<!-- 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. -->
wav2vec2phone-large-xlsr-jp-jdrt5N-demo3
This model is a fine-tuned version of jonatasgrosman/wav2vec2-large-xlsr-53-japanese on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.7083
- Wer: 0.6203
- Cer: 0.5618
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-06
- train_batch_size: 16
- eval_batch_size: 16
- seed: 4
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
Training results
Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
---|---|---|---|---|---|
1.658 | 1.0 | 567 | 1.5635 | 0.8506 | 0.6068 |
1.5568 | 2.0 | 1134 | 1.3338 | 0.8214 | 0.5796 |
1.4061 | 3.0 | 1701 | 1.1097 | 0.7830 | 0.5731 |
1.2577 | 4.0 | 2268 | 0.9200 | 0.7091 | 0.5671 |
1.2021 | 5.0 | 2835 | 0.8400 | 0.6761 | 0.5700 |
1.1708 | 6.0 | 3402 | 0.7893 | 0.6536 | 0.5711 |
1.2189 | 7.0 | 3969 | 0.7496 | 0.6395 | 0.5640 |
1.1133 | 8.0 | 4536 | 0.7273 | 0.6287 | 0.5674 |
1.1771 | 9.0 | 5103 | 0.7126 | 0.6213 | 0.5612 |
1.1452 | 10.0 | 5670 | 0.7083 | 0.6203 | 0.5618 |
Framework versions
- Transformers 4.31.0
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3