mms-1b-all-indic_voices-60250-hinglish_mixed_scripts-15_100-1e-4-hin-steps-15000-FT
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: 0.6399
- Global Wer: 37.7394
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: 32
- seed: 42
- optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: constant_with_warmup
- lr_scheduler_warmup_steps: 1500
- training_steps: 15000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Global Wer |
|---|---|---|---|---|
| 5.4926 | 0.6752 | 1875 | 1.1575 | 52.2858 |
| 1.3514 | 1.3504 | 3750 | 0.8499 | 44.4782 |
| 1.1159 | 2.0256 | 5625 | 0.7788 | 42.4444 |
| 0.9780 | 2.7008 | 7500 | 0.7223 | 41.0315 |
| 0.9249 | 3.3759 | 9375 | 0.6914 | 40.1209 |
| 0.9062 | 4.0511 | 11250 | 0.6690 | 39.0538 |
| 0.8778 | 4.7263 | 13125 | 0.6526 | 38.2085 |
| 0.8414 | 5.4015 | 15000 | 0.6399 | 37.7394 |
Framework versions
- Transformers 5.0.0.dev0
- Pytorch 2.9.0+cu126
- Datasets 3.6.0
- Tokenizers 0.22.2