad_norm": 2.2500391006469727, + "learning_rate": 5.841269841269842e-06, + "loss": 0.0247, + "step": 2660 + }, + { + "epoch": 4.489112227805695, + "grad_norm": 4.447635173797607, + "learning_rate": 5.80599647266314e-06, + "loss": 0.0263, + "step": 2680 + }, + { + "epoch": 4.522613065326633, + "grad_norm": 0.8300407528877258, + "learning_rate": 5.770723104056438e-06, + "loss": 0.0209, + "step": 2700 + }, + { + "epoch": 4.556113902847571, + "grad_norm": 1.6874111890792847, + "learning_rate": 5.735449735449736e-06, + "loss": 0.0195, + "step": 2720 + }, + { + "epoch": 4.589614740368509, + "grad_norm": 2.4045815467834473, + "learning_rate": 5.700176366843034e-06, + "loss": 0.0224, + "step": 2740 + }, + { + "epoch": 4.623115577889447, + "grad_norm": 2.3160908222198486, + "learning_rate": 5.664902998236332e-06, + "loss": 0.0179, + "step": 2760 + }, + { + "epoch": 4.656616415410385, + "grad_norm": 1.6684287786483765, + "learning_rate": 5.62962962962963e-06, + "loss": 0.0238, + "step": 2780 + }, + { + "epoch": 4.690117252931323, + "grad_norm": 1.973906397819519, + "learning_rate": 5.594356261022928e-06, + "loss": 0.0226, + "step": 2800 + }, + { + "epoch": 4.723618090452261, + "grad_norm": 2.270906686782837, + "learning_rate": 5.559082892416227e-06, + "loss": 0.0232, + "step": 2820 + }, + { + "epoch": 4.757118927973199, + "grad_norm": 1.8875011205673218, + "learning_rate": 5.523809523809525e-06, + "loss": 0.0218, + "step": 2840 + }, + { + "epoch": 4.790619765494137, + "grad_norm": 1.1312583684921265, + "learning_rate": 5.488536155202823e-06, + "loss": 0.0212, + "step": 2860 + }, + { + "epoch": 4.824120603015075, + "grad_norm": 0.864783525466919, + "learning_rate": 5.453262786596121e-06, + "loss": 0.023, + "step": 2880 + }, + { + "epoch": 4.857621440536013, + "grad_norm": 1.2935965061187744, + "learning_rate": 5.417989417989419e-06, + "loss": 0.019, + "step": 2900 + }, + { + "epoch": 4.891122278056951, + "grad_norm": 2.4576382637023926, + "learning_rate": 5.382716049382717e-06, + "loss": 0.0193, + "step": 2920 + }, + { + "epoch": 4.924623115577889, + "grad_norm": 2.71472430229187, + "learning_rate": 5.347442680776015e-06, + "loss": 0.0253, + "step": 2940 + }, + { + "epoch": 4.958123953098827, + "grad_norm": 2.84940505027771, + "learning_rate": 5.312169312169313e-06, + "loss": 0.0218, + "step": 2960 + }, + { + "epoch": 4.991624790619765, + "grad_norm": 1.8483999967575073, + "learning_rate": 5.276895943562611e-06, + "loss": 0.0226, + "step": 2980 + }, + { + "epoch": 5.025125628140704, + "grad_norm": 0.6126876473426819, + "learning_rate": 5.241622574955909e-06, + "loss": 0.0101, + "step": 3000 + }, + { + "epoch": 5.025125628140704, + "eval_loss": 0.09390027821063995, + "eval_runtime": 1767.4512, + "eval_samples_per_second": 2.702, + "eval_steps_per_second": 0.338, + "eval_wer": 5.500143170754987, + "step": 3000 + }, + { + "epoch": 5.058626465661642, + "grad_norm": 0.3711394965648651, + "learning_rate": 5.2063492063492076e-06, + "loss": 0.0119, + "step": 3020 + }, + { + "epoch": 5.09212730318258, + "grad_norm": 0.5738839507102966, + "learning_rate": 5.1710758377425055e-06, + "loss": 0.0086, + "step": 3040 + }, + { + "epoch": 5.125628140703517, + "grad_norm": 0.7609245777130127, + "learning_rate": 5.1358024691358035e-06, + "loss": 0.0093, + "step": 3060 + }, + { + "epoch": 5.159128978224456, + "grad_norm": 1.2764722108840942, + "learning_rate": 5.1005291005291015e-06, + "loss": 0.0111, + "step": 3080 + }, + { + "epoch": 5.192629815745394, + "grad_norm": 2.1169776916503906, + "learning_rate": 5.0652557319223995e-06, + "loss": 0.0107, + "step": 3100 + }, + { + "epoch": 5.226130653266332, + "grad_norm": 2.1893081665039062, + "learning_rate": 5.0299823633156974e-06, + "loss": 0.0097, + "step": 3120 + }, + { + "epoch": 5.259631490787269, + "grad_norm": 2.2419638633728027, + "learning_rate": 4.9947089947089946e-06, + "loss": 0.0098, + "step": 3140 + }, + { + "epoch": 5.293132328308207, + "grad_norm": 0.6479611992835999, + "learning_rate": 4.959435626102293e-06, + "loss": 0.0082, + "step": 3160 + }, + { + "epoch": 5.326633165829146, + "grad_norm": 0.4799642860889435, + "learning_rate": 4.924162257495591e-06, + "loss": 0.0119, + "step": 3180 + }, + { + "epoch": 5.360134003350084, + "grad_norm": 0.7716453075408936, + "learning_rate": 4.888888888888889e-06, + "loss": 0.0086, + "step": 3200 + }, + { + "epoch": 5.393634840871022, + "grad_norm": 1.2303547859191895, + "learning_rate": 4.853615520282187e-06, + "loss": 0.009, + "step": 3220 + }, + { + "epoch": 5.42713567839196, + "grad_norm": 0.6345349550247192, + "learning_rate": 4.818342151675485e-06, + "loss": 0.0085, + "step": 3240 + }, + { + "epoch": 5.460636515912898, + "grad_norm": 0.9741530418395996, + "learning_rate": 4.783068783068783e-06, + "loss": 0.0082, + "step": 3260 + }, + { + "epoch": 5.494137353433836, + "grad_norm": 1.1631624698638916, + "learning_rate": 4.747795414462081e-06, + "loss": 0.0086, + "step": 3280 + }, + { + "epoch": 5.527638190954773, + "grad_norm": 0.6502953767776489, + "learning_rate": 4.712522045855379e-06, + "loss": 0.0093, + "step": 3300 + }, + { + "epoch": 5.561139028475711, + "grad_norm": 0.7464337348937988, + "learning_rate": 4.677248677248677e-06, + "loss": 0.0086, + "step": 3320 + }, + { + "epoch": 5.594639865996649, + "grad_norm": 1.017751693725586, + "learning_rate": 4.641975308641975e-06, + "loss": 0.0091, + "step": 3340 + }, + { + "epoch": 5.628140703517588, + "grad_norm": 0.4273395836353302, + "learning_rate": 4.606701940035274e-06, + "loss": 0.0078, + "step": 3360 + }, + { + "epoch": 5.661641541038526, + "grad_norm": 0.6737497448921204, + "learning_rate": 4.571428571428572e-06, + "loss": 0.0106, + "step": 3380 + }, + { + "epoch": 5.695142378559464, + "grad_norm": 1.0791343450546265, + "learning_rate": 4.53615520282187e-06, + "loss": 0.0097, + "step": 3400 + }, + { + "epoch": 5.728643216080402, + "grad_norm": 1.0891772508621216, + "learning_rate": 4.500881834215168e-06, + "loss": 0.0109, + "step": 3420 + }, + { + "epoch": 5.76214405360134, + "grad_norm": 0.7465157508850098, + "learning_rate": 4.465608465608466e-06, + "loss": 0.0078, + "step": 3440 + }, + { + "epoch": 5.795644891122278, + "grad_norm": 0.7693866491317749, + "learning_rate": 4.430335097001764e-06, + "loss": 0.009, + "step": 3460 + }, + { + "epoch": 5.8291457286432165, + "grad_norm": 1.3295698165893555, + "learning_rate": 4.395061728395062e-06, + "loss": 0.016, + "step": 3480 + }, + { + "epoch": 5.8626465661641545, + "grad_norm": 2.4605352878570557, + "learning_rate": 4.35978835978836e-06, + "loss": 0.0081, + "step": 3500 + }, + { + "epoch": 5.8626465661641545, + "eval_loss": 0.09377142041921616, + "eval_runtime": 1773.01, + "eval_samples_per_second": 2.694, + "eval_steps_per_second": 0.337, + "eval_wer": 5.010976424549012, + "step": 3500 + }, + { + "epoch": 5.8961474036850925, + "grad_norm": 0.8250058889389038, + "learning_rate": 4.324514991181658e-06, + "loss": 0.0108, + "step": 3520 + }, + { + "epoch": 5.9296482412060305, + "grad_norm": 1.3606537580490112, + "learning_rate": 4.289241622574956e-06, + "loss": 0.0125, + "step": 3540 + }, + { + "epoch": 5.9631490787269685, + "grad_norm": 0.6893450021743774, + "learning_rate": 4.2539682539682546e-06, + "loss": 0.0101, + "step": 3560 + }, + { + "epoch": 5.9966499162479066, + "grad_norm": 0.8129726052284241, + "learning_rate": 4.2186948853615525e-06, + "loss": 0.0095, + "step": 3580 + }, + { + "epoch": 6.030150753768845, + "grad_norm": 0.5319514274597168, + "learning_rate": 4.1834215167548505e-06, + "loss": 0.0051, + "step": 3600 + }, + { + "epoch": 6.063651591289783, + "grad_norm": 0.20368462800979614, + "learning_rate": 4.1481481481481485e-06, + "loss": 0.0049, + "step": 3620 + }, + { + "epoch": 6.097152428810721, + "grad_norm": 1.1721038818359375, + "learning_rate": 4.1128747795414465e-06, + "loss": 0.0041, + "step": 3640 + }, + { + "epoch": 6.130653266331659, + "grad_norm": 0.17129285633563995, + "learning_rate": 4.0776014109347444e-06, + "loss": 0.0055, + "step": 3660 + }, + { + "epoch": 6.164154103852597, + "grad_norm": 0.31987234950065613, + "learning_rate": 4.042328042328042e-06, + "loss": 0.0037, + "step": 3680 + }, + { + "epoch": 6.197654941373535, + "grad_norm": 0.3214021921157837, + "learning_rate": 4.00705467372134e-06, + "loss": 0.0052, + "step": 3700 + }, + { + "epoch": 6.231155778894473, + "grad_norm": 1.3790876865386963, + "learning_rate": 3.971781305114638e-06, + "loss": 0.0053, + "step": 3720 + }, + { + "epoch": 6.264656616415411, + "grad_norm": 0.549566924571991, + "learning_rate": 3.936507936507936e-06, + "loss": 0.0048, + "step": 3740 + }, + { + "epoch": 6.298157453936349, + "grad_norm": 0.2458494007587433, + "learning_rate": 3.901234567901235e-06, + "loss": 0.006, + "step": 3760 + }, + { + "epoch": 6.331658291457287, + "grad_norm": 2.3661324977874756, + "learning_rate": 3.865961199294533e-06, + "loss": 0.0046, + "step": 3780 + }, + { + "epoch": 6.365159128978225, + "grad_norm": 0.7839031219482422, + "learning_rate": 3.830687830687831e-06, + "loss": 0.0059, + "step": 3800 + }, + { + "epoch": 6.398659966499163, + "grad_norm": 0.2562466263771057, + "learning_rate": 3.795414462081129e-06, + "loss": 0.0038, + "step": 3820 + }, + { + "epoch": 6.432160804020101, + "grad_norm": 0.9680606126785278, + "learning_rate": 3.760141093474427e-06, + "loss": 0.0053, + "step": 3840 + }, + { + "epoch": 6.465661641541039, + "grad_norm": 0.5647270083427429, + "learning_rate": 3.724867724867725e-06, + "loss": 0.0049, + "step": 3860 + }, + { + "epoch": 6.499162479061977, + "grad_norm": 0.5850030183792114, + "learning_rate": 3.689594356261023e-06, + "loss": 0.0039, + "step": 3880 + }, + { + "epoch": 6.532663316582915, + "grad_norm": 0.1745942085981369, + "learning_rate": 3.654320987654321e-06, + "loss": 0.0035, + "step": 3900 + }, + { + "epoch": 6.566164154103853, + "grad_norm": 0.2170235961675644, + "learning_rate": 3.6190476190476194e-06, + "loss": 0.0045, + "step": 3920 + }, + { + "epoch": 6.599664991624791, + "grad_norm": 0.30363383889198303, + "learning_rate": 3.5837742504409174e-06, + "loss": 0.0031, + "step": 3940 + }, + { + "epoch": 6.633165829145729, + "grad_norm": 0.22252851724624634, + "learning_rate": 3.5485008818342153e-06, + "loss": 0.004, + "step": 3960 + }, + { + "epoch": 6.666666666666667, + "grad_norm": 0.9289886951446533, + "learning_rate": 3.5132275132275133e-06, + "loss": 0.0055, + "step": 3980 + }, + { + "epoch": 6.700167504187605, + "grad_norm": 1.094429612159729, + "learning_rate": 3.4779541446208113e-06, + "loss": 0.0041, + "step": 4000 + }, + { + "epoch": 6.700167504187605, + "eval_loss": 0.09379494935274124, + "eval_runtime": 1761.7753, + "eval_samples_per_second": 2.711, + "eval_steps_per_second": 0.339, + "eval_wer": 4.89166746205975, + "step": 4000 + }, + { + "epoch": 6.733668341708543, + "grad_norm": 0.8961842656135559, + "learning_rate": 3.4426807760141097e-06, + "loss": 0.0034, + "step": 4020 + }, + { + "epoch": 6.767169179229481, + "grad_norm": 0.45882654190063477, + "learning_rate": 3.4074074074074077e-06, + "loss": 0.0054, + "step": 4040 + }, + { + "epoch": 6.800670016750419, + "grad_norm": 0.27223679423332214, + "learning_rate": 3.3721340388007056e-06, + "loss": 0.0034, + "step": 4060 + }, + { + "epoch": 6.834170854271357, + "grad_norm": 1.3323800563812256, + "learning_rate": 3.3368606701940036e-06, + "loss": 0.0055, + "step": 4080 + }, + { + "epoch": 6.867671691792295, + "grad_norm": 0.24382267892360687, + "learning_rate": 3.3015873015873016e-06, + "loss": 0.0038, + "step": 4100 + }, + { + "epoch": 6.901172529313233, + "grad_norm": 0.2211693376302719, + "learning_rate": 3.2663139329806e-06, + "loss": 0.0043, + "step": 4120 + }, + { + "epoch": 6.934673366834171, + "grad_norm": 0.20270536839962006, + "learning_rate": 3.231040564373898e-06, + "loss": 0.0047, + "step": 4140 + }, + { + "epoch": 6.968174204355109, + "grad_norm": 2.0083487033843994, + "learning_rate": 3.195767195767196e-06, + "loss": 0.0054, + "step": 4160 + }, + { + "epoch": 7.001675041876047, + "grad_norm": 0.1442304253578186, + "learning_rate": 3.160493827160494e-06, + "loss": 0.005, + "step": 4180 + }, + { + "epoch": 7.035175879396985, + "grad_norm": 0.38336917757987976, + "learning_rate": 3.126984126984127e-06, + "loss": 0.0031, + "step": 4200 + }, + { + "epoch": 7.068676716917923, + "grad_norm": 0.13572201132774353, + "learning_rate": 3.0917107583774254e-06, + "loss": 0.0027, + "step": 4220 + }, + { + "epoch": 7.102177554438861, + "grad_norm": 1.8334097862243652, + "learning_rate": 3.0564373897707234e-06, + "loss": 0.0025, + "step": 4240 + }, + { + "epoch": 7.135678391959799, + "grad_norm": 1.0188051462173462, + "learning_rate": 3.0211640211640214e-06, + "loss": 0.0032, + "step": 4260 + }, + { + "epoch": 7.169179229480737, + "grad_norm": 0.13007956743240356, + "learning_rate": 2.9858906525573194e-06, + "loss": 0.0033, + "step": 4280 + }, + { + "epoch": 7.202680067001675, + "grad_norm": 0.07188330590724945, + "learning_rate": 2.9506172839506173e-06, + "loss": 0.0025, + "step": 4300 + }, + { + "epoch": 7.236180904522613, + "grad_norm": 0.2624744176864624, + "learning_rate": 2.9153439153439157e-06, + "loss": 0.0022, + "step": 4320 + }, + { + "epoch": 7.269681742043551, + "grad_norm": 0.1817062646150589, + "learning_rate": 2.8800705467372137e-06, + "loss": 0.0027, + "step": 4340 + }, + { + "epoch": 7.303182579564489, + "grad_norm": 0.23730872571468353, + "learning_rate": 2.8447971781305117e-06, + "loss": 0.0028, + "step": 4360 + }, + { + "epoch": 7.336683417085427, + "grad_norm": 0.44462695717811584, + "learning_rate": 2.8095238095238096e-06, + "loss": 0.0027, + "step": 4380 + }, + { + "epoch": 7.370184254606365, + "grad_norm": 0.16056309640407562, + "learning_rate": 2.7742504409171076e-06, + "loss": 0.0021, + "step": 4400 + }, + { + "epoch": 7.403685092127303, + "grad_norm": 0.1740381121635437, + "learning_rate": 2.738977072310406e-06, + "loss": 0.0021, + "step": 4420 + }, + { + "epoch": 7.437185929648241, + "grad_norm": 0.2567131519317627, + "learning_rate": 2.703703703703704e-06, + "loss": 0.0025, + "step": 4440 + }, + { + "epoch": 7.4706867671691795, + "grad_norm": 0.22708427906036377, + "learning_rate": 2.668430335097002e-06, + "loss": 0.0018, + "step": 4460 + }, + { + "epoch": 7.5041876046901175, + "grad_norm": 0.14704178273677826, + "learning_rate": 2.6331569664903e-06, + "loss": 0.0023, + "step": 4480 + }, + { + "epoch": 7.5376884422110555, + "grad_norm": 0.26313552260398865, + "learning_rate": 2.597883597883598e-06, + "loss": 0.0026, + "step": 4500 + }, + { + "epoch": 7.5376884422110555, + "eval_loss": 0.09553142637014389, + "eval_runtime": 1387.1198, + "eval_samples_per_second": 3.443, + "eval_steps_per_second": 0.43, + "eval_wer": 4.50510642359454, + "step": 4500 + }, + { + "epoch": 7.5711892797319935, + "grad_norm": 0.09787946194410324, + "learning_rate": 2.5626102292768963e-06, + "loss": 0.0022, + "step": 4520 + }, + { + "epoch": 7.6046901172529315, + "grad_norm": 0.10577196627855301, + "learning_rate": 