metadata
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- common_voice
model-index:
- name: wav2vec2-large-xls-r-300m-pa-in
results: []
wav2vec2-large-xls-r-300m-pa-in
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: 1.2548
- Wer: 0.5677
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: 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_ratio: 0.12
- num_epochs: 120
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 6.4804 | 16.65 | 400 | 1.8461 | 1.0 |
| 0.474 | 33.33 | 800 | 1.1018 | 0.6624 |
| 0.1389 | 49.98 | 1200 | 1.1918 | 0.6103 |
| 0.0919 | 66.65 | 1600 | 1.1889 | 0.6058 |
| 0.0657 | 83.33 | 2000 | 1.2266 | 0.5931 |
| 0.0479 | 99.98 | 2400 | 1.2512 | 0.5902 |
| 0.0355 | 116.65 | 2800 | 1.2548 | 0.5677 |
Framework versions
- Transformers 4.15.0
- Pytorch 1.10.0+cu111
- Datasets 1.18.0
- Tokenizers 0.10.3