speecht5_finetuned_facebook_voxpopuli_french
This model is a fine-tuned version of microsoft/speecht5_tts on the voxpopuli dataset. It achieves the following results on the evaluation set:
- Loss: 0.4379
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: 5e-05
- 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: 30
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.4872 | 1.0 | 1584 | 0.4663 |
| 0.4656 | 2.0 | 3168 | 0.4642 |
| 0.4686 | 3.0 | 4752 | 0.4533 |
| 0.4576 | 4.0 | 6336 | 0.4479 |
| 0.4658 | 5.0 | 7920 | 0.4485 |
| 0.4536 | 6.0 | 9504 | 0.4443 |
| 0.4559 | 7.0 | 11088 | 0.4426 |
| 0.449 | 8.0 | 12672 | 0.4410 |
| 0.4469 | 9.0 | 14256 | 0.4420 |
| 0.4565 | 10.0 | 15840 | 0.4402 |
| 0.4428 | 11.0 | 17424 | 0.4470 |
| 0.4412 | 12.0 | 19008 | 0.4400 |
| 0.4437 | 13.0 | 20592 | 0.4396 |
| 0.4395 | 14.0 | 22176 | 0.4385 |
| 0.4461 | 15.0 | 23760 | 0.4407 |
| 0.4401 | 16.0 | 25344 | 0.4387 |
| 0.4407 | 17.0 | 26928 | 0.4379 |
| 0.4359 | 18.0 | 28512 | 0.4384 |
| 0.4338 | 19.0 | 30096 | 0.4387 |
| 0.4326 | 20.0 | 31680 | 0.4381 |
| 0.4406 | 21.0 | 33264 | 0.4390 |
| 0.437 | 22.0 | 34848 | 0.4387 |
| 0.4357 | 23.0 | 36432 | 0.4389 |
| 0.4309 | 24.0 | 38016 | 0.4387 |
| 0.441 | 25.0 | 39600 | 0.4379 |
| 0.4355 | 26.0 | 41184 | 0.4378 |
| 0.4312 | 27.0 | 42768 | 0.4380 |
| 0.4328 | 28.0 | 44352 | 0.4388 |
| 0.4289 | 29.0 | 45936 | 0.4380 |
| 0.4291 | 30.0 | 47520 | 0.4379 |
Framework versions
- Transformers 4.30.0.dev0
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3