metadata
library_name: transformers
license: cc-by-nc-4.0
base_model: facebook/mms-1b-all
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: test_mms
results: []
test_mms
This model is a fine-tuned version of facebook/mms-1b-all on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.8774
- Wer: 0.5800
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: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 32
- optimizer: Use OptimizerNames.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: 20
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.8977 | 3.128 | 100 | 0.9470 | 0.6404 |
| 0.7768 | 6.256 | 200 | 0.9301 | 0.5731 |
| 0.685 | 9.384 | 300 | 0.8985 | 0.5986 |
| 0.6113 | 12.512 | 400 | 0.8818 | 0.5708 |
| 0.5605 | 15.64 | 500 | 0.8921 | 0.5568 |
| 0.5259 | 18.768 | 600 | 0.8774 | 0.5800 |
Framework versions
- Transformers 4.51.3
- Pytorch 2.6.0+cu124
- Datasets 3.6.0
- Tokenizers 0.21.1