trocr-small-2style-hh104-hh1-stage1-frozen
This model is a fine-tuned version of cyttic/trocr-hebrew-small-untrained on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 4.8153
- Cer: 0.8034
- Wer: 1.0763
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: 5e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- 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: 0.1
- num_epochs: 1
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer | Wer |
|---|---|---|---|---|---|
| 20.9006 | 0.2581 | 2000 | 5.1620 | 0.7917 | 1.0318 |
| 20.2420 | 0.5161 | 4000 | 4.9747 | 0.8044 | 1.0749 |
| 19.6095 | 0.7742 | 6000 | 4.8615 | 0.8131 | 1.1179 |
| 19.3409 | 1.0 | 7750 | 4.8153 | 0.8034 | 1.0763 |
Framework versions
- Transformers 5.14.1
- Pytorch 2.10.0+cu128
- Datasets 5.0.0
- Tokenizers 0.22.2