mms-1b-all-bemgen-combined-train-drodat-gdro-1.65e-4-dat-0.019-no-sd-0.00-52
This model is a fine-tuned version of facebook/mms-1b-all on the BEMGEN - BEM dataset. It achieves the following results on the evaluation set:
- Loss: 12.0449
- Cer: 0.0516
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: 0.0003
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- 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: 500
- num_epochs: 30.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 266.4082 | 0.6349 | 500 | 264.2315 | 0.9285 |
| 28.711 | 1.2692 | 1000 | 14.1773 | 0.0644 |
| 30.0752 | 1.9041 | 1500 | 13.9650 | 0.0626 |
| 27.8423 | 2.5384 | 2000 | 13.5464 | 0.0604 |
| 24.7195 | 3.1727 | 2500 | 13.4000 | 0.0582 |
| 25.4105 | 3.8076 | 3000 | 13.0321 | 0.0580 |
| 25.0274 | 4.4419 | 3500 | 13.1646 | 0.0567 |
| 25.4865 | 5.0762 | 4000 | 12.8203 | 0.0571 |
| 22.8274 | 5.7111 | 4500 | 12.9346 | 0.0562 |
| 22.1437 | 6.3454 | 5000 | 12.7188 | 0.0560 |
| 22.6616 | 6.9803 | 5500 | 12.5919 | 0.0544 |
| 20.6103 | 7.6146 | 6000 | 12.5477 | 0.0548 |
| 21.7378 | 8.2489 | 6500 | 12.8438 | 0.0550 |
| 19.8644 | 8.8838 | 7000 | 12.3475 | 0.0534 |
| 20.2197 | 9.5181 | 7500 | 12.2833 | 0.0537 |
| 18.3269 | 10.1524 | 8000 | 12.4400 | 0.0534 |
| 19.7477 | 10.7873 | 8500 | 12.2490 | 0.0525 |
| 18.2041 | 11.4216 | 9000 | 12.5188 | 0.0529 |
| 19.6739 | 12.0559 | 9500 | 12.2462 | 0.0531 |
| 19.4572 | 12.6908 | 10000 | 12.6703 | 0.0537 |
| 17.9485 | 13.3251 | 10500 | 12.3024 | 0.0530 |
| 17.986 | 13.96 | 11000 | 12.1819 | 0.0522 |
| 17.1998 | 14.5943 | 11500 | 12.2845 | 0.0529 |
| 17.5421 | 15.2286 | 12000 | 12.1661 | 0.0526 |
| 17.6232 | 15.8635 | 12500 | 12.0449 | 0.0516 |
| 16.7226 | 16.4978 | 13000 | 12.2436 | 0.0519 |
| 17.3574 | 17.1321 | 13500 | 12.2116 | 0.0524 |
| 16.4421 | 17.7670 | 14000 | 12.2605 | 0.0521 |
Framework versions
- Transformers 4.53.0
- Pytorch 2.10.0+cu128
- Datasets 4.5.0
- Tokenizers 0.21.4