<!-- 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. -->
wav2vec2-owndata
This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.2515
- Wer: 0.3212
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: 0.0003
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 5
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
4.262 | 0.36 | 100 | 3.4482 | 0.9832 |
3.0032 | 0.72 | 200 | 2.9441 | 0.9832 |
2.9141 | 1.08 | 300 | 2.9393 | 0.9832 |
2.8585 | 1.44 | 400 | 2.8848 | 0.9627 |
2.2837 | 1.8 | 500 | 2.1732 | 1.0111 |
0.9834 | 2.16 | 600 | 0.8765 | 0.7345 |
0.7288 | 2.52 | 700 | 0.5741 | 0.5641 |
0.5521 | 2.88 | 800 | 0.3937 | 0.4467 |
0.3751 | 3.24 | 900 | 0.3484 | 0.4112 |
0.3733 | 3.6 | 1000 | 0.2964 | 0.3912 |
0.2443 | 3.96 | 1100 | 0.2673 | 0.3446 |
0.2667 | 4.32 | 1200 | 0.2657 | 0.3357 |
0.2237 | 4.68 | 1300 | 0.2515 | 0.3212 |
Framework versions
- Transformers 4.11.3
- Pytorch 1.10.1
- Datasets 2.9.0
- Tokenizers 0.10.3