2.5273368606701943e-06, + "loss": 0.0023, + "step": 4540 + }, + { + "epoch": 7.63819095477387, + "grad_norm": 0.11914186924695969, + "learning_rate": 2.4920634920634923e-06, + "loss": 0.0026, + "step": 4560 + }, + { + "epoch": 7.671691792294808, + "grad_norm": 0.6797345876693726, + "learning_rate": 2.4567901234567902e-06, + "loss": 0.0028, + "step": 4580 + }, + { + "epoch": 7.705192629815746, + "grad_norm": 0.11635720729827881, + "learning_rate": 2.4215167548500882e-06, + "loss": 0.0021, + "step": 4600 + }, + { + "epoch": 7.738693467336684, + "grad_norm": 0.0952395349740982, + "learning_rate": 2.3862433862433866e-06, + "loss": 0.0021, + "step": 4620 + }, + { + "epoch": 7.772194304857622, + "grad_norm": 0.11671995371580124, + "learning_rate": 2.3509700176366846e-06, + "loss": 0.0023, + "step": 4640 + }, + { + "epoch": 7.80569514237856, + "grad_norm": 0.10185588896274567, + "learning_rate": 2.3156966490299826e-06, + "loss": 0.003, + "step": 4660 + }, + { + "epoch": 7.839195979899498, + "grad_norm": 0.11902861297130585, + "learning_rate": 2.2804232804232805e-06, + "loss": 0.0021, + "step": 4680 + }, + { + "epoch": 7.872696817420436, + "grad_norm": 0.10562069714069366, + "learning_rate": 2.2451499118165785e-06, + "loss": 0.002, + "step": 4700 + }, + { + "epoch": 7.906197654941374, + "grad_norm": 0.1272854059934616, + "learning_rate": 2.209876543209877e-06, + "loss": 0.0024, + "step": 4720 + }, + { + "epoch": 7.939698492462312, + "grad_norm": 0.14873459935188293, + "learning_rate": 2.174603174603175e-06, + "loss": 0.0021, + "step": 4740 + }, + { + "epoch": 7.97319932998325, + "grad_norm": 0.09556525200605392, + "learning_rate": 2.139329805996473e-06, + "loss": 0.0021, + "step": 4760 + }, + { + "epoch": 8.006700167504187, + "grad_norm": 0.2545607388019562, + "learning_rate": 2.104056437389771e-06, + "loss": 0.0021, + "step": 4780 + }, + { + "epoch": 8.040201005025125, + "grad_norm": 0.09803763031959534, + "learning_rate": 2.068783068783069e-06, + "loss": 0.0017, + "step": 4800 + }, + { + "epoch": 8.073701842546063, + "grad_norm": 0.6409705281257629, + "learning_rate": 2.0335097001763672e-06, + "loss": 0.0019, + "step": 4820 + }, + { + "epoch": 8.107202680067001, + "grad_norm": 0.08693050593137741, + "learning_rate": 1.998236331569665e-06, + "loss": 0.0017, + "step": 4840 + }, + { + "epoch": 8.140703517587939, + "grad_norm": 0.09300371259450912, + "learning_rate": 1.962962962962963e-06, + "loss": 0.0022, + "step": 4860 + }, + { + "epoch": 8.174204355108877, + "grad_norm": 0.09886649250984192, + "learning_rate": 1.927689594356261e-06, + "loss": 0.0016, + "step": 4880 + }, + { + "epoch": 8.207705192629815, + "grad_norm": 0.14025329053401947, + "learning_rate": 1.8924162257495593e-06, + "loss": 0.0031, + "step": 4900 + }, + { + "epoch": 8.241206030150753, + "grad_norm": 0.06884673237800598, + "learning_rate": 1.8571428571428573e-06, + "loss": 0.0014, + "step": 4920 + }, + { + "epoch": 8.274706867671691, + "grad_norm": 0.08606573939323425, + "learning_rate": 1.8218694885361555e-06, + "loss": 0.0015, + "step": 4940 + }, + { + "epoch": 8.308207705192629, + "grad_norm": 0.5949509739875793, + "learning_rate": 1.7865961199294535e-06, + "loss": 0.0017, + "step": 4960 + }, + { + "epoch": 8.341708542713567, + "grad_norm": 0.07532797753810883, + "learning_rate": 1.7513227513227514e-06, + "loss": 0.0015, + "step": 4980 + }, + { + "epoch": 8.375209380234505, + "grad_norm": 0.07208308577537537, + "learning_rate": 1.7160493827160496e-06, + "loss": 0.0016, + "step": 5000 + }, + { + "epoch": 8.375209380234505, + "eval_loss": 0.0970502495765686, + "eval_runtime": 1406.3953, + "eval_samples_per_second": 3.396, + "eval_steps_per_second": 0.424, + "eval_wer": 4.457382838598836, + "step": 5000 + } + ], + "logging_steps": 20, + "max_steps": 5970, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "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": 4.616673800159232e+19, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-5000/training_args.bin b/checkpoint-5000/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6ffec31ca17b32691fb4cd1278f73919f3259cf4 --- /dev/null +++ b/checkpoint-5000/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41a3c645624f7463029286b7d5e1ad6924cfc44a0d153880431b77c40e5deb73 +size 5432 diff --git a/checkpoint-5500/config.json b/checkpoint-5500/config.json new file mode 100644 index 0000000000000000000000000000000000000000..337642e18341c758364c32c6b6159a090615329b --- /dev/null +++ b/checkpoint-5500/config.json @@ -0,0 +1,47 @@ +{ + "activation_dropout": 0.0, + "activation_function": "gelu", + "apply_spec_augment": false, + "architectures": [ + "WhisperForConditionalGeneration" + ], + "attention_dropout": 0.0, + "begin_suppress_tokens": null, + "bos_token_id": 50257, + "classifier_proj_size": 256, + "d_model": 768, + "decoder_attention_heads": 12, + "decoder_ffn_dim": 3072, + "decoder_layerdrop": 0.0, + "decoder_layers": 12, + "decoder_start_token_id": 50258, + "dropout": 0.0, + "encoder_attention_heads": 12, + "encoder_ffn_dim": 3072, + "encoder_layerdrop": 0.0, + "encoder_layers": 12, + "eos_token_id": 50257, + "forced_decoder_ids": null, + "init_std": 0.02, + "is_encoder_decoder": true, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "max_length": null, + "max_source_positions": 1500, + "max_target_positions": 448, + "median_filter_width": 7, + "model_type": "whisper", + "num_hidden_layers": 12, + "num_mel_bins": 80, + "pad_token_id": 50257, + "scale_embedding": false, + "torch_dtype": "float32", + "transformers_version": "4.51.3", + "use_cache": false, + "use_weighted_layer_sum": false, + "vocab_size": 51865 +} diff --git a/checkpoint-5500/generation_config.json b/checkpoint-5500/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..e1f4120825a303544ca23708d9dc52bbaa1dce16 --- /dev/null +++ b/checkpoint-5500/generation_config.json @@ -0,0 +1,9 @@ +{ + "bos_token_id": 50257, + "decoder_start_token_id": 50258, + "eos_token_id": 50257, + "pad_token_id": 50257, + "suppress_tokens": [], + "transformers_version": "4.51.3", + "use_cache": false +} diff --git a/checkpoint-5500/model.safetensors b/checkpoint-5500/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..eeb0c18df8e50b82daf3761c6af3c0c27f57969c --- /dev/null +++ b/checkpoint-5500/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e6bf7dcc74450d1c6ed6458e833ac290c00630bdbeba24c6c76661a1414f56b +size 966995080 diff --git a/checkpoint-5500/optimizer.pt b/checkpoint-5500/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..b9dfd0c631b6ef5974a84fe5ef040a83a4e3b711 --- /dev/null +++ b/checkpoint-5500/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a8faa76ee7e99dc93232e9b58d5ba4eea1968d59e7cfdfc261c955d0471cb6aa +size 1925064044 diff --git a/checkpoint-5500/preprocessor_config.json b/checkpoint-5500/preprocessor_config.json new file mode 100644 index 0000000000000000000000000000000000000000..d7c5b1dfd466c815664b16139e8f605620d22e4e --- /dev/null +++ b/checkpoint-5500/preprocessor_config.json @@ -0,0 +1,15 @@ +{ + "chunk_length": 30, + "dither": 0.0, + "feature_extractor_type": "WhisperFeatureExtractor", + "feature_size": 80, + "hop_length": 160, + "n_fft": 400, + "n_samples": 480000, + "nb_max_frames": 3000, + "padding_side": "right", + "padding_value": 0.0, + "processor_class": "WhisperProcessor", + "return_attention_mask": false, + "sampling_rate": 16000 +} diff --git a/checkpoint-5500/rng_state.pth b/checkpoint-5500/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..77f45f0fee8472a7ed82bdf09e2e13f93a5c8c1c --- /dev/null +++ b/checkpoint-5500/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86b85dc2a6aee07563abc3685f6902a60d386c6729c66a25d28ed82a981a3663 +size 14244 diff --git a/checkpoint-5500/scaler.pt b/checkpoint-5500/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..1080e5bb184770cd66c7c4b3b68d4758b814d72c --- /dev/null +++ b/checkpoint-5500/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77d80ee067af1cf06514bef9c00e3d21a1093d35e2d1213ec938fa7f90e938b2 +size 988 diff --git a/checkpoint-5500/scheduler.pt b/checkpoint-5500/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..2cf71cef91da1314100950e2a76d3deaae8b9470 --- /dev/null +++ b/checkpoint-5500/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:80d700041aa39b6c7e81ac66283cd97ce315a9a01fbd86086d7e4e8d4801a2cc +size 1064 diff --git a/checkpoint-5500/trainer_state.json b/checkpoint-5500/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..3434a5c7ef64557d6c0991829931306b3536bf65 --- /dev/null +++ b/checkpoint-5500/trainer_state.json @@ -0,0 +1,2058 @@ +{ + "best_global_step": 5500, + "best_metric": 4.431134866851197, + "best_model_checkpoint": "./SALAMA_C3/checkpoint-5500", + "epoch": 9.212730318257956, + "eval_steps": 500, + "global_step": 5500, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.03350083752093802, + "grad_norm": 11.757207870483398, + "learning_rate": 6.333333333333334e-07, + "loss": 0.5112, + "step": 20 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 10.197953224182129, + "learning_rate": 1.3e-06, + "loss": 0.5283, + "step": 40 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 12.811074256896973, + "learning_rate": 1.9666666666666668e-06, + "loss": 0.474, + "step": 60 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 8.559342384338379, + "learning_rate": 2.6e-06, + "loss": 0.424, + "step": 80 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 6.9629225730896, + "learning_rate": 3.266666666666667e-06, + "loss": 0.4465, + "step": 100 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 7.833058834075928, + "learning_rate": 3.9333333333333335e-06, + "loss": 0.3764, + "step": 120 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 6.882424831390381, + "learning_rate": 4.600000000000001e-06, + "loss": 0.3705, + "step": 140 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 5.4183244705200195, + "learning_rate": 5.2666666666666665e-06, + "loss": 0.4116, + "step": 160 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 6.103787899017334, + "learning_rate": 5.933333333333335e-06, + "loss": 0.3903, + "step": 180 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 6.084160804748535, + "learning_rate": 6.600000000000001e-06, + "loss": 0.3794, + "step": 200 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 4.531406879425049, + "learning_rate": 7.266666666666668e-06, + "loss": 0.3904, + "step": 220 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 6.702854156494141, + "learning_rate": 7.933333333333334e-06, + "loss": 0.3672, + "step": 240 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 6.386377811431885, + "learning_rate": 8.6e-06, + "loss": 0.3379, + "step": 260 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 5.33281135559082, + "learning_rate": 9.266666666666667e-06, + "loss": 0.4102, + "step": 280 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 6.031435012817383, + "learning_rate": 9.933333333333334e-06, + "loss": 0.3815, + "step": 300 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 4.8638176918029785, + "learning_rate": 9.968253968253969e-06, + "loss": 0.358, + "step": 320 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 5.352113723754883, + "learning_rate": 9.932980599647268e-06, + "loss": 0.3578, + "step": 340 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 5.560739040374756, + "learning_rate": 9.897707231040565e-06, + "loss": 0.3795, + "step": 360 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 6.186940670013428, + "learning_rate": 9.862433862433864e-06, + "loss": 0.3852, + "step": 380 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 5.37507438659668, + "learning_rate": 9.827160493827161e-06, + "loss": 0.3761, + "step": 400 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 6.373449802398682, + "learning_rate": 9.79188712522046e-06, + "loss": 0.3674, + "step": 420 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 5.702625274658203, + "learning_rate": 9.756613756613757e-06, + "loss": 0.4013, + "step": 440 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 5.5675153732299805, + "learning_rate": 9.721340388007056e-06, + "loss": 0.3495, + "step": 460 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 6.296374320983887, + "learning_rate": 9.686067019400353e-06, + "loss": 0.3704, + "step": 480 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 4.830463886260986, + "learning_rate": 9.650793650793652e-06, + "loss": 0.36, + "step": 500 + }, + { + "epoch": 0.8375209380234506, + "eval_loss": 0.27764827013015747, + "eval_runtime": 1752.3826, + "eval_samples_per_second": 2.725, + "eval_steps_per_second": 0.341, + "eval_wer": 19.59291781998664, + "step": 500 + }, + { + "epoch": 0.8710217755443886, + "grad_norm": 4.863058090209961, + "learning_rate": 9.61552028218695e-06, + "loss": 0.3874, + "step": 520 + }, + { + "epoch": 0.9045226130653267, + "grad_norm": 6.121027946472168, + "learning_rate": 9.580246913580248e-06, + "loss": 0.382, + "step": 540 + }, + { + "epoch": 0.9380234505862647, + "grad_norm": 4.538082599639893, + "learning_rate": 9.544973544973546e-06, + "loss": 0.3611, + "step": 560 + }, + { + "epoch": 0.9715242881072027, + "grad_norm": 4.8772382736206055, + "learning_rate": 9.509700176366844e-06, + "loss": 0.3584, + "step": 580 + }, + { + "epoch": 1.0050251256281406, + "grad_norm": 4.039212703704834, + "learning_rate": 9.474426807760142e-06, + "loss": 0.3701, + "step": 600 + }, + { + "epoch": 1.0385259631490786, + "grad_norm": 4.787687301635742, + "learning_rate": 9.43915343915344e-06, + "loss": 0.2058, + "step": 620 + }, + { + "epoch": 1.0720268006700167, + "grad_norm": 4.503021717071533, + "learning_rate": 9.403880070546738e-06, + "loss": 0.2455, + "step": 640 + }, + { + "epoch": 1.1055276381909547, + "grad_norm": 6.663857936859131, + "learning_rate": 9.368606701940036e-06, + "loss": 0.2179, + "step": 660 + }, + { + "epoch": 1.1390284757118927, + "grad_norm": 4.486196041107178, + "learning_rate": 9.333333333333334e-06, + "loss": 0.2196, + "step": 680 + }, + { + "epoch": 1.1725293132328307, + "grad_norm": 4.667060852050781, + "learning_rate": 9.298059964726633e-06, + "loss": 0.2183, + "step": 700 + }, + { + "epoch": 1.2060301507537687, + "grad_norm": 5.608316898345947, + "learning_rate": 9.26278659611993e-06, + "loss": 0.2161, + "step": 720 + }, + { + "epoch": 1.2395309882747068, + "grad_norm": 4.2184271812438965, + "learning_rate": 9.227513227513229e-06, + "loss": 0.2382, + "step": 740 + }, + { + "epoch": 1.2730318257956448, + "grad_norm": 3.871945381164551, + "learning_rate": 9.192239858906526e-06, + "loss": 0.2214, + "step": 760 + }, + { + "epoch": 1.3065326633165828, + "grad_norm": 3.730222225189209, + "learning_rate": 9.156966490299825e-06, + "loss": 0.2213, + "step": 780 + }, + { + "epoch": 1.3400335008375208, + "grad_norm": 4.740777015686035, + "learning_rate": 9.121693121693122e-06, + "loss": 0.2503, + "step": 800 + }, + { + "epoch": 1.3735343383584588, + "grad_norm": 4.123469352722168, + "learning_rate": 