metadata
license: mit
base_model: microsoft/speecht5_tts
tags:
- generated_from_trainer
model-index:
- name: ceb_b128_le4_s8000
results: []
ceb_b128_le4_s8000
This model is a fine-tuned version of microsoft/speecht5_tts on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.4096
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: 8
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 2000
- training_steps: 8000
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss |
---|---|---|---|
0.4432 | 39.6040 | 500 | 0.4058 |
0.4129 | 79.2079 | 1000 | 0.3996 |
0.3992 | 118.8119 | 1500 | 0.3986 |
0.3814 | 158.4158 | 2000 | 0.3966 |
0.3688 | 198.0198 | 2500 | 0.4021 |
0.3574 | 237.6238 | 3000 | 0.4000 |
0.3482 | 277.2277 | 3500 | 0.3990 |
0.3421 | 316.8317 | 4000 | 0.4014 |
0.3414 | 356.4356 | 4500 | 0.4059 |
0.3328 | 396.0396 | 5000 | 0.4065 |
0.3277 | 435.6436 | 5500 | 0.4062 |
0.3248 | 475.2475 | 6000 | 0.4069 |
0.3245 | 514.8515 | 6500 | 0.4071 |
0.3219 | 554.4554 | 7000 | 0.4096 |
0.3227 | 594.0594 | 7500 | 0.4112 |
0.3191 | 633.6634 | 8000 | 0.4096 |
Framework versions
- Transformers 4.41.0.dev0
- Pytorch 2.2.1+cu121
- Datasets 2.19.0
- Tokenizers 0.19.1