trocr-webfonts4
This model is a fine-tuned version of cyttic/exp2-frozen-benyehuda-cont on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.3953
- Cer: 0.0197
- Wer: 0.0562
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: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- 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: 4650
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer | Wer |
|---|---|---|---|---|---|
| 4.4492 | 0.1290 | 2000 | 1.9763 | 0.1759 | 0.3689 |
| 3.4292 | 0.2581 | 4000 | 1.4986 | 0.1173 | 0.2731 |
| 2.6827 | 0.3871 | 6000 | 1.2482 | 0.0917 | 0.2162 |
| 2.1522 | 0.5161 | 8000 | 0.9696 | 0.0692 | 0.1704 |
| 2.0142 | 0.6452 | 10000 | 0.8414 | 0.0558 | 0.1457 |
| 1.6561 | 0.7742 | 12000 | 0.7844 | 0.0516 | 0.1301 |
| 1.5216 | 0.9032 | 14000 | 0.6817 | 0.0428 | 0.1120 |
| 1.1658 | 1.0323 | 16000 | 0.6373 | 0.0373 | 0.1007 |
| 1.1099 | 1.1613 | 18000 | 0.5963 | 0.0336 | 0.0921 |
| 1.1093 | 1.2903 | 20000 | 0.5565 | 0.0311 | 0.0855 |
| 0.9545 | 1.4194 | 22000 | 0.5321 | 0.0294 | 0.0818 |
| 1.0267 | 1.5484 | 24000 | 0.5002 | 0.0274 | 0.0764 |
| 0.9041 | 1.6774 | 26000 | 0.4871 | 0.0269 | 0.0739 |
| 0.8422 | 1.8065 | 28000 | 0.4622 | 0.0249 | 0.0696 |
| 0.9088 | 1.9355 | 30000 | 0.4491 | 0.0236 | 0.0672 |
| 0.6078 | 2.0645 | 32000 | 0.4386 | 0.0236 | 0.0648 |
| 0.7209 | 2.1935 | 34000 | 0.4291 | 0.0226 | 0.0633 |
| 0.6248 | 2.3226 | 36000 | 0.4210 | 0.0215 | 0.0604 |
| 0.5949 | 2.4516 | 38000 | 0.4147 | 0.0215 | 0.0601 |
| 0.5598 | 2.5806 | 40000 | 0.4074 | 0.0202 | 0.0571 |
| 0.5955 | 2.7097 | 42000 | 0.4030 | 0.0203 | 0.0573 |
| 0.6750 | 2.8387 | 44000 | 0.3976 | 0.0199 | 0.0569 |
| 0.6013 | 2.9677 | 46000 | 0.3955 | 0.0199 | 0.0566 |
| 0.6690 | 3.0 | 46500 | 0.3953 | 0.0197 | 0.0562 |
Framework versions
- Transformers 5.15.0
- Pytorch 2.11.0+cu128
- Datasets 5.0.1
- Tokenizers 0.22.2