9.086419753086421e-06, + "loss": 0.2443, + "step": 820 + }, + { + "epoch": 1.4070351758793969, + "grad_norm": 3.988917827606201, + "learning_rate": 9.051146384479718e-06, + "loss": 0.241, + "step": 840 + }, + { + "epoch": 1.4405360134003349, + "grad_norm": 5.4730305671691895, + "learning_rate": 9.015873015873017e-06, + "loss": 0.2163, + "step": 860 + }, + { + "epoch": 1.474036850921273, + "grad_norm": 5.020652770996094, + "learning_rate": 8.980599647266314e-06, + "loss": 0.2387, + "step": 880 + }, + { + "epoch": 1.507537688442211, + "grad_norm": 4.231486797332764, + "learning_rate": 8.945326278659613e-06, + "loss": 0.2523, + "step": 900 + }, + { + "epoch": 1.541038525963149, + "grad_norm": 6.197975158691406, + "learning_rate": 8.910052910052912e-06, + "loss": 0.2255, + "step": 920 + }, + { + "epoch": 1.574539363484087, + "grad_norm": 5.489210605621338, + "learning_rate": 8.874779541446209e-06, + "loss": 0.2397, + "step": 940 + }, + { + "epoch": 1.608040201005025, + "grad_norm": 4.32737398147583, + "learning_rate": 8.839506172839508e-06, + "loss": 0.2298, + "step": 960 + }, + { + "epoch": 1.641541038525963, + "grad_norm": 4.504214763641357, + "learning_rate": 8.804232804232805e-06, + "loss": 0.2301, + "step": 980 + }, + { + "epoch": 1.675041876046901, + "grad_norm": 3.9694759845733643, + "learning_rate": 8.768959435626104e-06, + "loss": 0.2338, + "step": 1000 + }, + { + "epoch": 1.675041876046901, + "eval_loss": 0.18838582932949066, + "eval_runtime": 1770.3449, + "eval_samples_per_second": 2.698, + "eval_steps_per_second": 0.337, + "eval_wer": 14.071299035983584, + "step": 1000 + }, + { + "epoch": 1.708542713567839, + "grad_norm": 4.972934722900391, + "learning_rate": 8.7336860670194e-06, + "loss": 0.2388, + "step": 1020 + }, + { + "epoch": 1.742043551088777, + "grad_norm": 5.178994655609131, + "learning_rate": 8.6984126984127e-06, + "loss": 0.2531, + "step": 1040 + }, + { + "epoch": 1.775544388609715, + "grad_norm": 3.560372829437256, + "learning_rate": 8.663139329805997e-06, + "loss": 0.2187, + "step": 1060 + }, + { + "epoch": 1.809045226130653, + "grad_norm": 4.647324562072754, + "learning_rate": 8.627865961199296e-06, + "loss": 0.2413, + "step": 1080 + }, + { + "epoch": 1.8425460636515911, + "grad_norm": 3.6784422397613525, + "learning_rate": 8.592592592592593e-06, + "loss": 0.2242, + "step": 1100 + }, + { + "epoch": 1.8760469011725294, + "grad_norm": 5.928018093109131, + "learning_rate": 8.557319223985891e-06, + "loss": 0.2406, + "step": 1120 + }, + { + "epoch": 1.9095477386934674, + "grad_norm": 6.358487606048584, + "learning_rate": 8.52204585537919e-06, + "loss": 0.2482, + "step": 1140 + }, + { + "epoch": 1.9430485762144054, + "grad_norm": 4.409506797790527, + "learning_rate": 8.486772486772487e-06, + "loss": 0.2415, + "step": 1160 + }, + { + "epoch": 1.9765494137353434, + "grad_norm": 5.169639587402344, + "learning_rate": 8.451499118165786e-06, + "loss": 0.222, + "step": 1180 + }, + { + "epoch": 2.0100502512562812, + "grad_norm": 2.8841328620910645, + "learning_rate": 8.416225749559083e-06, + "loss": 0.2109, + "step": 1200 + }, + { + "epoch": 2.0435510887772192, + "grad_norm": 3.835498332977295, + "learning_rate": 8.380952380952382e-06, + "loss": 0.1284, + "step": 1220 + }, + { + "epoch": 2.0770519262981573, + "grad_norm": 2.584859848022461, + "learning_rate": 8.34567901234568e-06, + "loss": 0.1059, + "step": 1240 + }, + { + "epoch": 2.1105527638190953, + "grad_norm": 2.797682285308838, + "learning_rate": 8.310405643738978e-06, + "loss": 0.0967, + "step": 1260 + }, + { + "epoch": 2.1440536013400333, + "grad_norm": 3.2937309741973877, + "learning_rate": 8.275132275132275e-06, + "loss": 0.1267, + "step": 1280 + }, + { + "epoch": 2.1775544388609713, + "grad_norm": 3.344325065612793, + "learning_rate": 8.239858906525574e-06, + "loss": 0.1354, + "step": 1300 + }, + { + "epoch": 2.2110552763819094, + "grad_norm": 2.955522060394287, + "learning_rate": 8.204585537918873e-06, + "loss": 0.104, + "step": 1320 + }, + { + "epoch": 2.2445561139028474, + "grad_norm": 5.180908203125, + "learning_rate": 8.16931216931217e-06, + "loss": 0.1229, + "step": 1340 + }, + { + "epoch": 2.2780569514237854, + "grad_norm": 3.639417886734009, + "learning_rate": 8.134038800705469e-06, + "loss": 0.104, + "step": 1360 + }, + { + "epoch": 2.3115577889447234, + "grad_norm": 3.7380902767181396, + "learning_rate": 8.098765432098766e-06, + "loss": 0.1141, + "step": 1380 + }, + { + "epoch": 2.3450586264656614, + "grad_norm": 4.100318908691406, + "learning_rate": 8.063492063492065e-06, + "loss": 0.1123, + "step": 1400 + }, + { + "epoch": 2.3785594639865995, + "grad_norm": 2.8147048950195312, + "learning_rate": 8.028218694885362e-06, + "loss": 0.1039, + "step": 1420 + }, + { + "epoch": 2.4120603015075375, + "grad_norm": 3.7376410961151123, + "learning_rate": 7.99294532627866e-06, + "loss": 0.1179, + "step": 1440 + }, + { + "epoch": 2.4455611390284755, + "grad_norm": 4.320065975189209, + "learning_rate": 7.957671957671958e-06, + "loss": 0.1167, + "step": 1460 + }, + { + "epoch": 2.4790619765494135, + "grad_norm": 4.415127277374268, + "learning_rate": 7.922398589065257e-06, + "loss": 0.125, + "step": 1480 + }, + { + "epoch": 2.5125628140703515, + "grad_norm": 2.631763219833374, + "learning_rate": 7.887125220458554e-06, + "loss": 0.1188, + "step": 1500 + }, + { + "epoch": 2.5125628140703515, + "eval_loss": 0.13635103404521942, + "eval_runtime": 1806.1099, + "eval_samples_per_second": 2.644, + "eval_steps_per_second": 0.331, + "eval_wer": 11.713753937195762, + "step": 1500 + }, + { + "epoch": 2.5460636515912896, + "grad_norm": 3.608773946762085, + "learning_rate": 7.851851851851853e-06, + "loss": 0.1222, + "step": 1520 + }, + { + "epoch": 2.5795644891122276, + "grad_norm": 3.4494569301605225, + "learning_rate": 7.816578483245151e-06, + "loss": 0.1087, + "step": 1540 + }, + { + "epoch": 2.6130653266331656, + "grad_norm": 3.6112048625946045, + "learning_rate": 7.781305114638449e-06, + "loss": 0.1097, + "step": 1560 + }, + { + "epoch": 2.6465661641541036, + "grad_norm": 3.1978707313537598, + "learning_rate": 7.746031746031747e-06, + "loss": 0.1101, + "step": 1580 + }, + { + "epoch": 2.6800670016750416, + "grad_norm": 3.6376733779907227, + "learning_rate": 7.710758377425045e-06, + "loss": 0.1056, + "step": 1600 + }, + { + "epoch": 2.7135678391959797, + "grad_norm": 3.754915237426758, + "learning_rate": 7.675485008818343e-06, + "loss": 0.1103, + "step": 1620 + }, + { + "epoch": 2.7470686767169177, + "grad_norm": 3.329240322113037, + "learning_rate": 7.64021164021164e-06, + "loss": 0.1193, + "step": 1640 + }, + { + "epoch": 2.7805695142378557, + "grad_norm": 3.9300169944763184, + "learning_rate": 7.604938271604939e-06, + "loss": 0.1232, + "step": 1660 + }, + { + "epoch": 2.8140703517587937, + "grad_norm": 4.290626049041748, + "learning_rate": 7.569664902998237e-06, + "loss": 0.1374, + "step": 1680 + }, + { + "epoch": 2.8475711892797317, + "grad_norm": 3.8987998962402344, + "learning_rate": 7.534391534391535e-06, + "loss": 0.1062, + "step": 1700 + }, + { + "epoch": 2.8810720268006698, + "grad_norm": 4.234960079193115, + "learning_rate": 7.499118165784833e-06, + "loss": 0.111, + "step": 1720 + }, + { + "epoch": 2.914572864321608, + "grad_norm": 3.103458881378174, + "learning_rate": 7.463844797178131e-06, + "loss": 0.1076, + "step": 1740 + }, + { + "epoch": 2.948073701842546, + "grad_norm": 3.270204782485962, + "learning_rate": 7.428571428571429e-06, + "loss": 0.1224, + "step": 1760 + }, + { + "epoch": 2.981574539363484, + "grad_norm": 4.261337757110596, + "learning_rate": 7.393298059964727e-06, + "loss": 0.1224, + "step": 1780 + }, + { + "epoch": 3.0150753768844223, + "grad_norm": 1.9531301259994507, + "learning_rate": 7.358024691358025e-06, + "loss": 0.0913, + "step": 1800 + }, + { + "epoch": 3.0485762144053603, + "grad_norm": 1.866215467453003, + "learning_rate": 7.322751322751324e-06, + "loss": 0.0466, + "step": 1820 + }, + { + "epoch": 3.0820770519262983, + "grad_norm": 3.9456610679626465, + "learning_rate": 7.287477954144622e-06, + "loss": 0.0451, + "step": 1840 + }, + { + "epoch": 3.1155778894472363, + "grad_norm": 1.9518849849700928, + "learning_rate": 7.25220458553792e-06, + "loss": 0.0414, + "step": 1860 + }, + { + "epoch": 3.1490787269681744, + "grad_norm": 2.232792854309082, + "learning_rate": 7.216931216931218e-06, + "loss": 0.0502, + "step": 1880 + }, + { + "epoch": 3.1825795644891124, + "grad_norm": 2.074127674102783, + "learning_rate": 7.181657848324516e-06, + "loss": 0.057, + "step": 1900 + }, + { + "epoch": 3.2160804020100504, + "grad_norm": 3.5036733150482178, + "learning_rate": 7.146384479717814e-06, + "loss": 0.0515, + "step": 1920 + }, + { + "epoch": 3.2495812395309884, + "grad_norm": 2.324014186859131, + "learning_rate": 7.111111111111112e-06, + "loss": 0.0454, + "step": 1940 + }, + { + "epoch": 3.2830820770519265, + "grad_norm": 2.710326671600342, + "learning_rate": 7.07583774250441e-06, + "loss": 0.0473, + "step": 1960 + }, + { + "epoch": 3.3165829145728645, + "grad_norm": 2.8943896293640137, + "learning_rate": 7.040564373897708e-06, + "loss": 0.046, + "step": 1980 + }, + { + "epoch": 3.3500837520938025, + "grad_norm": 2.319986581802368, + "learning_rate": 7.005291005291006e-06, + "loss": 0.0449, + "step": 2000 + }, + { + "epoch": 3.3500837520938025, + "eval_loss": 0.10934468358755112, + "eval_runtime": 1762.4506, + "eval_samples_per_second": 2.71, + "eval_steps_per_second": 0.339, + "eval_wer": 7.172854824854443, + "step": 2000 + }, + { + "epoch": 3.3835845896147405, + "grad_norm": 3.317129135131836, + "learning_rate": 6.9700176366843046e-06, + "loss": 0.0645, + "step": 2020 + }, + { + "epoch": 3.4170854271356785, + "grad_norm": 1.9533768892288208, + "learning_rate": 6.9347442680776025e-06, + "loss": 0.0432, + "step": 2040 + }, + { + "epoch": 3.4505862646566166, + "grad_norm": 2.0508453845977783, + "learning_rate": 6.8994708994709005e-06, + "loss": 0.0521, + "step": 2060 + }, + { + "epoch": 3.4840871021775546, + "grad_norm": 2.163236141204834, + "learning_rate": 6.8641975308641985e-06, + "loss": 0.0529, + "step": 2080 + }, + { + "epoch": 3.5175879396984926, + "grad_norm": 2.7154581546783447, + "learning_rate": 6.8289241622574965e-06, + "loss": 0.0452, + "step": 2100 + }, + { + "epoch": 3.5510887772194306, + "grad_norm": 3.0822432041168213, + "learning_rate": 6.7936507936507944e-06, + "loss": 0.0546, + "step": 2120 + }, + { + "epoch": 3.5845896147403686, + "grad_norm": 4.19010591506958, + "learning_rate": 6.758377425044092e-06, + "loss": 0.0529, + "step": 2140 + }, + { + "epoch": 3.6180904522613067, + "grad_norm": 2.9883594512939453, + "learning_rate": 6.72310405643739e-06, + "loss": 0.0503, + "step": 2160 + }, + { + "epoch": 3.6515912897822447, + "grad_norm": 2.3664371967315674, + "learning_rate": 6.687830687830688e-06, + "loss": 0.0498, + "step": 2180 + }, + { + "epoch": 3.6850921273031827, + "grad_norm": 2.0549991130828857, + "learning_rate": 6.652557319223986e-06, + "loss": 0.051, + "step": 2200 + }, + { + "epoch": 3.7185929648241207, + "grad_norm": 2.5339038372039795, + "learning_rate": 6.617283950617285e-06, + "loss": 0.0568, + "step": 2220 + }, + { + "epoch": 3.7520938023450587, + "grad_norm": 1.9988099336624146, + "learning_rate": 6.582010582010583e-06, + "loss": 0.051, + "step": 2240 + }, + { + "epoch": 3.7855946398659968, + "grad_norm": 2.5243782997131348, + "learning_rate": 6.546737213403881e-06, + "loss": 0.056, + "step": 2260 + }, + { + "epoch": 3.819095477386935, + "grad_norm": 3.157158136367798, + "learning_rate": 6.511463844797179e-06, + "loss": 0.0497, + "step": 2280 + }, + { + "epoch": 3.852596314907873, + "grad_norm": 1.9286202192306519, + "learning_rate": 6.476190476190477e-06, + "loss": 0.0426, + "step": 2300 + }, + { + "epoch": 3.886097152428811, + "grad_norm": 3.808802604675293, + "learning_rate": 6.440917107583775e-06, + "loss": 0.0499, + "step": 2320 + }, + { + "epoch": 3.919597989949749, + "grad_norm": 2.506671667098999, + "learning_rate": 6.405643738977073e-06, + "loss": 0.052, + "step": 2340 + }, + { + "epoch": 3.953098827470687, + "grad_norm": 2.9451920986175537, + "learning_rate": 6.370370370370371e-06, + "loss": 0.0552, + "step": 2360 + }, + { + "epoch": 3.986599664991625, + "grad_norm": 2.592744827270508, + "learning_rate": 6.335097001763669e-06, + "loss": 0.0527, + "step": 2380 + }, + { + "epoch": 4.0201005025125625, + "grad_norm": 1.8891575336456299, + "learning_rate": 6.299823633156967e-06, + "loss": 0.0289, + "step": 2400 + }, + { + "epoch": 4.0536013400335005, + "grad_norm": 1.8053243160247803, + "learning_rate": 6.264550264550266e-06, + "loss": 0.0192, + "step": 2420 + }, + { + "epoch": 4.0871021775544385, + "grad_norm": 2.0084407329559326, + "learning_rate": 6.229276895943564e-06, + "loss": 0.0242, + "step": 2440 + }, + { + "epoch": 4.1206030150753765, + "grad_norm": 1.5919119119644165, + "learning_rate": 6.194003527336862e-06, + "loss": 0.0211, + "step": 2460 + }, + { + "epoch": 4.1541038525963145, + "grad_norm": 1.9214613437652588, + "learning_rate": 6.15873015873016e-06, + "loss": 0.0233, + "step": 2480 + }, + { + "epoch": 4.187604690117253, + "grad_norm": 1.2652311325073242, + "learning_rate": 6.123456790123458e-06, + "loss": 0.0199, + "step": 2500 + }, + { + "epoch": 4.187604690117253, + "eval_loss": 0.0981329157948494, + "eval_runtime": 1779.0213, + "eval_samples_per_second": 2.685, + "eval_steps_per_second": 0.336, + "eval_wer": 6.707549871146321, + "step": 2500 + }, + { + "epoch": 4.221105527638191, + "grad_norm": 2.026528835296631, + "learning_rate": 6.088183421516756e-06, + "loss": 0.0217, + "step": 2520 + }, + { + "epoch": 4.254606365159129, + "grad_norm": 1.596919059753418, + "learning_rate": 6.052910052910054e-06, + "loss": 0.0167, + "step": 2540 + }, + { + "epoch": 4.288107202680067, + "grad_norm": 2.9445090293884277, + "learning_rate": 6.017636684303352e-06, + "loss": 0.0225, + "step": 2560 + }, + { + "epoch": 4.321608040201005, + "grad_norm": 2.4160282611846924, + "learning_rate": 5.9823633156966496e-06, + "loss": 0.0253, + "step": 2580 + }, + { + "epoch": 4.355108877721943, + "grad_norm": 1.461127758026123, + "learning_rate": 5.9470899470899475e-06, + "loss": 0.0197, + "step": 2600 + }, + { + "epoch": 4.388609715242881, + "grad_norm": 2.7892863750457764, + "learning_rate": 5.911816578483246e-06, + "loss": 0.022, + "step": 2620 + }, + { + "epoch": 4.422110552763819, + "grad_norm": 1.651208758354187, + "learning_rate": 5.876543209876544e-06, + "loss": 0.0215, + "step": 2640 + }, + { + "epoch": 4.455611390284757, + "grad_norm": 2.2500391006469727, + "learning_rate": 5.841269841269842e-06, + "loss": 0.0247, + "step": 2660 + }, + { + "epoch": 4.489112227805695, + "grad_norm": 4.447635173797607, + "learning_rate": 5.80599647266314e-06, + "loss": 0.0263, + "step": 2680 + }, + { + "epoch": 4.522613065326633, + "grad_norm": 0.8300407528877258, + "learning_rate": 5.770723104056438e-06, + "loss": 0.0209, + "step": 2700 + }, + { + "epoch": 