SpeechT5 STT Wav2Vec2
This model is a fine-tuned version of facebook/wav2vec2-base-960h on the Lj-Speech dataset. It achieves the following results on the evaluation set:
- Loss: 509.3571
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0001
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 19536.3975 | 0.1376 | 50 | 8125.4639 |
| 2673.9081 | 0.2751 | 100 | 909.8571 |
| 1958.4278 | 0.4127 | 150 | 544.6085 |
| 1268.7548 | 0.5502 | 200 | 555.2729 |
| 1504.7081 | 0.6878 | 250 | 520.7637 |
| 1322.1669 | 0.8253 | 300 | 572.5987 |
| 1331.9734 | 0.9629 | 350 | 514.8672 |
| 1149.1491 | 1.1004 | 400 | 525.9183 |
| 1063.02 | 1.2380 | 450 | 511.6159 |
| 1063.2695 | 1.3755 | 500 | 521.9377 |
| 1037.6037 | 1.5131 | 550 | 511.7293 |
| 1065.5638 | 1.6506 | 600 | 510.2425 |
| 1025.7576 | 1.7882 | 650 | 506.2704 |
| 1132.412 | 1.9257 | 700 | 525.5427 |
| 1033.8723 | 2.0633 | 750 | 506.9381 |
| 1027.0328 | 2.2008 | 800 | 513.5829 |
| 1024.9632 | 2.3384 | 850 | 518.4105 |
| 1023.1637 | 2.4759 | 900 | 515.6079 |
| 1006.7498 | 2.6135 | 950 | 513.5686 |
| 1026.8645 | 2.7510 | 1000 | 507.8027 |
| 1026.9354 | 2.8886 | 1050 | 509.3571 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.3.0+cu121
- Datasets 2.19.2
- Tokenizers 0.19.1
- Downloads last month
- 8