metadata
library_name: transformers
language:
- bn
license: mit
base_model: microsoft/speecht5_tts
tags:
- Bengali
- generated_from_trainer
datasets:
- mozilla-foundation/common_voice_17_0
model-index:
- name: SpeechT5-tuned-bn
results: []
SpeechT5-tuned-bn
This model is a fine-tuned version of microsoft/speecht5_tts on the common_voice_17_0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.5420
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: 4
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 32
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- num_epochs: 3.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 5.7778 | 0.3410 | 100 | 0.6444 |
| 5.2875 | 0.6820 | 200 | 0.6133 |
| 4.977 | 1.0230 | 300 | 0.5806 |
| 4.8498 | 1.3640 | 400 | 0.5748 |
| 4.8067 | 1.7050 | 500 | 0.5627 |
| 4.6922 | 2.0460 | 600 | 0.5534 |
| 4.6079 | 2.3870 | 700 | 0.5460 |
| 4.6776 | 2.7280 | 800 | 0.5420 |
Framework versions
- Transformers 4.46.0.dev0
- Pytorch 2.4.1+cu121
- Datasets 3.0.1
- Tokenizers 0.20.1