4.556113902847571, + "grad_norm": 1.6874111890792847, + "learning_rate": 5.735449735449736e-06, + "loss": 0.0195, + "step": 2720 + }, + { + "epoch": 4.589614740368509, + "grad_norm": 2.4045815467834473, + "learning_rate": 5.700176366843034e-06, + "loss": 0.0224, + "step": 2740 + }, + { + "epoch": 4.623115577889447, + "grad_norm": 2.3160908222198486, + "learning_rate": 5.664902998236332e-06, + "loss": 0.0179, + "step": 2760 + }, + { + "epoch": 4.656616415410385, + "grad_norm": 1.6684287786483765, + "learning_rate": 5.62962962962963e-06, + "loss": 0.0238, + "step": 2780 + }, + { + "epoch": 4.690117252931323, + "grad_norm": 1.973906397819519, + "learning_rate": 5.594356261022928e-06, + "loss": 0.0226, + "step": 2800 + }, + { + "epoch": 4.723618090452261, + "grad_norm": 2.270906686782837, + "learning_rate": 5.559082892416227e-06, + "loss": 0.0232, + "step": 2820 + }, + { + "epoch": 4.757118927973199, + "grad_norm": 1.8875011205673218, + "learning_rate": 5.523809523809525e-06, + "loss": 0.0218, + "step": 2840 + }, + { + "epoch": 4.790619765494137, + "grad_norm": 1.1312583684921265, + "learning_rate": 5.488536155202823e-06, + "loss": 0.0212, + "step": 2860 + }, + { + "epoch": 4.824120603015075, + "grad_norm": 0.864783525466919, + "learning_rate": 5.453262786596121e-06, + "loss": 0.023, + "step": 2880 + }, + { + "epoch": 4.857621440536013, + "grad_norm": 1.2935965061187744, + "learning_rate": 5.417989417989419e-06, + "loss": 0.019, + "step": 2900 + }, + { + "epoch": 4.891122278056951, + "grad_norm": 2.4576382637023926, + "learning_rate": 5.382716049382717e-06, + "loss": 0.0193, + "step": 2920 + }, + { + "epoch": 4.924623115577889, + "grad_norm": 2.71472430229187, + "learning_rate": 5.347442680776015e-06, + "loss": 0.0253, + "step": 2940 + }, + { + "epoch": 4.958123953098827, + "grad_norm": 2.84940505027771, + "learning_rate": 5.312169312169313e-06, + "loss": 0.0218, + "step": 2960 + }, + { + "epoch": 4.991624790619765, + "grad_norm": 1.8483999967575073, + "learning_rate": 5.276895943562611e-06, + "loss": 0.0226, + "step": 2980 + }, + { + "epoch": 5.025125628140704, + "grad_norm": 0.6126876473426819, + "learning_rate": 5.241622574955909e-06, + "loss": 0.0101, + "step": 3000 + }, + { + "epoch": 5.025125628140704, + "eval_loss": 0.09390027821063995, + "eval_runtime": 1767.4512, + "eval_samples_per_second": 2.702, + "eval_steps_per_second": 0.338, + "eval_wer": 5.500143170754987, + "step": 3000 + }, + { + "epoch": 5.058626465661642, + "grad_norm": 0.3711394965648651, + "learning_rate": 5.2063492063492076e-06, + "loss": 0.0119, + "step": 3020 + }, + { + "epoch": 5.09212730318258, + "grad_norm": 0.5738839507102966, + "learning_rate": 5.1710758377425055e-06, + "loss": 0.0086, + "step": 3040 + }, + { + "epoch": 5.125628140703517, + "grad_norm": 0.7609245777130127, + "learning_rate": 5.1358024691358035e-06, + "loss": 0.0093, + "step": 3060 + }, + { + "epoch": 5.159128978224456, + "grad_norm": 1.2764722108840942, + "learning_rate": 5.1005291005291015e-06, + "loss": 0.0111, + "step": 3080 + }, + { + "epoch": 5.192629815745394, + "grad_norm": 2.1169776916503906, + "learning_rate": 5.0652557319223995e-06, + "loss": 0.0107, + "step": 3100 + }, + { + "epoch": 5.226130653266332, + "grad_norm": 2.1893081665039062, + "learning_rate": 5.0299823633156974e-06, + "loss": 0.0097, + "step": 3120 + }, + { + "epoch": 5.259631490787269, + "grad_norm": 2.2419638633728027, + "learning_rate": 4.9947089947089946e-06, + "loss": 0.0098, + "step": 3140 + }, + { + "epoch": 5.293132328308207, + "grad_norm": 0.6479611992835999, + "learning_rate": 4.959435626102293e-06, + "loss": 0.0082, + "step": 3160 + }, + { + "epoch": 5.326633165829146, + "grad_norm": 0.4799642860889435, + "learning_rate": 4.924162257495591e-06, + "loss": 0.0119, + "step": 3180 + }, + { + "epoch": 5.360134003350084, + "grad_norm": 0.7716453075408936, + "learning_rate": 4.888888888888889e-06, + "loss": 0.0086, + "step": 3200 + }, + { + "epoch": 5.393634840871022, + "grad_norm": 1.2303547859191895, + "learning_rate": 4.853615520282187e-06, + "loss": 0.009, + "step": 3220 + }, + { + "epoch": 5.42713567839196, + "grad_norm": 0.6345349550247192, + "learning_rate": 4.818342151675485e-06, + "loss": 0.0085, + "step": 3240 + }, + { + "epoch": 5.460636515912898, + "grad_norm": 0.9741530418395996, + "learning_rate": 4.783068783068783e-06, + "loss": 0.0082, + "step": 3260 + }, + { + "epoch": 5.494137353433836, + "grad_norm": 1.1631624698638916, + "learning_rate": 4.747795414462081e-06, + "loss": 0.0086, + "step": 3280 + }, + { + "epoch": 5.527638190954773, + "grad_norm": 0.6502953767776489, + "learning_rate": 4.712522045855379e-06, + "loss": 0.0093, + "step": 3300 + }, + { + "epoch": 5.561139028475711, + "grad_norm": 0.7464337348937988, + "learning_rate": 4.677248677248677e-06, + "loss": 0.0086, + "step": 3320 + }, + { + "epoch": 5.594639865996649, + "grad_norm": 1.017751693725586, + "learning_rate": 4.641975308641975e-06, + "loss": 0.0091, + "step": 3340 + }, + { + "epoch": 5.628140703517588, + "grad_norm": 0.4273395836353302, + "learning_rate": 4.606701940035274e-06, + "loss": 0.0078, + "step": 3360 + }, + { + "epoch": 5.661641541038526, + "grad_norm": 0.6737497448921204, + "learning_rate": 4.571428571428572e-06, + "loss": 0.0106, + "step": 3380 + }, + { + "epoch": 5.695142378559464, + "grad_norm": 1.0791343450546265, + "learning_rate": 4.53615520282187e-06, + "loss": 0.0097, + "step": 3400 + }, + { + "epoch": 5.728643216080402, + "grad_norm": 1.0891772508621216, + "learning_rate": 4.500881834215168e-06, + "loss": 0.0109, + "step": 3420 + }, + { + "epoch": 5.76214405360134, + "grad_norm": 0.7465157508850098, + "learning_rate": 4.465608465608466e-06, + "loss": 0.0078, + "step": 3440 + }, + { + "epoch": 5.795644891122278, + "grad_norm": 0.7693866491317749, + "learning_rate": 4.430335097001764e-06, + "loss": 0.009, + "step": 3460 + }, + { + "epoch": 5.8291457286432165, + "grad_norm": 1.3295698165893555, + "learning_rate": 4.395061728395062e-06, + "loss": 0.016, + "step": 3480 + }, + { + "epoch": 5.8626465661641545, + "grad_norm": 2.4605352878570557, + "learning_rate": 4.35978835978836e-06, + "loss": 0.0081, + "step": 3500 + }, + { + "epoch": 5.8626465661641545, + "eval_loss": 0.09377142041921616, + "eval_runtime": 1773.01, + "eval_samples_per_second": 2.694, + "eval_steps_per_second": 0.337, + "eval_wer": 5.010976424549012, + "step": 3500 + }, + { + "epoch": 5.8961474036850925, + "grad_norm": 0.8250058889389038, + "learning_rate": 4.324514991181658e-06, + "loss": 0.0108, + "step": 3520 + }, + { + "epoch": 5.9296482412060305, + "grad_norm": 1.3606537580490112, + "learning_rate": 4.289241622574956e-06, + "loss": 0.0125, + "step": 3540 + }, + { + "epoch": 5.9631490787269685, + "grad_norm": 0.6893450021743774, + "learning_rate": 4.2539682539682546e-06, + "loss": 0.0101, + "step": 3560 + }, + { + "epoch": 5.9966499162479066, + "grad_norm": 0.8129726052284241, + "learning_rate": 4.2186948853615525e-06, + "loss": 0.0095, + "step": 3580 + }, + { + "epoch": 6.030150753768845, + "grad_norm": 0.5319514274597168, + "learning_rate": 4.1834215167548505e-06, + "loss": 0.0051, + "step": 3600 + }, + { + "epoch": 6.063651591289783, + "grad_norm": 0.20368462800979614, + "learning_rate": 4.1481481481481485e-06, + "loss": 0.0049, + "step": 3620 + }, + { + "epoch": 6.097152428810721, + "grad_norm": 1.1721038818359375, + "learning_rate": 4.1128747795414465e-06, + "loss": 0.0041, + "step": 3640 + }, + { + "epoch": 6.130653266331659, + "grad_norm": 0.17129285633563995, + "learning_rate": 4.0776014109347444e-06, + "loss": 0.0055, + "step": 3660 + }, + { + "epoch": 6.164154103852597, + "grad_norm": 0.31987234950065613, + "learning_rate": 4.042328042328042e-06, + "loss": 0.0037, + "step": 3680 + }, + { + "epoch": 6.197654941373535, + "grad_norm": 0.3214021921157837, + "learning_rate": 4.00705467372134e-06, + "loss": 0.0052, + "step": 3700 + }, + { + "epoch": 6.231155778894473, + "grad_norm": 1.3790876865386963, + "learning_rate": 3.971781305114638e-06, + "loss": 0.0053, + "step": 3720 + }, + { + "epoch": 6.264656616415411, + "grad_norm": 0.549566924571991, + "learning_rate": 3.936507936507936e-06, + "loss": 0.0048, + "step": 3740 + }, + { + "epoch": 6.298157453936349, + "grad_norm": 0.2458494007587433, + "learning_rate": 3.901234567901235e-06, + "loss": 0.006, + "step": 3760 + }, + { + "epoch": 6.331658291457287, + "grad_norm": 2.3661324977874756, + "learning_rate": 3.865961199294533e-06, + "loss": 0.0046, + "step": 3780 + }, + { + "epoch": 6.365159128978225, + "grad_norm": 0.7839031219482422, + "learning_rate": 3.830687830687831e-06, + "loss": 0.0059, + "step": 3800 + }, + { + "epoch": 6.398659966499163, + "grad_norm": 0.2562466263771057, + "learning_rate": 3.795414462081129e-06, + "loss": 0.0038, + "step": 3820 + }, + { + "epoch": 6.432160804020101, + "grad_norm": 0.9680606126785278, + "learning_rate": 3.760141093474427e-06, + "loss": 0.0053, + "step": 3840 + }, + { + "epoch": 6.465661641541039, + "grad_norm": 0.5647270083427429, + "learning_rate": 3.724867724867725e-06, + "loss": 0.0049, + "step": 3860 + }, + { + "epoch": 6.499162479061977, + "grad_norm": 0.5850030183792114, + "learning_rate": 3.689594356261023e-06, + "loss": 0.0039, + "step": 3880 + }, + { + "epoch": 6.532663316582915, + "grad_norm": 0.1745942085981369, + "learning_rate": 3.654320987654321e-06, + "loss": 0.0035, + "step": 3900 + }, + { + "epoch": 6.566164154103853, + "grad_norm": 0.2170235961675644, + "learning_rate": 3.6190476190476194e-06, + "loss": 0.0045, + "step": 3920 + }, + { + "epoch": 6.599664991624791, + "grad_norm": 0.30363383889198303, + "learning_rate": 3.5837742504409174e-06, + "loss": 0.0031, + "step": 3940 + }, + { + "epoch": 6.633165829145729, + "grad_norm": 0.22252851724624634, + "learning_rate": 3.5485008818342153e-06, + "loss": 0.004, + "step": 3960 + }, + { + "epoch": 6.666666666666667, + "grad_norm": 0.9289886951446533, + "learning_rate": 3.5132275132275133e-06, + "loss": 0.0055, + "step": 3980 + }, + { + "epoch": 6.700167504187605, + "grad_norm": 1.094429612159729, + "learning_rate": 3.4779541446208113e-06, + "loss": 0.0041, + "step": 4000 + }, + { + "epoch": 6.700167504187605, + "eval_loss": 0.09379494935274124, + "eval_runtime": 1761.7753, + "eval_samples_per_second": 2.711, + "eval_steps_per_second": 0.339, + "eval_wer": 4.89166746205975, + "step": 4000 + }, + { + "epoch": 6.733668341708543, + "grad_norm": 0.8961842656135559, + "learning_rate": 3.4426807760141097e-06, + "loss": 0.0034, + "step": 4020 + }, + { + "epoch": 6.767169179229481, + "grad_norm": 0.45882654190063477, + "learning_rate": 3.4074074074074077e-06, + "loss": 0.0054, + "step": 4040 + }, + { + "epoch": 6.800670016750419, + "grad_norm": 0.27223679423332214, + "learning_rate": 3.3721340388007056e-06, + "loss": 0.0034, + "step": 4060 + }, + { + "epoch": 6.834170854271357, + "grad_norm": 1.3323800563812256, + "learning_rate": 3.3368606701940036e-06, + "loss": 0.0055, + "step": 4080 + }, + { + "epoch": 6.867671691792295, + "grad_norm": 0.24382267892360687, + "learning_rate": 3.3015873015873016e-06, + "loss": 0.0038, + "step": 4100 + }, + { + "epoch": 6.901172529313233, + "grad_norm": 0.2211693376302719, + "learning_rate": 3.2663139329806e-06, + "loss": 0.0043, + "step": 4120 + }, + { + "epoch": 6.934673366834171, + "grad_norm": 0.20270536839962006, + "learning_rate": 3.231040564373898e-06, + "loss": 0.0047, + "step": 4140 + }, + { + "epoch": 6.968174204355109, + "grad_norm": 2.0083487033843994, + "learning_rate": 3.195767195767196e-06, + "loss": 0.0054, + "step": 4160 + }, + { + "epoch": 7.001675041876047, + "grad_norm": 0.1442304253578186, + "learning_rate": 3.160493827160494e-06, + "loss": 0.005, + "step": 4180 + }, + { + "epoch": 7.035175879396985, + "grad_norm": 0.38336917757987976, + "learning_rate": 3.126984126984127e-06, + "loss": 0.0031, + "step": 4200 + }, + { + "epoch": 7.068676716917923, + "grad_norm": 0.13572201132774353, + "learning_rate": 3.0917107583774254e-06, + "loss": 0.0027, + "step": 4220 + }, + { + "epoch": 7.102177554438861, + "grad_norm": 1.8334097862243652, + "learning_rate": 3.0564373897707234e-06, + "loss": 0.0025, + "step": 4240 + }, + { + "epoch": 7.135678391959799, + "grad_norm": 1.0188051462173462, + "learning_rate": 3.0211640211640214e-06, + "loss": 0.0032, + "step": 4260 + }, + { + "epoch": 7.169179229480737, + "grad_norm": 0.13007956743240356, + "learning_rate": 2.9858906525573194e-06, + "loss": 0.0033, + "step": 4280 + }, + { + "epoch": 7.202680067001675, + "grad_norm": 0.07188330590724945, + "learning_rate": 2.9506172839506173e-06, + "loss": 0.0025, + "step": 4300 + }, + { + "epoch": 7.236180904522613, + "grad_norm": 0.2624744176864624, + "learning_rate": 2.9153439153439157e-06, + "loss": 0.0022, + "step": 4320 + }, + { + "epoch": 7.269681742043551, + "grad_norm": 0.1817062646150589, + "learning_rate": 2.8800705467372137e-06, + "loss": 0.0027, + "step": 4340 + }, + { + "epoch": 7.303182579564489, + "grad_norm": 0.23730872571468353, + "learning_rate": 2.8447971781305117e-06, + "loss": 0.0028, + "step": 4360 + }, + { + "epoch": 7.336683417085427, + "grad_norm": 0.44462695717811584, + "learning_rate": 2.8095238095238096e-06, + "loss": 0.0027, + "step": 4380 + }, + { + "epoch": 7.370184254606365, + "grad_norm": 0.16056309640407562, + "learning_rate": 2.7742504409171076e-06, + "loss": 0.0021, + "step": 4400 + }, + { + "epoch": 7.403685092127303, + "grad_norm": 0.1740381121635437, + "learning_rate": 2.738977072310406e-06, + "loss": 0.0021, + "step": 4420 + }, + { + "epoch": 7.437185929648241, + "grad_norm": 0.2567131519317627, + "learning_rate": 2.703703703703704e-06, + "loss": 0.0025, + "step": 4440 + }, + { + "epoch": 7.4706867671691795, + "grad_norm": 0.22708427906036377, + "learning_rate": 2.668430335097002e-06, + "loss": 0.0018, + "step": 4460 + }, + { + "epoch": 7.5041876046901175, + "grad_norm": 0.14704178273677826, + "learning_rate": 2.6331569664903e-06, + "loss": 0.0023, + "step": 4480 + }, + { + "epoch": 7.5376884422110555, + "grad_norm": 0.26313552260398865, + "learning_rate": 2.597883597883598e-06, + "loss": 0.0026, + "step": 4500 + }, + { + "epoch": 7.5376884422110555, + "eval_loss": 0.09553142637014389, + "eval_runtime": 1387.1198, + "eval_samples_per_second": 3.443, + "eval_steps_per_second": 0.43, + "eval_wer": 4.50510642359454, + "step": 4500 + }, + { + "epoch": 7.5711892797319935, + "grad_norm": 0.09787946194410324, + "learning_rate": 2.5626102292768963e-06, + "loss": 0.0022, + "step": 4520 + }, + { + "epoch": 7.6046901172529315, + "grad_norm": 0.10577196627855301, + "learning_rate": 2.5273368606701943e-06, + "loss": 0.0023, + "step": 4540 + }, + { + "epoch": 7.63819095477387, + "grad_norm": 0.11914186924695969, + "learning_rate": 2.4920634920634923e-06, + "loss": 0.0026, + "step": 4560 + }, + { + "epoch": 7.671691792294808, + "grad_norm": 0.6797345876693726, + "learning_rate": 2.4567901234567902e-06, + "loss": 0.0028, + "step": 4580 + }, + { + "epoch": 7.705192629815746, + "grad_norm": 