<!-- 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-xlsr-53-espeak-cv-ft-tat-ntsema-colab
This model is a fine-tuned version of facebook/wav2vec2-xlsr-53-espeak-cv-ft on the audiofolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.2976
- Wer: 0.2834
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: 8
- seed: 42
- gradient_accumulation_steps: 8
- 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 |
|---|---|---|---|---|
| 3.5013 | 3.57 | 400 | 0.4017 | 0.4837 |
| 0.3368 | 7.14 | 800 | 0.2774 | 0.3693 |
| 0.1942 | 10.71 | 1200 | 0.3054 | 0.3386 |
| 0.1449 | 14.28 | 1600 | 0.3085 | 0.3246 |
| 0.1147 | 17.85 | 2000 | 0.3134 | 0.3037 |
| 0.0944 | 21.43 | 2400 | 0.3046 | 0.2933 |
| 0.0778 | 24.99 | 2800 | 0.3057 | 0.2927 |
| 0.0643 | 28.57 | 3200 | 0.2976 | 0.2834 |
Framework versions
- Transformers 4.24.0
- Pytorch 1.14.0.dev20221107+cu116
- Datasets 2.6.1
- Tokenizers 0.13.2
