git-base-500
This model is a fine-tuned version of microsoft/git-base on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.4804
- Wer Score: 8.3286
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: 2
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 16
- 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
- num_epochs: 50
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer Score |
|---|---|---|---|---|
| 3.6377 | 0.8 | 50 | 4.5566 | 8.9044 |
| 1.224 | 1.592 | 100 | 0.7253 | 1.1039 |
| 0.2503 | 2.384 | 150 | 0.4574 | 0.9295 |
| 0.2197 | 3.176 | 200 | 0.4328 | 1.8787 |
| 0.1877 | 3.976 | 250 | 0.4220 | 1.2766 |
| 0.1712 | 4.768 | 300 | 0.4191 | 5.0394 |
| 0.1551 | 5.5600 | 350 | 0.4194 | 3.4952 |
| 0.1555 | 6.352 | 400 | 0.4185 | 3.5920 |
| 0.1405 | 7.144 | 450 | 0.4188 | 8.9391 |
| 0.1339 | 7.944 | 500 | 0.4157 | 3.1930 |
| 0.1236 | 8.736 | 550 | 0.4161 | 3.3608 |
| 0.1179 | 9.528 | 600 | 0.4179 | 4.5639 |
| 0.1143 | 10.32 | 650 | 0.4203 | 7.8746 |
| 0.0984 | 11.112 | 700 | 0.4218 | 3.6105 |
| 0.0977 | 11.912 | 750 | 0.4210 | 6.4683 |
| 0.0896 | 12.704 | 800 | 0.4242 | 8.9421 |
| 0.0876 | 13.496 | 850 | 0.4268 | 6.4528 |
| 0.0791 | 14.288 | 900 | 0.4296 | 8.9367 |
| 0.0861 | 15.08 | 950 | 0.4295 | 8.9462 |
| 0.0713 | 15.88 | 1000 | 0.4345 | 5.5215 |
| 0.0706 | 16.672 | 1050 | 0.4319 | 8.9839 |
| 0.0659 | 17.464 | 1100 | 0.4378 | 7.1995 |
| 0.0633 | 18.256 | 1150 | 0.4366 | 8.9056 |
| 0.0612 | 19.048 | 1200 | 0.4425 | 7.9755 |
| 0.0601 | 19.848 | 1250 | 0.4428 | 7.8967 |
| 0.054 | 20.64 | 1300 | 0.4432 | 8.9534 |
| 0.053 | 21.432 | 1350 | 0.4470 | 8.7013 |
| 0.0528 | 22.224 | 1400 | 0.4489 | 7.5454 |
| 0.0522 | 23.016 | 1450 | 0.4488 | 4.6481 |
| 0.0471 | 23.816 | 1500 | 0.4528 | 8.6278 |
| 0.0458 | 24.608 | 1550 | 0.4552 | 8.7969 |
| 0.0414 | 25.4 | 1600 | 0.4549 | 8.6983 |
| 0.0415 | 26.192 | 1650 | 0.4553 | 5.8256 |
| 0.0398 | 26.992 | 1700 | 0.4584 | 7.7479 |
| 0.0395 | 27.784 | 1750 | 0.4587 | 8.6667 |
| 0.0359 | 28.576 | 1800 | 0.4622 | 8.4624 |
| 0.0347 | 29.368 | 1850 | 0.4610 | 6.7873 |
| 0.034 | 30.16 | 1900 | 0.4634 | 8.4725 |
| 0.0328 | 30.96 | 1950 | 0.4655 | 8.4480 |
| 0.0322 | 31.752 | 2000 | 0.4658 | 8.4839 |
| 0.0301 | 32.544 | 2050 | 0.4655 | 8.4976 |
| 0.0308 | 33.336 | 2100 | 0.4673 | 4.3256 |
| 0.0275 | 34.128 | 2150 | 0.4699 | 7.0305 |
| 0.0286 | 34.928 | 2200 | 0.4708 | 4.2640 |
| 0.0277 | 35.72 | 2250 | 0.4722 | 8.4534 |
| 0.0279 | 36.512 | 2300 | 0.4712 | 8.4665 |
| 0.0262 | 37.304 | 2350 | 0.4723 | 5.3303 |
| 0.026 | 38.096 | 2400 | 0.4735 | 8.4020 |
| 0.0241 | 38.896 | 2450 | 0.4750 | 8.4779 |
| 0.0241 | 39.688 | 2500 | 0.4750 | 8.4194 |
| 0.024 | 40.48 | 2550 | 0.4765 | 7.0854 |
| 0.0226 | 41.272 | 2600 | 0.4788 | 8.4164 |
| 0.0215 | 42.064 | 2650 | 0.4766 | 3.8375 |
| 0.0227 | 42.864 | 2700 | 0.4768 | 3.4355 |
| 0.0223 | 43.656 | 2750 | 0.4770 | 3.7461 |
| 0.0211 | 44.448 | 2800 | 0.4780 | 4.7300 |
| 0.0222 | 45.24 | 2850 | 0.4785 | 3.8919 |
| 0.0211 | 46.032 | 2900 | 0.4796 | 7.5161 |
| 0.02 | 46.832 | 2950 | 0.4797 | 7.9827 |
| 0.0212 | 47.624 | 3000 | 0.4797 | 8.0400 |
| 0.0202 | 48.416 | 3050 | 0.4807 | 8.3190 |
| 0.02 | 49.208 | 3100 | 0.4799 | 8.3375 |
| 0.0206 | 50.0 | 3150 | 0.4804 | 8.3286 |
Framework versions
- Transformers 4.52.4
- Pytorch 2.6.0+cu124
- Datasets 3.6.0
- Tokenizers 0.21.2