<!-- 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-large-xlsr-129-turkish-colab
This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-129 on the common_voice dataset. It achieves the following results on the evaluation set:
- Loss: 0.3149
- Wer: 0.4748
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-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- 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
- lr_scheduler_warmup_steps: 500
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
6.4837 | 3.67 | 400 | 3.2526 | 1.0 |
3.0896 | 7.34 | 800 | 2.8037 | 1.0 |
1.5604 | 11.01 | 1200 | 0.5688 | 0.6613 |
0.6511 | 14.68 | 1600 | 0.3998 | 0.5580 |
0.4798 | 18.35 | 2000 | 0.3505 | 0.5118 |
0.4047 | 22.02 | 2400 | 0.3273 | 0.4858 |
0.3519 | 25.69 | 2800 | 0.3224 | 0.4796 |
0.343 | 29.36 | 3200 | 0.3149 | 0.4748 |
Framework versions
- Transformers 4.11.3
- Pytorch 1.10.0+cu102
- Datasets 1.13.3
- Tokenizers 0.10.3