0.11635720729827881, + "learning_rate": 2.4215167548500882e-06, + "loss": 0.0021, + "step": 4600 + }, + { + "epoch": 7.738693467336684, + "grad_norm": 0.0952395349740982, + "learning_rate": 2.3862433862433866e-06, + "loss": 0.0021, + "step": 4620 + }, + { + "epoch": 7.772194304857622, + "grad_norm": 0.11671995371580124, + "learning_rate": 2.3509700176366846e-06, + "loss": 0.0023, + "step": 4640 + }, + { + "epoch": 7.80569514237856, + "grad_norm": 0.10185588896274567, + "learning_rate": 2.3156966490299826e-06, + "loss": 0.003, + "step": 4660 + }, + { + "epoch": 7.839195979899498, + "grad_norm": 0.11902861297130585, + "learning_rate": 2.2804232804232805e-06, + "loss": 0.0021, + "step": 4680 + }, + { + "epoch": 7.872696817420436, + "grad_norm": 0.10562069714069366, + "learning_rate": 2.2451499118165785e-06, + "loss": 0.002, + "step": 4700 + }, + { + "epoch": 7.906197654941374, + "grad_norm": 0.1272854059934616, + "learning_rate": 2.209876543209877e-06, + "loss": 0.0024, + "step": 4720 + }, + { + "epoch": 7.939698492462312, + "grad_norm": 0.14873459935188293, + "learning_rate": 2.174603174603175e-06, + "loss": 0.0021, + "step": 4740 + }, + { + "epoch": 7.97319932998325, + "grad_norm": 0.09556525200605392, + "learning_rate": 2.139329805996473e-06, + "loss": 0.0021, + "step": 4760 + }, + { + "epoch": 8.006700167504187, + "grad_norm": 0.2545607388019562, + "learning_rate": 2.104056437389771e-06, + "loss": 0.0021, + "step": 4780 + }, + { + "epoch": 8.040201005025125, + "grad_norm": 0.09803763031959534, + "learning_rate": 2.068783068783069e-06, + "loss": 0.0017, + "step": 4800 + }, + { + "epoch": 8.073701842546063, + "grad_norm": 0.6409705281257629, + "learning_rate": 2.0335097001763672e-06, + "loss": 0.0019, + "step": 4820 + }, + { + "epoch": 8.107202680067001, + "grad_norm": 0.08693050593137741, + "learning_rate": 1.998236331569665e-06, + "loss": 0.0017, + "step": 4840 + }, + { + "epoch": 8.140703517587939, + "grad_norm": 0.09300371259450912, + "learning_rate": 1.962962962962963e-06, + "loss": 0.0022, + "step": 4860 + }, + { + "epoch": 8.174204355108877, + "grad_norm": 0.09886649250984192, + "learning_rate": 1.927689594356261e-06, + "loss": 0.0016, + "step": 4880 + }, + { + "epoch": 8.207705192629815, + "grad_norm": 0.14025329053401947, + "learning_rate": 1.8924162257495593e-06, + "loss": 0.0031, + "step": 4900 + }, + { + "epoch": 8.241206030150753, + "grad_norm": 0.06884673237800598, + "learning_rate": 1.8571428571428573e-06, + "loss": 0.0014, + "step": 4920 + }, + { + "epoch": 8.274706867671691, + "grad_norm": 0.08606573939323425, + "learning_rate": 1.8218694885361555e-06, + "loss": 0.0015, + "step": 4940 + }, + { + "epoch": 8.308207705192629, + "grad_norm": 0.5949509739875793, + "learning_rate": 1.7865961199294535e-06, + "loss": 0.0017, + "step": 4960 + }, + { + "epoch": 8.341708542713567, + "grad_norm": 0.07532797753810883, + "learning_rate": 1.7513227513227514e-06, + "loss": 0.0015, + "step": 4980 + }, + { + "epoch": 8.375209380234505, + "grad_norm": 0.07208308577537537, + "learning_rate": 1.7160493827160496e-06, + "loss": 0.0016, + "step": 5000 + }, + { + "epoch": 8.375209380234505, + "eval_loss": 0.0970502495765686, + "eval_runtime": 1406.3953, + "eval_samples_per_second": 3.396, + "eval_steps_per_second": 0.424, + "eval_wer": 4.457382838598836, + "step": 5000 + }, + { + "epoch": 8.408710217755443, + "grad_norm": 0.07992870360612869, + "learning_rate": 1.6807760141093476e-06, + "loss": 0.0015, + "step": 5020 + }, + { + "epoch": 8.442211055276381, + "grad_norm": 0.1169973835349083, + "learning_rate": 1.6455026455026458e-06, + "loss": 0.0018, + "step": 5040 + }, + { + "epoch": 8.47571189279732, + "grad_norm": 0.07965467125177383, + "learning_rate": 1.6102292768959438e-06, + "loss": 0.0015, + "step": 5060 + }, + { + "epoch": 8.509212730318257, + "grad_norm": 0.08486379683017731, + "learning_rate": 1.5749559082892417e-06, + "loss": 0.0016, + "step": 5080 + }, + { + "epoch": 8.542713567839195, + "grad_norm": 0.08790017664432526, + "learning_rate": 1.53968253968254e-06, + "loss": 0.0017, + "step": 5100 + }, + { + "epoch": 8.576214405360133, + "grad_norm": 0.07054130733013153, + "learning_rate": 1.504409171075838e-06, + "loss": 0.0015, + "step": 5120 + }, + { + "epoch": 8.609715242881071, + "grad_norm": 0.09546579420566559, + "learning_rate": 1.469135802469136e-06, + "loss": 0.0017, + "step": 5140 + }, + { + "epoch": 8.64321608040201, + "grad_norm": 0.09177026897668839, + "learning_rate": 1.433862433862434e-06, + "loss": 0.0015, + "step": 5160 + }, + { + "epoch": 8.676716917922947, + "grad_norm": 0.11043737083673477, + "learning_rate": 1.398589065255732e-06, + "loss": 0.0015, + "step": 5180 + }, + { + "epoch": 8.710217755443885, + "grad_norm": 0.09833870083093643, + "learning_rate": 1.3633156966490302e-06, + "loss": 0.0015, + "step": 5200 + }, + { + "epoch": 8.743718592964823, + "grad_norm": 0.07351676374673843, + "learning_rate": 1.3280423280423282e-06, + "loss": 0.0015, + "step": 5220 + }, + { + "epoch": 8.777219430485761, + "grad_norm": 0.09649737179279327, + "learning_rate": 1.2927689594356264e-06, + "loss": 0.0015, + "step": 5240 + }, + { + "epoch": 8.8107202680067, + "grad_norm": 0.08271702378988266, + "learning_rate": 1.2574955908289244e-06, + "loss": 0.0017, + "step": 5260 + }, + { + "epoch": 8.844221105527637, + "grad_norm": 0.08492777496576309, + "learning_rate": 1.2222222222222223e-06, + "loss": 0.0014, + "step": 5280 + }, + { + "epoch": 8.877721943048575, + "grad_norm": 0.11445468664169312, + "learning_rate": 1.1869488536155203e-06, + "loss": 0.0016, + "step": 5300 + }, + { + "epoch": 8.911222780569513, + "grad_norm": 0.08237680047750473, + "learning_rate": 1.1516754850088185e-06, + "loss": 0.0021, + "step": 5320 + }, + { + "epoch": 8.944723618090451, + "grad_norm": 0.3206549882888794, + "learning_rate": 1.1164021164021165e-06, + "loss": 0.0015, + "step": 5340 + }, + { + "epoch": 8.97822445561139, + "grad_norm": 0.0848088189959526, + "learning_rate": 1.0811287477954144e-06, + "loss": 0.0025, + "step": 5360 + }, + { + "epoch": 9.011725293132328, + "grad_norm": 0.06675776094198227, + "learning_rate": 1.0458553791887126e-06, + "loss": 0.0014, + "step": 5380 + }, + { + "epoch": 9.045226130653266, + "grad_norm": 0.054239727556705475, + "learning_rate": 1.0105820105820106e-06, + "loss": 0.0013, + "step": 5400 + }, + { + "epoch": 9.078726968174204, + "grad_norm": 0.1410382091999054, + "learning_rate": 9.753086419753088e-07, + "loss": 0.0013, + "step": 5420 + }, + { + "epoch": 9.112227805695142, + "grad_norm": 0.06337971985340118, + "learning_rate": 9.400352733686068e-07, + "loss": 0.0013, + "step": 5440 + }, + { + "epoch": 9.14572864321608, + "grad_norm": 0.06749581545591354, + "learning_rate": 9.047619047619048e-07, + "loss": 0.0014, + "step": 5460 + }, + { + "epoch": 9.179229480737018, + "grad_norm": 0.07229887694120407, + "learning_rate": 8.694885361552028e-07, + "loss": 0.0013, + "step": 5480 + }, + { + "epoch": 9.212730318257956, + "grad_norm": 0.08598683774471283, + "learning_rate": 8.342151675485009e-07, + "loss": 0.0013, + "step": 5500 + }, + { + "epoch": 9.212730318257956, + "eval_loss": 0.09753644466400146, + "eval_runtime": 1385.4537, + "eval_samples_per_second": 3.447, + "eval_steps_per_second": 0.431, + "eval_wer": 4.431134866851197, + "step": 5500 + } + ], + "logging_steps": 20, + "max_steps": 5970, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "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": 5.078323865051136e+19, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-5500/training_args.bin b/checkpoint-5500/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6ffec31ca17b32691fb4cd1278f73919f3259cf4 --- /dev/null +++ b/checkpoint-5500/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41a3c645624f7463029286b7d5e1ad6924cfc44a0d153880431b77c40e5deb73 +size 5432 diff --git a/checkpoint-5970/config.json b/checkpoint-5970/config.json new file mode 100644 index 0000000000000000000000000000000000000000..337642e18341c758364c32c6b6159a090615329b --- /dev/null +++ b/checkpoint-5970/config.json @@ -0,0 +1,47 @@ +{ + "activation_dropout": 0.0, + "activation_function": "gelu", + "apply_spec_augment": false, + "architectures": [ + "WhisperForConditionalGeneration" + ], + "attention_dropout": 0.0, + "begin_suppress_tokens": null, + "bos_token_id": 50257, + "classifier_proj_size": 256, + "d_model": 768, + "decoder_attention_heads": 12, + "decoder_ffn_dim": 3072, + "decoder_layerdrop": 0.0, + "decoder_layers": 12, + "decoder_start_token_id": 50258, + "dropout": 0.0, + "encoder_attention_heads": 12, + "encoder_ffn_dim": 3072, + "encoder_layerdrop": 0.0, + "encoder_layers": 12, + "eos_token_id": 50257, + "forced_decoder_ids": null, + "init_std": 0.02, + "is_encoder_decoder": true, + "mask_feature_length": 10, + "mask_feature_min_masks": 0, + "mask_feature_prob": 0.0, + "mask_time_length": 10, + "mask_time_min_masks": 2, + "mask_time_prob": 0.05, + "max_length": null, + "max_source_positions": 1500, + "max_target_positions": 448, + "median_filter_width": 7, + "model_type": "whisper", + "num_hidden_layers": 12, + "num_mel_bins": 80, + "pad_token_id": 50257, + "scale_embedding": false, + "torch_dtype": "float32", + "transformers_version": "4.51.3", + "use_cache": false, + "use_weighted_layer_sum": false, + "vocab_size": 51865 +} diff --git a/checkpoint-5970/generation_config.json b/checkpoint-5970/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..e1f4120825a303544ca23708d9dc52bbaa1dce16 --- /dev/null +++ b/checkpoint-5970/generation_config.json @@ -0,0 +1,9 @@ +{ + "bos_token_id": 50257, + "decoder_start_token_id": 50258, + "eos_token_id": 50257, + "pad_token_id": 50257, + "suppress_tokens": [], + "transformers_version": "4.51.3", + "use_cache": false +} diff --git a/checkpoint-5970/model.safetensors b/checkpoint-5970/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c58a28a4896de4e122a7f4d96a95e1848bdd11f2 --- /dev/null +++ b/checkpoint-5970/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e28027f2f07117bb9a07942d71f74a3f5235a8bcc96e3a1b4663de3d697666ac +size 966995080 diff --git a/checkpoint-5970/optimizer.pt b/checkpoint-5970/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..c146f6cd01aa5e177ca9bbd47807a6d02c0c3cf3 --- /dev/null +++ b/checkpoint-5970/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4715ee784a94e5357e7d9f5893559054e6ea8d39c87685a27b2893960bfe9a1e +size 1925064044 diff --git a/checkpoint-5970/preprocessor_config.json b/checkpoint-5970/preprocessor_config.json new file mode 100644 index 0000000000000000000000000000000000000000..d7c5b1dfd466c815664b16139e8f605620d22e4e --- /dev/null +++ b/checkpoint-5970/preprocessor_config.json @@ -0,0 +1,15 @@ +{ + "chunk_length": 30, + "dither": 0.0, + "feature_extractor_type": "WhisperFeatureExtractor", + "feature_size": 80, + "hop_length": 160, + "n_fft": 400, + "n_samples": 480000, + "nb_max_frames": 3000, + "padding_side": "right", + "padding_value": 0.0, + "processor_class": "WhisperProcessor", + "return_attention_mask": false, + "sampling_rate": 16000 +} diff --git a/checkpoint-5970/rng_state.pth b/checkpoint-5970/rng_state.pth new file mode 100644 index 0000000000000000000000000000000000000000..040b14b3d91c388a7e1170a221fd0e721109e51d --- /dev/null +++ b/checkpoint-5970/rng_state.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c13bc7cb569fdca884d1a372d168250f9b5593c86e43b09288bbd5ba0834a1a +size 14244 diff --git a/checkpoint-5970/scaler.pt b/checkpoint-5970/scaler.pt new file mode 100644 index 0000000000000000000000000000000000000000..a33dfe6c6039d32049ba8352d730509e8e998b47 --- /dev/null +++ b/checkpoint-5970/scaler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b0cac195d0cc2ca1bcbb4f6eea998f8369cc362d392d823c034405bde50ef1c +size 988 diff --git a/checkpoint-5970/scheduler.pt b/checkpoint-5970/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..b1d7d4dcd70cc24fba3011913ff061fc3b0238b2 --- /dev/null +++ b/checkpoint-5970/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:53ed52a9dbb53965f2982bd8752c670af0f0b2b4eb5f29f36ad54f9c413f21bc +size 1064 diff --git a/checkpoint-5970/trainer_state.json b/checkpoint-5970/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..4a07cae6e5b016a5bb844a8f011d1d2493323970 --- /dev/null +++ b/checkpoint-5970/trainer_state.json @@ -0,0 +1,2219 @@ +{ + "best_global_step": 5500, + "best_metric": 4.431134866851197, + "best_model_checkpoint": "./SALAMA_C3/checkpoint-5500", + "epoch": 10.0, + "eval_steps": 500, + "global_step": 5970, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.03350083752093802, + "grad_norm": 11.757207870483398, + "learning_rate": 6.333333333333334e-07, + "loss": 0.5112, + "step": 20 + }, + { + "epoch": 0.06700167504187604, + "grad_norm": 10.197953224182129, + "learning_rate": 1.3e-06, + "loss": 0.5283, + "step": 40 + }, + { + "epoch": 0.10050251256281408, + "grad_norm": 12.811074256896973, + "learning_rate": 1.9666666666666668e-06, + "loss": 0.474, + "step": 60 + }, + { + "epoch": 0.13400335008375208, + "grad_norm": 8.559342384338379, + "learning_rate": 2.6e-06, + "loss": 0.424, + "step": 80 + }, + { + "epoch": 0.16750418760469013, + "grad_norm": 6.9629225730896, + "learning_rate": 3.266666666666667e-06, + "loss": 0.4465, + "step": 100 + }, + { + "epoch": 0.20100502512562815, + "grad_norm": 7.833058834075928, + "learning_rate": 3.9333333333333335e-06, + "loss": 0.3764, + "step": 120 + }, + { + "epoch": 0.23450586264656617, + "grad_norm": 6.882424831390381, + "learning_rate": 4.600000000000001e-06, + "loss": 0.3705, + "step": 140 + }, + { + "epoch": 0.26800670016750416, + "grad_norm": 5.4183244705200195, + "learning_rate": 5.2666666666666665e-06, + "loss": 0.4116, + "step": 160 + }, + { + "epoch": 0.3015075376884422, + "grad_norm": 6.103787899017334, + "learning_rate": 5.933333333333335e-06, + "loss": 0.3903, + "step": 180 + }, + { + "epoch": 0.33500837520938026, + "grad_norm": 6.084160804748535, + "learning_rate": 6.600000000000001e-06, + "loss": 0.3794, + "step": 200 + }, + { + "epoch": 0.3685092127303183, + "grad_norm": 4.531406879425049, + "learning_rate": 7.266666666666668e-06, + "loss": 0.3904, + "step": 220 + }, + { + "epoch": 0.4020100502512563, + "grad_norm": 6.702854156494141, + "learning_rate": 7.933333333333334e-06, + "loss": 0.3672, + "step": 240 + }, + { + "epoch": 0.4355108877721943, + "grad_norm": 6.386377811431885, + "learning_rate": 8.6e-06, + "loss": 0.3379, + "step": 260 + }, + { + "epoch": 0.46901172529313234, + "grad_norm": 5.33281135559082, + "learning_rate": 9.266666666666667e-06, + "loss": 0.4102, + "step": 280 + }, + { + "epoch": 0.5025125628140703, + "grad_norm": 6.031435012817383, + "learning_rate": 9.933333333333334e-06, + "loss": 0.3815, + "step": 300 + }, + { + "epoch": 0.5360134003350083, + "grad_norm": 4.8638176918029785, + "learning_rate": 9.968253968253969e-06, + "loss": 0.358, + "step": 320 + }, + { + "epoch": 0.5695142378559463, + "grad_norm": 5.352113723754883, + "learning_rate": 