ssc-tob-mms-model-mix-adapt-max
This model is a fine-tuned version of facebook/mms-1b-all on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.7327
- Cer: 0.1791
- Wer: 0.6105
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: 0.001
- train_batch_size: 8
- eval_batch_size: 6
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer | Wer |
|---|---|---|---|---|---|
| 0.8544 | 0.7648 | 200 | 0.8135 | 0.1852 | 0.6245 |
| 0.7251 | 1.5277 | 400 | 0.7700 | 0.1858 | 0.6342 |
| 0.6716 | 2.2906 | 600 | 0.7553 | 0.1804 | 0.6147 |
| 0.6364 | 3.0535 | 800 | 0.7441 | 0.1782 | 0.6124 |
| 0.602 | 3.8184 | 1000 | 0.7339 | 0.1791 | 0.6115 |
| 0.6208 | 4.5813 | 1200 | 0.7327 | 0.1791 | 0.6105 |
Framework versions
- Transformers 4.57.2
- Pytorch 2.9.1+cu128
- Datasets 3.6.0
- Tokenizers 0.22.0