mms-1b-all-bigc_mini-drodat-gdro-eta-3.15e-4-dat-0.1454-no-sd-0.00-52
This model is a fine-tuned version of facebook/mms-1b-all on the BIGC_MINI - BEM dataset. It achieves the following results on the evaluation set:
- Loss: 35.5156
- Cer: 0.1254
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 |
|---|---|---|---|---|
| 189.4137 | 0.9579 | 500 | 177.5592 | 0.9943 |
| 42.5653 | 1.9157 | 1000 | 42.4759 | 0.1418 |
| 40.4784 | 2.8736 | 1500 | 40.9850 | 0.1376 |
| 39.818 | 3.8314 | 2000 | 40.0258 | 0.1355 |
| 37.5954 | 4.7893 | 2500 | 40.8527 | 0.1501 |
| 36.5888 | 5.7471 | 3000 | 39.5709 | 0.1325 |
| 34.5414 | 6.7050 | 3500 | 41.2238 | 0.1639 |
| 32.9927 | 7.6628 | 4000 | 37.4801 | 0.1361 |
| 32.7713 | 8.6207 | 4500 | 37.6774 | 0.1471 |
| 35.2549 | 9.5785 | 5000 | 36.5540 | 0.1323 |
| 34.9088 | 10.5364 | 5500 | 36.0874 | 0.1276 |
| 31.3475 | 11.4943 | 6000 | 35.5150 | 0.1254 |
| 34.6237 | 12.4521 | 6500 | 35.7899 | 0.1264 |
| 31.1917 | 13.4100 | 7000 | 35.7217 | 0.1260 |
| 29.9408 | 14.3678 | 7500 | 36.1161 | 0.1365 |
Framework versions
- Transformers 4.53.0
- Pytorch 2.10.0+cu128
- Datasets 4.5.0
- Tokenizers 0.21.4