9.932980599647268e-06, + "loss": 0.3578, + "step": 340 + }, + { + "epoch": 0.6030150753768844, + "grad_norm": 5.560739040374756, + "learning_rate": 9.897707231040565e-06, + "loss": 0.3795, + "step": 360 + }, + { + "epoch": 0.6365159128978225, + "grad_norm": 6.186940670013428, + "learning_rate": 9.862433862433864e-06, + "loss": 0.3852, + "step": 380 + }, + { + "epoch": 0.6700167504187605, + "grad_norm": 5.37507438659668, + "learning_rate": 9.827160493827161e-06, + "loss": 0.3761, + "step": 400 + }, + { + "epoch": 0.7035175879396985, + "grad_norm": 6.373449802398682, + "learning_rate": 9.79188712522046e-06, + "loss": 0.3674, + "step": 420 + }, + { + "epoch": 0.7370184254606366, + "grad_norm": 5.702625274658203, + "learning_rate": 9.756613756613757e-06, + "loss": 0.4013, + "step": 440 + }, + { + "epoch": 0.7705192629815746, + "grad_norm": 5.5675153732299805, + "learning_rate": 9.721340388007056e-06, + "loss": 0.3495, + "step": 460 + }, + { + "epoch": 0.8040201005025126, + "grad_norm": 6.296374320983887, + "learning_rate": 9.686067019400353e-06, + "loss": 0.3704, + "step": 480 + }, + { + "epoch": 0.8375209380234506, + "grad_norm": 4.830463886260986, + "learning_rate": 9.650793650793652e-06, + "loss": 0.36, + "step": 500 + }, + { + "epoch": 0.8375209380234506, + "eval_loss": 0.27764827013015747, + "eval_runtime": 1752.3826, + "eval_samples_per_second": 2.725, + "eval_steps_per_second": 0.341, + "eval_wer": 19.59291781998664, + "step": 500 + }, + { + "epoch": 0.8710217755443886, + "grad_norm": 4.863058090209961, + "learning_rate": 9.61552028218695e-06, + "loss": 0.3874, + "step": 520 + }, + { + "epoch": 0.9045226130653267, + "grad_norm": 6.121027946472168, + "learning_rate": 9.580246913580248e-06, + "loss": 0.382, + "step": 540 + }, + { + "epoch": 0.9380234505862647, + "grad_norm": 4.538082599639893, + "learning_rate": 9.544973544973546e-06, + "loss": 0.3611, + "step": 560 + }, + { + "epoch": 0.9715242881072027, + "grad_norm": 4.8772382736206055, + "learning_rate": 9.509700176366844e-06, + "loss": 0.3584, + "step": 580 + }, + { + "epoch": 1.0050251256281406, + "grad_norm": 4.039212703704834, + "learning_rate": 9.474426807760142e-06, + "loss": 0.3701, + "step": 600 + }, + { + "epoch": 1.0385259631490786, + "grad_norm": 4.787687301635742, + "learning_rate": 9.43915343915344e-06, + "loss": 0.2058, + "step": 620 + }, + { + "epoch": 1.0720268006700167, + "grad_norm": 4.503021717071533, + "learning_rate": 9.403880070546738e-06, + "loss": 0.2455, + "step": 640 + }, + { + "epoch": 1.1055276381909547, + "grad_norm": 6.663857936859131, + "learning_rate": 9.368606701940036e-06, + "loss": 0.2179, + "step": 660 + }, + { + "epoch": 1.1390284757118927, + "grad_norm": 4.486196041107178, + "learning_rate": 9.333333333333334e-06, + "loss": 0.2196, + "step": 680 + }, + { + "epoch": 1.1725293132328307, + "grad_norm": 4.667060852050781, + "learning_rate": 9.298059964726633e-06, + "loss": 0.2183, + "step": 700 + }, + { + "epoch": 1.2060301507537687, + "grad_norm": 5.608316898345947, + "learning_rate": 9.26278659611993e-06, + "loss": 0.2161, + "step": 720 + }, + { + "epoch": 1.2395309882747068, + "grad_norm": 4.2184271812438965, + "learning_rate": 9.227513227513229e-06, + "loss": 0.2382, + "step": 740 + }, + { + "epoch": 1.2730318257956448, + "grad_norm": 3.871945381164551, + "learning_rate": 9.192239858906526e-06, + "loss": 0.2214, + "step": 760 + }, + { + "epoch": 1.3065326633165828, + "grad_norm": 3.730222225189209, + "learning_rate": 9.156966490299825e-06, + "loss": 0.2213, + "step": 780 + }, + { + "epoch": 1.3400335008375208, + "grad_norm": 4.740777015686035, + "learning_rate": 9.121693121693122e-06, + "loss": 0.2503, + "step": 800 + }, + { + "epoch": 1.3735343383584588, + "grad_norm": 4.123469352722168, + "learning_rate": 9.086419753086421e-06, + "loss": 0.2443, + "step": 820 + }, + { + "epoch": 1.4070351758793969, + "grad_norm": 3.988917827606201, + "learning_rate": 9.051146384479718e-06, + "loss": 0.241, + "step": 840 + }, + { + "epoch": 1.4405360134003349, + "grad_norm": 5.4730305671691895, + "learning_rate": 9.015873015873017e-06, + "loss": 0.2163, + "step": 860 + }, + { + "epoch": 1.474036850921273, + "grad_norm": 5.020652770996094, + "learning_rate": 8.980599647266314e-06, + "loss": 0.2387, + "step": 880 + }, + { + "epoch": 1.507537688442211, + "grad_norm": 4.231486797332764, + "learning_rate": 8.945326278659613e-06, + "loss": 0.2523, + "step": 900 + }, + { + "epoch": 1.541038525963149, + "grad_norm": 6.197975158691406, + "learning_rate": 8.910052910052912e-06, + "loss": 0.2255, + "step": 920 + }, + { + "epoch": 1.574539363484087, + "grad_norm": 5.489210605621338, + "learning_rate": 8.874779541446209e-06, + "loss": 0.2397, + "step": 940 + }, + { + "epoch": 1.608040201005025, + "grad_norm": 4.32737398147583, + "learning_rate": 8.839506172839508e-06, + "loss": 0.2298, + "step": 960 + }, + { + "epoch": 1.641541038525963, + "grad_norm": 4.504214763641357, + "learning_rate": 8.804232804232805e-06, + "loss": 0.2301, + "step": 980 + }, + { + "epoch": 1.675041876046901, + "grad_norm": 3.9694759845733643, + "learning_rate": 8.768959435626104e-06, + "loss": 0.2338, + "step": 1000 + }, + { + "epoch": 1.675041876046901, + "eval_loss": 0.18838582932949066, + "eval_runtime": 1770.3449, + "eval_samples_per_second": 2.698, + "eval_steps_per_second": 0.337, + "eval_wer": 14.071299035983584, + "step": 1000 + }, + { + "epoch": 1.708542713567839, + "grad_norm": 4.972934722900391, + "learning_rate": 8.7336860670194e-06, + "loss": 0.2388, + "step": 1020 + }, + { + "epoch": 1.742043551088777, + "grad_norm": 5.178994655609131, + "learning_rate": 8.6984126984127e-06, + "loss": 0.2531, + "step": 1040 + }, + { + "epoch": 1.775544388609715, + "grad_norm": 3.560372829437256, + "learning_rate": 8.663139329805997e-06, + "loss": 0.2187, + "step": 1060 + }, + { + "epoch": 1.809045226130653, + "grad_norm": 4.647324562072754, + "learning_rate": 8.627865961199296e-06, + "loss": 0.2413, + "step": 1080 + }, + { + "epoch": 1.8425460636515911, + "grad_norm": 3.6784422397613525, + "learning_rate": 8.592592592592593e-06, + "loss": 0.2242, + "step": 1100 + }, + { + "epoch": 1.8760469011725294, + "grad_norm": 5.928018093109131, + "learning_rate": 8.557319223985891e-06, + "loss": 0.2406, + "step": 1120 + }, + { + "epoch": 1.9095477386934674, + "grad_norm": 6.358487606048584, + "learning_rate": 8.52204585537919e-06, + "loss": 0.2482, + "step": 1140 + }, + { + "epoch": 1.9430485762144054, + "grad_norm": 4.409506797790527, + "learning_rate": 8.486772486772487e-06, + "loss": 0.2415, + "step": 1160 + }, + { + "epoch": 1.9765494137353434, + "grad_norm": 5.169639587402344, + "learning_rate": 8.451499118165786e-06, + "loss": 0.222, + "step": 1180 + }, + { + "epoch": 2.0100502512562812, + "grad_norm": 2.8841328620910645, + "learning_rate": 8.416225749559083e-06, + "loss": 0.2109, + "step": 1200 + }, + { + "epoch": 2.0435510887772192, + "grad_norm": 3.835498332977295, + "learning_rate": 8.380952380952382e-06, + "loss": 0.1284, + "step": 1220 + }, + { + "epoch": 2.0770519262981573, + "grad_norm": 2.584859848022461, + "learning_rate": 8.34567901234568e-06, + "loss": 0.1059, + "step": 1240 + }, + { + "epoch": 2.1105527638190953, + "grad_norm": 2.797682285308838, + "learning_rate": 8.310405643738978e-06, + "loss": 0.0967, + "step": 1260 + }, + { + "epoch": 2.1440536013400333, + "grad_norm": 3.2937309741973877, + "learning_rate": 8.275132275132275e-06, + "loss": 0.1267, + "step": 1280 + }, + { + "epoch": 2.1775544388609713, + "grad_norm": 3.344325065612793, + "learning_rate": 8.239858906525574e-06, + "loss": 0.1354, + "step": 1300 + }, + { + "epoch": 2.2110552763819094, + "grad_norm": 2.955522060394287, + "learning_rate": 8.204585537918873e-06, + "loss": 0.104, + "step": 1320 + }, + { + "epoch": 2.2445561139028474, + "grad_norm": 5.180908203125, + "learning_rate": 8.16931216931217e-06, + "loss": 0.1229, + "step": 1340 + }, + { + "epoch": 2.2780569514237854, + "grad_norm": 3.639417886734009, + "learning_rate": 8.134038800705469e-06, + "loss": 0.104, + "step": 1360 + }, + { + "epoch": 2.3115577889447234, + "grad_norm": 3.7380902767181396, + "learning_rate": 8.098765432098766e-06, + "loss": 0.1141, + "step": 1380 + }, + { + "epoch": 2.3450586264656614, + "grad_norm": 4.100318908691406, + "learning_rate": 8.063492063492065e-06, + "loss": 0.1123, + "step": 1400 + }, + { + "epoch": 2.3785594639865995, + "grad_norm": 2.8147048950195312, + "learning_rate": 8.028218694885362e-06, + "loss": 0.1039, + "step": 1420 + }, + { + "epoch": 2.4120603015075375, + "grad_norm": 3.7376410961151123, + "learning_rate": 7.99294532627866e-06, + "loss": 0.1179, + "step": 1440 + }, + { + "epoch": 2.4455611390284755, + "grad_norm": 4.320065975189209, + "learning_rate": 7.957671957671958e-06, + "loss": 0.1167, + "step": 1460 + }, + { + "epoch": 2.4790619765494135, + "grad_norm": 4.415127277374268, + "learning_rate": 7.922398589065257e-06, + "loss": 0.125, + "step": 1480 + }, + { + "epoch": 2.5125628140703515, + "grad_norm": 2.631763219833374, + "learning_rate": 7.887125220458554e-06, + "loss": 0.1188, + "step": 1500 + }, + { + "epoch": 2.5125628140703515, + "eval_loss": 0.13635103404521942, + "eval_runtime": 1806.1099, + "eval_samples_per_second": 2.644, + "eval_steps_per_second": 0.331, + "eval_wer": 11.713753937195762, + "step": 1500 + }, + { + "epoch": 2.5460636515912896, + "grad_norm": 3.608773946762085, + "learning_rate": 7.851851851851853e-06, + "loss": 0.1222, + "step": 1520 + }, + { + "epoch": 2.5795644891122276, + "grad_norm": 3.4494569301605225, + "learning_rate": 7.816578483245151e-06, + "loss": 0.1087, + "step": 1540 + }, + { + "epoch": 2.6130653266331656, + "grad_norm": 3.6112048625946045, + "learning_rate": 7.781305114638449e-06, + "loss": 0.1097, + "step": 1560 + }, + { + "epoch": 2.6465661641541036, + "grad_norm": 3.1978707313537598, + "learning_rate": 7.746031746031747e-06, + "loss": 0.1101, + "step": 1580 + }, + { + "epoch": 2.6800670016750416, + "grad_norm": 3.6376733779907227, + "learning_rate": 7.710758377425045e-06, + "loss": 0.1056, + "step": 1600 + }, + { + "epoch": 2.7135678391959797, + "grad_norm": 3.754915237426758, + "learning_rate": 7.675485008818343e-06, + "loss": 0.1103, + "step": 1620 + }, + { + "epoch": 2.7470686767169177, + "grad_norm": 3.329240322113037, + "learning_rate": 7.64021164021164e-06, + "loss": 0.1193, + "step": 1640 + }, + { + "epoch": 2.7805695142378557, + "grad_norm": 3.9300169944763184, + "learning_rate": 7.604938271604939e-06, + "loss": 0.1232, + "step": 1660 + }, + { + "epoch": 2.8140703517587937, + "grad_norm": 4.290626049041748, + "learning_rate": 7.569664902998237e-06, + "loss": 0.1374, + "step": 1680 + }, + { + "epoch": 2.8475711892797317, + "grad_norm": 3.8987998962402344, + "learning_rate": 7.534391534391535e-06, + "loss": 0.1062, + "step": 1700 + }, + { + "epoch": 2.8810720268006698, + "grad_norm": 4.234960079193115, + "learning_rate": 7.499118165784833e-06, + "loss": 0.111, + "step": 1720 + }, + { + "epoch": 2.914572864321608, + "grad_norm": 3.103458881378174, + "learning_rate": 7.463844797178131e-06, + "loss": 0.1076, + "step": 1740 + }, + { + "epoch": 2.948073701842546, + "grad_norm": 3.270204782485962, + "learning_rate": 7.428571428571429e-06, + "loss": 0.1224, + "step": 1760 + }, + { + "epoch": 2.981574539363484, + "grad_norm": 4.261337757110596, + "learning_rate": 7.393298059964727e-06, + "loss": 0.1224, + "step": 1780 + }, + { + "epoch": 3.0150753768844223, + "grad_norm": 1.9531301259994507, + "learning_rate": 7.358024691358025e-06, + "loss": 0.0913, + "step": 1800 + }, + { + "epoch": 3.0485762144053603, + "grad_norm": 1.866215467453003, + "learning_rate": 7.322751322751324e-06, + "loss": 0.0466, + "step": 1820 + }, + { + "epoch": 3.0820770519262983, + "grad_norm": 3.9456610679626465, + "learning_rate": 7.287477954144622e-06, + "loss": 0.0451, + "step": 1840 + }, + { + "epoch": 3.1155778894472363, + "grad_norm": 1.9518849849700928, + "learning_rate": 7.25220458553792e-06, + "loss": 0.0414, + "step": 1860 + }, + { + "epoch": 3.1490787269681744, + "grad_norm": 2.232792854309082, + "learning_rate": 7.216931216931218e-06, + "loss": 0.0502, + "step": 1880 + }, + { + "epoch": 3.1825795644891124, + "grad_norm": 2.074127674102783, + "learning_rate": 7.181657848324516e-06, + "loss": 0.057, + "step": 1900 + }, + { + "epoch": 3.2160804020100504, + "grad_norm": 3.5036733150482178, + "learning_rate": 7.146384479717814e-06, + "loss": 0.0515, + "step": 1920 + }, + { + "epoch": 3.2495812395309884, + "grad_norm": 2.324014186859131, + "learning_rate": 7.111111111111112e-06, + "loss": 0.0454, + "step": 1940 + }, + { + "epoch": 3.2830820770519265, + "grad_norm": 2.710326671600342, + "learning_rate": 7.07583774250441e-06, + "loss": 0.0473, + "step": 1960 + }, + { + "epoch": 3.3165829145728645, + "grad_norm": 2.8943896293640137, + "learning_rate": 7.040564373897708e-06, + "loss": 0.046, + "step": 1980 + }, + { + "epoch": 3.3500837520938025, + "grad_norm": 2.319986581802368, + "learning_rate": 7.005291005291006e-06, + "loss": 0.0449, + "step": 2000 + }, + { + "epoch": 3.3500837520938025, + "eval_loss": 0.10934468358755112, + "eval_runtime": 1762.4506, + "eval_samples_per_second": 2.71, + "eval_steps_per_second": 0.339, + "eval_wer": 7.172854824854443, + "step": 2000 + }, + { + "epoch": 3.3835845896147405, + "grad_norm": 3.317129135131836, + "learning_rate": 6.9700176366843046e-06, + "loss": 0.0645, + "step": 2020 + }, + { + "epoch": 3.4170854271356785, + "grad_norm": 1.9533768892288208, + "learning_rate": 6.9347442680776025e-06, + "loss": 0.0432, + "step": 2040 + }, + { + "epoch": 3.4505862646566166, + "grad_norm": 2.0508453845977783, + "learning_rate": 6.8994708994709005e-06, + "loss": 0.0521, + "step": 2060 + }, + { + "epoch": 3.4840871021775546, + "grad_norm": 2.163236141204834, + "learning_rate": 6.8641975308641985e-06, + "loss": 0.0529, + "step": 2080 + }, + { + "epoch": 3.5175879396984926, + "grad_norm": 2.7154581546783447, + "learning_rate": 6.8289241622574965e-06, + "loss": 0.0452, + "step": 2100 + }, + { + "epoch": 3.5510887772194306, + "grad_norm": 3.0822432041168213, + "learning_rate": 6.7936507936507944e-06, + "loss": 0.0546, + "step": 2120 + }, + { + "epoch": 3.5845896147403686, + "grad_norm": 4.19010591506958, + "learning_rate": 6.758377425044092e-06, + "loss": 0.0529, + "step": 2140 + }, + { + "epoch": 3.6180904522613067, + "grad_norm": 2.9883594512939453, + "learning_rate": 6.72310405643739e-06, + "loss": 0.0503, + "step": 2160 + }, + { + "epoch": 3.6515912897822447, + "grad_norm": 2.3664371967315674, + "learning_rate": 6.687830687830688e-06, + "loss": 0.0498, + "step": 2180 + }, + { + "epoch": 3.6850921273031827, + "grad_norm": 2.0549991130828857, + "learning_rate": 6.652557319223986e-06, + "loss": 0.051, + "step": 2200 + }, + { + "epoch": 3.7185929648241207, + "grad_norm": 2.5339038372039795, + "learning_rate": 6.617283950617285e-06, + "loss": 0.0568, + "step": 2220 + }, + { + "epoch": 3.7520938023450587, + "grad_norm": 