metadata
library_name: transformers
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: MSP-Visual
results: []
MSP-Visual
This model is a fine-tuned version of on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.3334
- Wer: 0.6493
- Cer: 0.3725
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: 0.0001
- train_batch_size: 32
- eval_batch_size: 32
- 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: cosine
- lr_scheduler_warmup_steps: 1000.0
- training_steps: 20000
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 2.5771 | 0.05 | 1000 | 2.3988 | 0.9791 | 0.6114 |
| 2.0987 | 0.1 | 2000 | 1.7770 | 0.8288 | 0.4786 |
| 1.9764 | 0.15 | 3000 | 1.6459 | 0.7827 | 0.4485 |
| 1.9493 | 0.2 | 4000 | 1.6139 | 0.7614 | 0.4388 |
| 1.9069 | 0.25 | 5000 | 1.5498 | 0.7351 | 0.4191 |
| 1.9144 | 0.3 | 6000 | 1.5212 | 0.7212 | 0.4142 |
| 1.8289 | 0.35 | 7000 | 1.4857 | 0.7139 | 0.4063 |
| 1.8845 | 0.4 | 8000 | 1.4832 | 0.6958 | 0.3979 |
| 1.7763 | 0.45 | 9000 | 1.4466 | 0.6938 | 0.3946 |
| 1.9370 | 0.5 | 10000 | 1.4235 | 0.6825 | 0.3916 |
| 1.7678 | 0.55 | 11000 | 1.4164 | 0.6784 | 0.3857 |
| 1.8403 | 0.6 | 12000 | 1.3981 | 0.6696 | 0.3868 |
| 1.6723 | 0.65 | 13000 | 1.3849 | 0.6631 | 0.3769 |
| 1.7040 | 0.7 | 14000 | 1.3884 | 0.6579 | 0.3763 |
| 1.5821 | 0.75 | 15000 | 1.3599 | 0.6588 | 0.3756 |
| 1.5721 | 0.8 | 16000 | 1.3480 | 0.6519 | 0.3728 |
| 1.6635 | 0.85 | 17000 | 1.3432 | 0.6527 | 0.3732 |
| 1.6557 | 0.9 | 18000 | 1.3501 | 0.6525 | 0.3748 |
| 1.7992 | 0.95 | 19000 | 1.3334 | 0.6493 | 0.3725 |
| 1.7090 | 1.0 | 20000 | 1.3328 | 0.6499 | 0.3724 |
Framework versions
- Transformers 5.10.2
- Pytorch 2.8.0+cu128
- Datasets 4.0.0
- Tokenizers 0.22.2