<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
med_v1_M07
This model is a fine-tuned version of facebook/wav2vec2-base on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 3.0346
- Wer: 1.0
- Cer: 0.9247
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: 0.0002
- train_batch_size: 10
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 2575
- num_epochs: 100
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
---|---|---|---|---|---|
40.0631 | 5.0 | 2575 | 3.4774 | 1.0 | 1.0 |
3.1345 | 10.0 | 5150 | 3.2955 | 1.0 | 0.9983 |
2.8502 | 15.0 | 7725 | 3.1890 | 1.0 | 0.9922 |
2.7005 | 20.0 | 10300 | 3.0885 | 1.0 | 0.9784 |
2.6038 | 25.0 | 12875 | 3.0523 | 1.0 | 0.9628 |
2.5369 | 30.0 | 15450 | 3.0120 | 1.0 | 0.9506 |
2.4897 | 35.0 | 18025 | 3.0113 | 1.0 | 0.9463 |
2.4539 | 40.0 | 20600 | 3.0255 | 1.0 | 0.9411 |
2.4249 | 45.0 | 23175 | 2.9973 | 1.0 | 0.9377 |
2.401 | 50.0 | 25750 | 3.0020 | 1.0 | 0.9351 |
2.3787 | 55.0 | 28325 | 3.0206 | 1.0 | 0.9359 |
2.3686 | 60.0 | 30900 | 3.0332 | 1.0 | 0.9333 |
2.3519 | 65.0 | 33475 | 3.0239 | 1.0 | 0.9307 |
2.3448 | 70.0 | 36050 | 3.0209 | 1.0 | 0.9247 |
2.3341 | 75.0 | 38625 | 3.0215 | 1.0 | 0.9273 |
2.3214 | 80.0 | 41200 | 3.0338 | 1.0 | 0.9290 |
2.3182 | 85.0 | 43775 | 3.0332 | 1.0 | 0.9281 |
2.3169 | 90.0 | 46350 | 3.0343 | 1.0 | 0.9238 |
2.313 | 95.0 | 48925 | 3.0355 | 1.0 | 0.9247 |
2.3115 | 100.0 | 51500 | 3.0346 | 1.0 | 0.9247 |
Framework versions
- Transformers 4.18.0
- Pytorch 1.10.2+cu102
- Datasets 2.3.2
- Tokenizers 0.12.1