1.9988099336624146, + "learning_rate": 6.582010582010583e-06, + "loss": 0.051, + "step": 2240 + }, + { + "epoch": 3.7855946398659968, + "grad_norm": 2.5243782997131348, + "learning_rate": 6.546737213403881e-06, + "loss": 0.056, + "step": 2260 + }, + { + "epoch": 3.819095477386935, + "grad_norm": 3.157158136367798, + "learning_rate": 6.511463844797179e-06, + "loss": 0.0497, + "step": 2280 + }, + { + "epoch": 3.852596314907873, + "grad_norm": 1.9286202192306519, + "learning_rate": 6.476190476190477e-06, + "loss": 0.0426, + "step": 2300 + }, + { + "epoch": 3.886097152428811, + "grad_norm": 3.808802604675293, + "learning_rate": 6.440917107583775e-06, + "loss": 0.0499, + "step": 2320 + }, + { + "epoch": 3.919597989949749, + "grad_norm": 2.506671667098999, + "learning_rate": 6.405643738977073e-06, + "loss": 0.052, + "step": 2340 + }, + { + "epoch": 3.953098827470687, + "grad_norm": 2.9451920986175537, + "learning_rate": 6.370370370370371e-06, + "loss": 0.0552, + "step": 2360 + }, + { + "epoch": 3.986599664991625, + "grad_norm": 2.592744827270508, + "learning_rate": 6.335097001763669e-06, + "loss": 0.0527, + "step": 2380 + }, + { + "epoch": 4.0201005025125625, + "grad_norm": 1.8891575336456299, + "learning_rate": 6.299823633156967e-06, + "loss": 0.0289, + "step": 2400 + }, + { + "epoch": 4.0536013400335005, + "grad_norm": 1.8053243160247803, + "learning_rate": 6.264550264550266e-06, + "loss": 0.0192, + "step": 2420 + }, + { + "epoch": 4.0871021775544385, + "grad_norm": 2.0084407329559326, + "learning_rate": 6.229276895943564e-06, + "loss": 0.0242, + "step": 2440 + }, + { + "epoch": 4.1206030150753765, + "grad_norm": 1.5919119119644165, + "learning_rate": 6.194003527336862e-06, + "loss": 0.0211, + "step": 2460 + }, + { + "epoch": 4.1541038525963145, + "grad_norm": 1.9214613437652588, + "learning_rate": 6.15873015873016e-06, + "loss": 0.0233, + "step": 2480 + }, + { + "epoch": 4.187604690117253, + "grad_norm": 1.2652311325073242, + "learning_rate": 6.123456790123458e-06, + "loss": 0.0199, + "step": 2500 + }, + { + "epoch": 4.187604690117253, + "eval_loss": 0.0981329157948494, + "eval_runtime": 1779.0213, + "eval_samples_per_second": 2.685, + "eval_steps_per_second": 0.336, + "eval_wer": 6.707549871146321, + "step": 2500 + }, + { + "epoch": 4.221105527638191, + "grad_norm": 2.026528835296631, + "learning_rate": 6.088183421516756e-06, + "loss": 0.0217, + "step": 2520 + }, + { + "epoch": 4.254606365159129, + "grad_norm": 1.596919059753418, + "learning_rate": 6.052910052910054e-06, + "loss": 0.0167, + "step": 2540 + }, + { + "epoch": 4.288107202680067, + "grad_norm": 2.9445090293884277, + "learning_rate": 6.017636684303352e-06, + "loss": 0.0225, + "step": 2560 + }, + { + "epoch": 4.321608040201005, + "grad_norm": 2.4160282611846924, + "learning_rate": 5.9823633156966496e-06, + "loss": 0.0253, + "step": 2580 + }, + { + "epoch": 4.355108877721943, + "grad_norm": 1.461127758026123, + "learning_rate": 5.9470899470899475e-06, + "loss": 0.0197, + "step": 2600 + }, + { + "epoch": 4.388609715242881, + "grad_norm": 2.7892863750457764, + "learning_rate": 5.911816578483246e-06, + "loss": 0.022, + "step": 2620 + }, + { + "epoch": 4.422110552763819, + "grad_norm": 1.651208758354187, + "learning_rate": 5.876543209876544e-06, + "loss": 0.0215, + "step": 2640 + }, + { + "epoch": 4.455611390284757, + "grad_norm": 2.2500391006469727, + "learning_rate": 5.841269841269842e-06, + "loss": 0.0247, + "step": 2660 + }, + { + "epoch": 4.489112227805695, + "grad_norm": 4.447635173797607, + "learning_rate": 5.80599647266314e-06, + "loss": 0.0263, + "step": 2680 + }, + { + "epoch": 4.522613065326633, + "grad_norm": 0.8300407528877258, + "learning_rate": 5.770723104056438e-06, + "loss": 0.0209, + "step": 2700 + }, + { + "epoch": 4.556113902847571, + "grad_norm": 1.6874111890792847, + "learning_rate": 5.735449735449736e-06, + "loss": 0.0195, + "step": 2720 + }, + { + "epoch": 4.589614740368509, + "grad_norm": 2.4045815467834473, + "learning_rate": 5.700176366843034e-06, + "loss": 0.0224, + "step": 2740 + }, + { + "epoch": 4.623115577889447, + "grad_norm": 2.3160908222198486, + "learning_rate": 5.664902998236332e-06, + "loss": 0.0179, + "step": 2760 + }, + { + "epoch": 4.656616415410385, + "grad_norm": 1.6684287786483765, + "learning_rate": 5.62962962962963e-06, + "loss": 0.0238, + "step": 2780 + }, + { + "epoch": 4.690117252931323, + "grad_norm": 1.973906397819519, + "learning_rate": 5.594356261022928e-06, + "loss": 0.0226, + "step": 2800 + }, + { + "epoch": 4.723618090452261, + "grad_norm": 2.270906686782837, + "learning_rate": 5.559082892416227e-06, + "loss": 0.0232, + "step": 2820 + }, + { + "epoch": 4.757118927973199, + "grad_norm": 1.8875011205673218, + "learning_rate": 5.523809523809525e-06, + "loss": 0.0218, + "step": 2840 + }, + { + "epoch": 4.790619765494137, + "grad_norm": 1.1312583684921265, + "learning_rate": 5.488536155202823e-06, + "loss": 0.0212, + "step": 2860 + }, + { + "epoch": 4.824120603015075, + "grad_norm": 0.864783525466919, + "learning_rate": 5.453262786596121e-06, + "loss": 0.023, + "step": 2880 + }, + { + "epoch": 4.857621440536013, + "grad_norm": 1.2935965061187744, + "learning_rate": 5.417989417989419e-06, + "loss": 0.019, + "step": 2900 + }, + { + "epoch": 4.891122278056951, + "grad_norm": 2.4576382637023926, + "learning_rate": 5.382716049382717e-06, + "loss": 0.0193, + "step": 2920 + }, + { + "epoch": 4.924623115577889, + "grad_norm": 2.71472430229187, + "learning_rate": 5.347442680776015e-06, + "loss": 0.0253, + "step": 2940 + }, + { + "epoch": 4.958123953098827, + "grad_norm": 2.84940505027771, + "learning_rate": 5.312169312169313e-06, + "loss": 0.0218, + "step": 2960 + }, + { + "epoch": 4.991624790619765, + "grad_norm": 1.8483999967575073, + "learning_rate": 5.276895943562611e-06, + "loss": 0.0226, + "step": 2980 + }, + { + "epoch": 5.025125628140704, + "grad_norm": 0.6126876473426819, + "learning_rate": 5.241622574955909e-06, + "loss": 0.0101, + "step": 3000 + }, + { + "epoch": 5.025125628140704, + "eval_loss": 0.09390027821063995, + "eval_runtime": 1767.4512, + "eval_samples_per_second": 2.702, + "eval_steps_per_second": 0.338, + "eval_wer": 5.500143170754987, + "step": 3000 + }, + { + "epoch": 5.058626465661642, + "grad_norm": 0.3711394965648651, + "learning_rate": 5.2063492063492076e-06, + "loss": 0.0119, + "step": 3020 + }, + { + "epoch": 5.09212730318258, + "grad_norm": 0.5738839507102966, + "learning_rate": 5.1710758377425055e-06, + "loss": 0.0086, + "step": 3040 + }, + { + "epoch": 5.125628140703517, + "grad_norm": 0.7609245777130127, + "learning_rate": 5.1358024691358035e-06, + "loss": 0.0093, + "step": 3060 + }, + { + "epoch": 5.159128978224456, + "grad_norm": 1.2764722108840942, + "learning_rate": 5.1005291005291015e-06, + "loss": 0.0111, + "step": 3080 + }, + { + "epoch": 5.192629815745394, + "grad_norm": 2.1169776916503906, + "learning_rate": 5.0652557319223995e-06, + "loss": 0.0107, + "step": 3100 + }, + { + "epoch": 5.226130653266332, + "grad_norm": 2.1893081665039062, + "learning_rate": 5.0299823633156974e-06, + "loss": 0.0097, + "step": 3120 + }, + { + "epoch": 5.259631490787269, + "grad_norm": 2.2419638633728027, + "learning_rate": 4.9947089947089946e-06, + "loss": 0.0098, + "step": 3140 + }, + { + "epoch": 5.293132328308207, + "grad_norm": 0.6479611992835999, + "learning_rate": 4.959435626102293e-06, + "loss": 0.0082, + "step": 3160 + }, + { + "epoch": 5.326633165829146, + "grad_norm": 0.4799642860889435, + "learning_rate": 4.924162257495591e-06, + "loss": 0.0119, + "step": 3180 + }, + { + "epoch": 5.360134003350084, + "grad_norm": 0.7716453075408936, + "learning_rate": 4.888888888888889e-06, + "loss": 0.0086, + "step": 3200 + }, + { + "epoch": 5.393634840871022, + "grad_norm": 1.2303547859191895, + "learning_rate": 4.853615520282187e-06, + "loss": 0.009, + "step": 3220 + }, + { + "epoch": 5.42713567839196, + "grad_norm": 0.6345349550247192, + "learning_rate": 4.818342151675485e-06, + "loss": 0.0085, + "step": 3240 + }, + { + "epoch": 5.460636515912898, + "grad_norm": 0.9741530418395996, + "learning_rate": 4.783068783068783e-06, + "loss": 0.0082, + "step": 3260 + }, + { + "epoch": 5.494137353433836, + "grad_norm": 1.1631624698638916, + "learning_rate": 4.747795414462081e-06, + "loss": 0.0086, + "step": 3280 + }, + { + "epoch": 5.527638190954773, + "grad_norm": 0.6502953767776489, + "learning_rate": 4.712522045855379e-06, + "loss": 0.0093, + "step": 3300 + }, + { + "epoch": 5.561139028475711, + "grad_norm": 0.7464337348937988, + "learning_rate": 4.677248677248677e-06, + "loss": 0.0086, + "step": 3320 + }, + { + "epoch": 5.594639865996649, + "grad_norm": 1.017751693725586, + "learning_rate": 4.641975308641975e-06, + "loss": 0.0091, + "step": 3340 + }, + { + "epoch": 5.628140703517588, + "grad_norm": 0.4273395836353302, + "learning_rate": 4.606701940035274e-06, + "loss": 0.0078, + "step": 3360 + }, + { + "epoch": 5.661641541038526, + "grad_norm": 0.6737497448921204, + "learning_rate": 4.571428571428572e-06, + "loss": 0.0106, + "step": 3380 + }, + { + "epoch": 5.695142378559464, + "grad_norm": 1.0791343450546265, + "learning_rate": 4.53615520282187e-06, + "loss": 0.0097, + "step": 3400 + }, + { + "epoch": 5.728643216080402, + "grad_norm": 1.0891772508621216, + "learning_rate": 4.500881834215168e-06, + "loss": 0.0109, + "step": 3420 + }, + { + "epoch": 5.76214405360134, + "grad_norm": 0.7465157508850098, + "learning_rate": 4.465608465608466e-06, + "loss": 0.0078, + "step": 3440 + }, + { + "epoch": 5.795644891122278, + "grad_norm": 0.7693866491317749, + "learning_rate": 4.430335097001764e-06, + "loss": 0.009, + "step": 3460 + }, + { + "epoch": 5.8291457286432165, + "grad_norm": 1.3295698165893555, + "learning_rate": 4.395061728395062e-06, + "loss": 0.016, + "step": 3480 + }, + { + "epoch": 5.8626465661641545, + "grad_norm": 2.4605352878570557, + "learning_rate": 4.35978835978836e-06, + "loss": 0.0081, + "step": 3500 + }, + { + "epoch": 5.8626465661641545, + "eval_loss": 0.09377142041921616, + "eval_runtime": 1773.01, + "eval_samples_per_second": 2.694, + "eval_steps_per_second": 0.337, + "eval_wer": 5.010976424549012, + "step": 3500 + }, + { + "epoch": 5.8961474036850925, + "grad_norm": 0.8250058889389038, + "learning_rate": 4.324514991181658e-06, + "loss": 0.0108, + "step": 3520 + }, + { + "epoch": 5.9296482412060305, + "grad_norm": 1.3606537580490112, + "learning_rate": 4.289241622574956e-06, + "loss": 0.0125, + "step": 3540 + }, + { + "epoch": 5.9631490787269685, + "grad_norm": 0.6893450021743774, + "learning_rate": 4.2539682539682546e-06, + "loss": 0.0101, + "step": 3560 + }, + { + "epoch": 5.9966499162479066, + "grad_norm": 0.8129726052284241, + "learning_rate": 4.2186948853615525e-06, + "loss": 0.0095, + "step": 3580 + }, + { + "epoch": 6.030150753768845, + "grad_norm": 0.5319514274597168, + "learning_rate": 4.1834215167548505e-06, + "loss": 0.0051, + "step": 3600 + }, + { + "epoch": 6.063651591289783, + "grad_norm": 0.20368462800979614, + "learning_rate": 4.1481481481481485e-06, + "loss": 0.0049, + "step": 3620 + }, + { + "epoch": 6.097152428810721, + "grad_norm": 1.1721038818359375, + "learning_rate": 4.1128747795414465e-06, + "loss": 0.0041, + "step": 3640 + }, + { + "epoch": 6.130653266331659, + "grad_norm": 0.17129285633563995, + "learning_rate": 4.0776014109347444e-06, + "loss": 0.0055, + "step": 3660 + }, + { + "epoch": 6.164154103852597, + "grad_norm": 0.31987234950065613, + "learning_rate": 4.042328042328042e-06, + "loss": 0.0037, + "step": 3680 + }, + { + "epoch": 6.197654941373535, + "grad_norm": 0.3214021921157837, + "learning_rate": 4.00705467372134e-06, + "loss": 0.0052, + "step": 3700 + }, + { + "epoch": 6.231155778894473, + "grad_norm": 1.3790876865386963, + "learning_rate": 3.971781305114638e-06, + "loss": 0.0053, + "step": 3720 + }, + { + "epoch": 6.264656616415411, + "grad_norm": 0.549566924571991, + "learning_rate": 3.936507936507936e-06, + "loss": 0.0048, + "step": 3740 + }, + { + "epoch": 6.298157453936349, + "grad_norm": 0.2458494007587433, + "learning_rate": 3.901234567901235e-06, + "loss": 0.006, + "step": 3760 + }, + { + "epoch": 6.331658291457287, + "grad_norm": 2.3661324977874756, + "learning_rate": 3.865961199294533e-06, + "loss": 0.0046, + "step": 3780 + }, + { + "epoch": 6.365159128978225, + "grad_norm": 0.7839031219482422, + "learning_rate": 3.830687830687831e-06, + "loss": 0.0059, + "step": 3800 + }, + { + "epoch": 6.398659966499163, + "grad_norm": 0.2562466263771057, + "learning_rate": 3.795414462081129e-06, + "loss": 0.0038, + "step": 3820 + }, + { + "epoch": 6.432160804020101, + "grad_norm": 0.9680606126785278, + "learning_rate": 3.760141093474427e-06, + "loss": 0.0053, + "step": 3840 + }, + { + "epoch": 6.465661641541039, + "grad_norm": 0.5647270083427429, + "learning_rate": 3.724867724867725e-06, + "loss": 0.0049, + "step": 3860 + }, + { + "epoch": 6.499162479061977, + "grad_norm": 0.5850030183792114, + "learning_rate": 3.689594356261023e-06, + "loss": 0.0039, + "step": 3880 + }, + { + "epoch": 6.532663316582915, + "grad_norm": 0.1745942085981369, + "learning_rate": 3.654320987654321e-06, + "loss": 0.0035, + "step": 3900 + }, + { + "epoch": 6.566164154103853, + "grad_norm": 0.2170235961675644, + "learning_rate": 3.6190476190476194e-06, + "loss": 0.0045, + "step": 3920 + }, + { + "epoch": 6.599664991624791, + "grad_norm": 0.30363383889198303, + "learning_rate": 3.5837742504409174e-06, + "loss": 0.0031, + "step": 3940 + }, + { + "epoch": 6.633165829145729, + "grad_norm": 0.22252851724624634, + "learning_rate": 3.5485008818342153e-06, + "loss": 0.004, + "step": 3960 + }, + { + "epoch": 6.666666666666667, + "grad_norm": 0.9289886951446533, + "learning_rate": 3.5132275132275133e-06, + "loss": 0.0055, + "step": 3980 + }, + { + "epoch": 6.700167504187605, + "grad_norm": 1.094429612159729, + "learning_rate": 3.4779541446208113e-06, + "loss": 0.0041, + "step": 4000 + }, + { + "epoch": 6.700167504187605, + "eval_loss": 0.09379494935274124, + "eval_runtime": 1761.7753, + "eval_samples_per_second": 2.711, + "eval_steps_per_second": 0.339, + "eval_wer": 4.89166746205975, + "step": 4000 + }, + { + "epoch": 6.733668341708543, + "grad_norm": 0.8961842656135559, + "learning_rate": 3.4426807760141097e-06, + "loss": 0.0034, + "step": 4020 + }, + { + "epoch": 6.767169179229481, + "grad_norm": 0.45882654190063477, + "learning_rate": 3.4074074074074077e-06, + "loss": 0.0054, + "step": 4040 + }, + { + "epoch": 6.800670016750419, + "grad_norm": 0.27223679423332214, + "learning_rate": 3.3721340388007056e-06, + "loss": 0.0034, + "step": 4060 + }, + { + "epoch": 6.834170854271357, + "grad_norm": 1.3323800563812256, + "learning_rate": 3.3368606701940036e-06, + "loss": 0.0055, + "step": 4080 + }, + { + "epoch": 6.867671691792295, + "grad_norm": 0.24382267892360687, + "learning_rate": 3.3015873015873016e-06, + "loss": 0.0038, + "step": 4100 + }, + { + "epoch": 6.901172529313233, + "grad_norm": 0.2211693376302719, + "learning_rate": 3.2663139329806e-06, + "loss": 0.0043, + "step": 