<!-- 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-xls-r-bengali_v1
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice dataset. It achieves the following results on the evaluation set:
- Loss: 3.2973
- Wer: 1.0
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: 3e-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: 6
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
8.7896 | 0.8 | 500 | 3.8455 | 1.0 |
3.3871 | 1.6 | 1000 | 3.2862 | 1.0 |
3.3302 | 2.4 | 1500 | 3.3086 | 1.0 |
3.3259 | 3.2 | 2000 | 3.2973 | 1.0 |
3.325 | 4.0 | 2500 | 3.2973 | 1.0 |
3.3178 | 4.8 | 3000 | 3.2973 | 1.0 |
3.3226 | 5.6 | 3500 | 3.2973 | 1.0 |
Framework versions
- Transformers 4.18.0
- Pytorch 1.11.0
- Datasets 2.1.0
- Tokenizers 0.12.1