exp26-composed1m-stage1-frozen
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: 1.5778
- Cer: 0.2897
- Wer: 0.5149
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
- 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: 0.05
- num_epochs: 1
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer | Wer |
|---|---|---|---|---|---|
| 4.1680 | 0.0401 | 2500 | 4.0129 | 0.5637 | 0.8836 |
| 3.8516 | 0.0802 | 5000 | 3.5477 | 0.5362 | 0.8356 |
| 3.4272 | 0.1202 | 7500 | 3.3947 | 0.5073 | 0.7942 |
| 3.2142 | 0.1603 | 10000 | 2.9534 | 0.4854 | 0.7841 |
| 2.9221 | 0.2004 | 12500 | 2.8703 | 0.4708 | 0.7600 |
| 2.7607 | 0.2405 | 15000 | 2.6533 | 0.4470 | 0.7375 |
| 2.6423 | 0.2806 | 17500 | 2.4823 | 0.4345 | 0.7206 |
| 2.5646 | 0.3206 | 20000 | 2.3793 | 0.4062 | 0.6851 |
| 2.5511 | 0.3607 | 22500 | 2.2873 | 0.4083 | 0.6807 |
| 2.4239 | 0.4008 | 25000 | 2.2064 | 0.3872 | 0.6626 |
| 2.3392 | 0.4409 | 27500 | 2.1431 | 0.3828 | 0.6413 |
| 2.2148 | 0.4810 | 30000 | 2.0546 | 0.3737 | 0.6341 |
| 2.0883 | 0.5210 | 32500 | 1.9973 | 0.3574 | 0.6044 |
| 2.1791 | 0.5611 | 35000 | 1.9270 | 0.3557 | 0.6149 |
| 2.0269 | 0.6012 | 37500 | 1.8892 | 0.3371 | 0.5820 |
| 1.9764 | 0.6413 | 40000 | 1.8271 | 0.3287 | 0.5711 |
| 2.0094 | 0.6814 | 42500 | 1.7845 | 0.3262 | 0.5665 |
| 1.9356 | 0.7214 | 45000 | 1.7627 | 0.3225 | 0.5577 |
| 1.9026 | 0.7615 | 47500 | 1.7126 | 0.3112 | 0.5450 |
| 1.8774 | 0.8016 | 50000 | 1.6698 | 0.3107 | 0.5433 |
| 1.8034 | 0.8417 | 52500 | 1.6496 | 0.3005 | 0.5323 |
| 1.8590 | 0.8818 | 55000 | 1.6276 | 0.2965 | 0.5203 |
| 1.7828 | 0.9218 | 57500 | 1.6020 | 0.2905 | 0.5120 |
| 1.7827 | 0.9619 | 60000 | 1.5880 | 0.2912 | 0.5141 |
| 1.8088 | 1.0 | 62375 | 1.5778 | 0.2897 | 0.5149 |
Framework versions
- Transformers 5.12.1
- Pytorch 2.11.0+cu128
- Datasets 4.0.0
- Tokenizers 0.22.2