4120 + }, + { + "epoch": 6.934673366834171, + "grad_norm": 0.20270536839962006, + "learning_rate": 3.231040564373898e-06, + "loss": 0.0047, + "step": 4140 + }, + { + "epoch": 6.968174204355109, + "grad_norm": 2.0083487033843994, + "learning_rate": 3.195767195767196e-06, + "loss": 0.0054, + "step": 4160 + }, + { + "epoch": 7.001675041876047, + "grad_norm": 0.1442304253578186, + "learning_rate": 3.160493827160494e-06, + "loss": 0.005, + "step": 4180 + }, + { + "epoch": 7.035175879396985, + "grad_norm": 0.38336917757987976, + "learning_rate": 3.126984126984127e-06, + "loss": 0.0031, + "step": 4200 + }, + { + "epoch": 7.068676716917923, + "grad_norm": 0.13572201132774353, + "learning_rate": 3.0917107583774254e-06, + "loss": 0.0027, + "step": 4220 + }, + { + "epoch": 7.102177554438861, + "grad_norm": 1.8334097862243652, + "learning_rate": 3.0564373897707234e-06, + "loss": 0.0025, + "step": 4240 + }, + { + "epoch": 7.135678391959799, + "grad_norm": 1.0188051462173462, + "learning_rate": 3.0211640211640214e-06, + "loss": 0.0032, + "step": 4260 + }, + { + "epoch": 7.169179229480737, + "grad_norm": 0.13007956743240356, + "learning_rate": 2.9858906525573194e-06, + "loss": 0.0033, + "step": 4280 + }, + { + "epoch": 7.202680067001675, + "grad_norm": 0.07188330590724945, + "learning_rate": 2.9506172839506173e-06, + "loss": 0.0025, + "step": 4300 + }, + { + "epoch": 7.236180904522613, + "grad_norm": 0.2624744176864624, + "learning_rate": 2.9153439153439157e-06, + "loss": 0.0022, + "step": 4320 + }, + { + "epoch": 7.269681742043551, + "grad_norm": 0.1817062646150589, + "learning_rate": 2.8800705467372137e-06, + "loss": 0.0027, + "step": 4340 + }, + { + "epoch": 7.303182579564489, + "grad_norm": 0.23730872571468353, + "learning_rate": 2.8447971781305117e-06, + "loss": 0.0028, + "step": 4360 + }, + { + "epoch": 7.336683417085427, + "grad_norm": 0.44462695717811584, + "learning_rate": 2.8095238095238096e-06, + "loss": 0.0027, + "step": 4380 + }, + { + "epoch": 7.370184254606365, + "grad_norm": 0.16056309640407562, + "learning_rate": 2.7742504409171076e-06, + "loss": 0.0021, + "step": 4400 + }, + { + "epoch": 7.403685092127303, + "grad_norm": 0.1740381121635437, + "learning_rate": 2.738977072310406e-06, + "loss": 0.0021, + "step": 4420 + }, + { + "epoch": 7.437185929648241, + "grad_norm": 0.2567131519317627, + "learning_rate": 2.703703703703704e-06, + "loss": 0.0025, + "step": 4440 + }, + { + "epoch": 7.4706867671691795, + "grad_norm": 0.22708427906036377, + "learning_rate": 2.668430335097002e-06, + "loss": 0.0018, + "step": 4460 + }, + { + "epoch": 7.5041876046901175, + "grad_norm": 0.14704178273677826, + "learning_rate": 2.6331569664903e-06, + "loss": 0.0023, + "step": 4480 + }, + { + "epoch": 7.5376884422110555, + "grad_norm": 0.26313552260398865, + "learning_rate": 2.597883597883598e-06, + "loss": 0.0026, + "step": 4500 + }, + { + "epoch": 7.5376884422110555, + "eval_loss": 0.09553142637014389, + "eval_runtime": 1387.1198, + "eval_samples_per_second": 3.443, + "eval_steps_per_second": 0.43, + "eval_wer": 4.50510642359454, + "step": 4500 + }, + { + "epoch": 7.5711892797319935, + "grad_norm": 0.09787946194410324, + "learning_rate": 2.5626102292768963e-06, + "loss": 0.0022, + "step": 4520 + }, + { + "epoch": 7.6046901172529315, + "grad_norm": 0.10577196627855301, + "learning_rate": 2.5273368606701943e-06, + "loss": 0.0023, + "step": 4540 + }, + { + "epoch": 7.63819095477387, + "grad_norm": 0.11914186924695969, + "learning_rate": 2.4920634920634923e-06, + "loss": 0.0026, + "step": 4560 + }, + { + "epoch": 7.671691792294808, + "grad_norm": 0.6797345876693726, + "learning_rate": 2.4567901234567902e-06, + "loss": 0.0028, + "step": 4580 + }, + { + "epoch": 7.705192629815746, + "grad_norm": 0.11635720729827881, + "learning_rate": 2.4215167548500882e-06, + "loss": 0.0021, + "step": 4600 + }, + { + "epoch": 7.738693467336684, + "grad_norm": 0.0952395349740982, + "learning_rate": 2.3862433862433866e-06, + "loss": 0.0021, + "step": 4620 + }, + { + "epoch": 7.772194304857622, + "grad_norm": 0.11671995371580124, + "learning_rate": 2.3509700176366846e-06, + "loss": 0.0023, + "step": 4640 + }, + { + "epoch": 7.80569514237856, + "grad_norm": 0.10185588896274567, + "learning_rate": 2.3156966490299826e-06, + "loss": 0.003, + "step": 4660 + }, + { + "epoch": 7.839195979899498, + "grad_norm": 0.11902861297130585, + "learning_rate": 2.2804232804232805e-06, + "loss": 0.0021, + "step": 4680 + }, + { + "epoch": 7.872696817420436, + "grad_norm": 0.10562069714069366, + "learning_rate": 2.2451499118165785e-06, + "loss": 0.002, + "step": 4700 + }, + { + "epoch": 7.906197654941374, + "grad_norm": 0.1272854059934616, + "learning_rate": 2.209876543209877e-06, + "loss": 0.0024, + "step": 4720 + }, + { + "epoch": 7.939698492462312, + "grad_norm": 0.14873459935188293, + "learning_rate": 2.174603174603175e-06, + "loss": 0.0021, + "step": 4740 + }, + { + "epoch": 7.97319932998325, + "grad_norm": 0.09556525200605392, + "learning_rate": 2.139329805996473e-06, + "loss": 0.0021, + "step": 4760 + }, + { + "epoch": 8.006700167504187, + "grad_norm": 0.2545607388019562, + "learning_rate": 2.104056437389771e-06, + "loss": 0.0021, + "step": 4780 + }, + { + "epoch": 8.040201005025125, + "grad_norm": 0.09803763031959534, + "learning_rate": 2.068783068783069e-06, + "loss": 0.0017, + "step": 4800 + }, + { + "epoch": 8.073701842546063, + "grad_norm": 0.6409705281257629, + "learning_rate": 2.0335097001763672e-06, + "loss": 0.0019, + "step": 4820 + }, + { + "epoch": 8.107202680067001, + "grad_norm": 0.08693050593137741, + "learning_rate": 1.998236331569665e-06, + "loss": 0.0017, + "step": 4840 + }, + { + "epoch": 8.140703517587939, + "grad_norm": 0.09300371259450912, + "learning_rate": 1.962962962962963e-06, + "loss": 0.0022, + "step": 4860 + }, + { + "epoch": 8.174204355108877, + "grad_norm": 0.09886649250984192, + "learning_rate": 1.927689594356261e-06, + "loss": 0.0016, + "step": 4880 + }, + { + "epoch": 8.207705192629815, + "grad_norm": 0.14025329053401947, + "learning_rate": 1.8924162257495593e-06, + "loss": 0.0031, + "step": 4900 + }, + { + "epoch": 8.241206030150753, + "grad_norm": 0.06884673237800598, + "learning_rate": 1.8571428571428573e-06, + "loss": 0.0014, + "step": 4920 + }, + { + "epoch": 8.274706867671691, + "grad_norm": 0.08606573939323425, + "learning_rate": 1.8218694885361555e-06, + "loss": 0.0015, + "step": 4940 + }, + { + "epoch": 8.308207705192629, + "grad_norm": 0.5949509739875793, + "learning_rate": 1.7865961199294535e-06, + "loss": 0.0017, + "step": 4960 + }, + { + "epoch": 8.341708542713567, + "grad_norm": 0.07532797753810883, + "learning_rate": 1.7513227513227514e-06, + "loss": 0.0015, + "step": 4980 + }, + { + "epoch": 8.375209380234505, + "grad_norm": 0.07208308577537537, + "learning_rate": 1.7160493827160496e-06, + "loss": 0.0016, + "step": 5000 + }, + { + "epoch": 8.375209380234505, + "eval_loss": 0.0970502495765686, + "eval_runtime": 1406.3953, + "eval_samples_per_second": 3.396, + "eval_steps_per_second": 0.424, + "eval_wer": 4.457382838598836, + "step": 5000 + }, + { + "epoch": 8.408710217755443, + "grad_norm": 0.07992870360612869, + "learning_rate": 1.6807760141093476e-06, + "loss": 0.0015, + "step": 5020 + }, + { + "epoch": 8.442211055276381, + "grad_norm": 0.1169973835349083, + "learning_rate": 1.6455026455026458e-06, + "loss": 0.0018, + "step": 5040 + }, + { + "epoch": 8.47571189279732, + "grad_norm": 0.07965467125177383, + "learning_rate": 1.6102292768959438e-06, + "loss": 0.0015, + "step": 5060 + }, + { + "epoch": 8.509212730318257, + "grad_norm": 0.08486379683017731, + "learning_rate": 1.5749559082892417e-06, + "loss": 0.0016, + "step": 5080 + }, + { + "epoch": 8.542713567839195, + "grad_norm": 0.08790017664432526, + "learning_rate": 1.53968253968254e-06, + "loss": 0.0017, + "step": 5100 + }, + { + "epoch": 8.576214405360133, + "grad_norm": 0.07054130733013153, + "learning_rate": 1.504409171075838e-06, + "loss": 0.0015, + "step": 5120 + }, + { + "epoch": 8.609715242881071, + "grad_norm": 0.09546579420566559, + "learning_rate": 1.469135802469136e-06, + "loss": 0.0017, + "step": 5140 + }, + { + "epoch": 8.64321608040201, + "grad_norm": 0.09177026897668839, + "learning_rate": 1.433862433862434e-06, + "loss": 0.0015, + "step": 5160 + }, + { + "epoch": 8.676716917922947, + "grad_norm": 0.11043737083673477, + "learning_rate": 1.398589065255732e-06, + "loss": 0.0015, + "step": 5180 + }, + { + "epoch": 8.710217755443885, + "grad_norm": 0.09833870083093643, + "learning_rate": 1.3633156966490302e-06, + "loss": 0.0015, + "step": 5200 + }, + { + "epoch": 8.743718592964823, + "grad_norm": 0.07351676374673843, + "learning_rate": 1.3280423280423282e-06, + "loss": 0.0015, + "step": 5220 + }, + { + "epoch": 8.777219430485761, + "grad_norm": 0.09649737179279327, + "learning_rate": 1.2927689594356264e-06, + "loss": 0.0015, + "step": 5240 + }, + { + "epoch": 8.8107202680067, + "grad_norm": 0.08271702378988266, + "learning_rate": 1.2574955908289244e-06, + "loss": 0.0017, + "step": 5260 + }, + { + "epoch": 8.844221105527637, + "grad_norm": 0.08492777496576309, + "learning_rate": 1.2222222222222223e-06, + "loss": 0.0014, + "step": 5280 + }, + { + "epoch": 8.877721943048575, + "grad_norm": 0.11445468664169312, + "learning_rate": 1.1869488536155203e-06, + "loss": 0.0016, + "step": 5300 + }, + { + "epoch": 8.911222780569513, + "grad_norm": 0.08237680047750473, + "learning_rate": 1.1516754850088185e-06, + "loss": 0.0021, + "step": 5320 + }, + { + "epoch": 8.944723618090451, + "grad_norm": 0.3206549882888794, + "learning_rate": 1.1164021164021165e-06, + "loss": 0.0015, + "step": 5340 + }, + { + "epoch": 8.97822445561139, + "grad_norm": 0.0848088189959526, + "learning_rate": 1.0811287477954144e-06, + "loss": 0.0025, + "step": 5360 + }, + { + "epoch": 9.011725293132328, + "grad_norm": 0.06675776094198227, + "learning_rate": 1.0458553791887126e-06, + "loss": 0.0014, + "step": 5380 + }, + { + "epoch": 9.045226130653266, + "grad_norm": 0.054239727556705475, + "learning_rate": 1.0105820105820106e-06, + "loss": 0.0013, + "step": 5400 + }, + { + "epoch": 9.078726968174204, + "grad_norm": 0.1410382091999054, + "learning_rate": 9.753086419753088e-07, + "loss": 0.0013, + "step": 5420 + }, + { + "epoch": 9.112227805695142, + "grad_norm": 0.06337971985340118, + "learning_rate": 9.400352733686068e-07, + "loss": 0.0013, + "step": 5440 + }, + { + "epoch": 9.14572864321608, + "grad_norm": 0.06749581545591354, + "learning_rate": 9.047619047619048e-07, + "loss": 0.0014, + "step": 5460 + }, + { + "epoch": 9.179229480737018, + "grad_norm": 0.07229887694120407, + "learning_rate": 8.694885361552028e-07, + "loss": 0.0013, + "step": 5480 + }, + { + "epoch": 9.212730318257956, + "grad_norm": 0.08598683774471283, + "learning_rate": 8.342151675485009e-07, + "loss": 0.0013, + "step": 5500 + }, + { + "epoch": 9.212730318257956, + "eval_loss": 0.09753644466400146, + "eval_runtime": 1385.4537, + "eval_samples_per_second": 3.447, + "eval_steps_per_second": 0.431, + "eval_wer": 4.431134866851197, + "step": 5500 + }, + { + "epoch": 9.246231155778894, + "grad_norm": 0.06778629124164581, + "learning_rate": 7.98941798941799e-07, + "loss": 0.0019, + "step": 5520 + }, + { + "epoch": 9.279731993299832, + "grad_norm": 0.10274141281843185, + "learning_rate": 7.636684303350971e-07, + "loss": 0.0021, + "step": 5540 + }, + { + "epoch": 9.31323283082077, + "grad_norm": 0.07176781445741653, + "learning_rate": 7.283950617283951e-07, + "loss": 0.0013, + "step": 5560 + }, + { + "epoch": 9.346733668341708, + "grad_norm": 0.07372064888477325, + "learning_rate": 6.931216931216931e-07, + "loss": 0.0014, + "step": 5580 + }, + { + "epoch": 9.380234505862646, + "grad_norm": 0.06419791281223297, + "learning_rate": 6.578483245149912e-07, + "loss": 0.0012, + "step": 5600 + }, + { + "epoch": 9.413735343383584, + "grad_norm": 0.08370620012283325, + "learning_rate": 6.225749559082893e-07, + "loss": 0.0013, + "step": 5620 + }, + { + "epoch": 9.447236180904522, + "grad_norm": 0.07443846017122269, + "learning_rate": 5.873015873015874e-07, + "loss": 0.0013, + "step": 5640 + }, + { + "epoch": 9.48073701842546, + "grad_norm": 0.07495580613613129, + "learning_rate": 5.520282186948854e-07, + "loss": 0.0014, + "step": 5660 + }, + { + "epoch": 9.514237855946398, + "grad_norm": 0.07814870774745941, + "learning_rate": 5.167548500881834e-07, + "loss": 0.0013, + "step": 5680 + }, + { + "epoch": 9.547738693467338, + "grad_norm": 0.06474769860506058, + "learning_rate": 4.814814814814815e-07, + "loss": 0.0014, + "step": 5700 + }, + { + "epoch": 9.581239530988274, + "grad_norm": 0.0897350087761879, + "learning_rate": 4.462081128747796e-07, + "loss": 0.0013, + "step": 5720 + }, + { + "epoch": 9.614740368509214, + "grad_norm": 0.061975665390491486, + "learning_rate": 4.1093474426807766e-07, + "loss": 0.0018, + "step": 5740 + }, + { + "epoch": 9.64824120603015, + "grad_norm": 0.0621723048388958, + "learning_rate": 3.756613756613757e-07, + "loss": 0.0013, + "step": 5760 + }, + { + "epoch": 9.68174204355109, + "grad_norm": 0.06962548196315765, + "learning_rate": 3.4038800705467377e-07, + "loss": 0.0013, + "step": 5780 + }, + { + "epoch": 9.715242881072026, + "grad_norm": 0.0762791708111763, + "learning_rate": 3.051146384479718e-07, + "loss": 0.0014, + "step": 5800 + }, + { + "epoch": 9.748743718592966, + "grad_norm": 0.0735374167561531, + "learning_rate": 2.698412698412699e-07, + "loss": 0.0014, + "step": 5820 + }, + { + "epoch": 9.782244556113902, + "grad_norm": 0.07293818145990372, + "learning_rate": 2.3456790123456793e-07, + "loss": 0.0014, + "step": 5840 + }, + { + "epoch": 9.815745393634842, + "grad_norm": 0.06484886258840561, + "learning_rate": 1.9929453262786599e-07, + "loss": 0.0014, + "step": 5860 + }, + { + "epoch": 9.849246231155778, + "grad_norm": 0.06984104216098785, + "learning_rate": 1.6402116402116401e-07, + "loss": 0.0013, + "step": 5880 + }, + { + "epoch": 9.882747068676718, + "grad_norm": 0.06675717979669571, + "learning_rate": 1.287477954144621e-07, + "loss": 0.0013, + "step": 5900 + }, + { + "epoch": 9.916247906197654, + "grad_norm": 0.06275812536478043, + "learning_rate": 9.347442680776015e-08, + "loss": 0.0013, + "step": 5920 + }, + { + "epoch": 9.949748743718594, + "grad_norm": 0.07500205188989639, + "learning_rate": 5.8201058201058204e-08, + "loss": 0.0014, + "step": 5940 + }, + { + "epoch": 9.98324958123953, + "grad_norm": 0.06431729346513748, + "learning_rate": 2.2927689594356265e-08, + "loss": 0.0013, + "step": 5960 + } + ], + "logging_steps": 20, + "max_steps": 5970, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 5.51226973151232e+19, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-5970/training_args.bin b/checkpoint-5970/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6ffec31ca17b32691fb4cd1278f73919f3259cf4 --- /dev/null +++ b/checkpoint-5970/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41a3c645624f7463029286b7d5e1ad6924cfc44a0d153880431b77c40e5deb73 +size 5432