trocr-mix-font30-s7
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: 0.3770
- Cer: 0.0199
- Wer: 0.0592
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: 7
- 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 |
|---|---|---|---|---|---|
| 5.1224 | 0.1290 | 2000 | 2.2814 | 0.2291 | 0.4434 |
| 3.7824 | 0.2581 | 4000 | 1.7096 | 0.1497 | 0.3183 |
| 3.0008 | 0.3871 | 6000 | 1.3229 | 0.1069 | 0.2508 |
| 2.3375 | 0.5161 | 8000 | 1.1015 | 0.0831 | 0.2016 |
| 2.1721 | 0.6452 | 10000 | 0.9292 | 0.0660 | 0.1689 |
| 1.8688 | 0.7742 | 12000 | 0.8224 | 0.0585 | 0.1494 |
| 1.6774 | 0.9032 | 14000 | 0.7401 | 0.0506 | 0.1324 |
| 1.3136 | 1.0323 | 16000 | 0.6698 | 0.0421 | 0.1147 |
| 1.0305 | 1.1613 | 18000 | 0.6350 | 0.0404 | 0.1094 |
| 1.0998 | 1.2903 | 20000 | 0.5982 | 0.0364 | 0.0993 |
| 1.1230 | 1.4194 | 22000 | 0.5486 | 0.0339 | 0.0940 |
| 1.0980 | 1.5484 | 24000 | 0.5143 | 0.0310 | 0.0860 |
| 0.9279 | 1.6774 | 26000 | 0.4898 | 0.0286 | 0.0817 |
| 0.9557 | 1.8065 | 28000 | 0.4631 | 0.0263 | 0.0759 |
| 0.8726 | 1.9355 | 30000 | 0.4475 | 0.0249 | 0.0724 |
| 0.7155 | 2.0645 | 32000 | 0.4292 | 0.0248 | 0.0712 |
| 0.6998 | 2.1935 | 34000 | 0.4197 | 0.0229 | 0.0668 |
| 0.6664 | 2.3226 | 36000 | 0.4096 | 0.0223 | 0.0660 |
| 0.6647 | 2.4516 | 38000 | 0.3989 | 0.0219 | 0.0640 |
| 0.6329 | 2.5806 | 40000 | 0.3911 | 0.0212 | 0.0623 |
| 0.6395 | 2.7097 | 42000 | 0.3860 | 0.0207 | 0.0621 |
| 0.5718 | 2.8387 | 44000 | 0.3793 | 0.0200 | 0.0598 |
| 0.6878 | 2.9677 | 46000 | 0.3772 | 0.0200 | 0.0594 |
| 0.6329 | 3.0 | 46500 | 0.3770 | 0.0199 | 0.0592 |
Framework versions
- Transformers 5.15.0
- Pytorch 2.11.0+cu128
- Datasets 5.0.1
- Tokenizers 0.22.2