<!-- 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. -->
wavlm2phone-jdrt-fw07-5N-holdout
This model is a fine-tuned version of jonatasgrosman/exp_w2v2t_ja_wavlm_s729 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.3838
- Wer: 0.8698
- Cer: 0.6738
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: 64
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 15
Training results
Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
---|---|---|---|---|---|
6.5546 | 1.0 | 242 | 3.8755 | 0.9769 | 0.9718 |
1.7774 | 2.0 | 484 | 1.6824 | 0.8979 | 0.7440 |
1.672 | 3.0 | 726 | 1.5816 | 0.8869 | 0.7067 |
1.6271 | 4.0 | 968 | 1.5313 | 0.8838 | 0.7010 |
1.6116 | 5.0 | 1210 | 1.4975 | 0.8803 | 0.6955 |
1.5788 | 6.0 | 1452 | 1.4718 | 0.8830 | 0.6945 |
1.5747 | 7.0 | 1694 | 1.4493 | 0.8764 | 0.6869 |
1.5625 | 8.0 | 1936 | 1.4322 | 0.8738 | 0.6834 |
1.5531 | 9.0 | 2178 | 1.4194 | 0.8730 | 0.6821 |
1.5504 | 10.0 | 2420 | 1.4081 | 0.8733 | 0.6803 |
1.5405 | 11.0 | 2662 | 1.3992 | 0.8723 | 0.6781 |
1.5433 | 12.0 | 2904 | 1.3921 | 0.8711 | 0.6765 |
1.5365 | 13.0 | 3146 | 1.3874 | 0.8719 | 0.6761 |
1.5345 | 14.0 | 3388 | 1.3847 | 0.8699 | 0.6743 |
1.5351 | 15.0 | 3630 | 1.3838 | 0.8698 | 0.6738 |
Framework versions
- Transformers 4.34.0.dev0
- Pytorch 2.0.1+cu118
- Datasets 2.14.5
- Tokenizers 0.13.3