metadata
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- common_voice
model-index:
- name: wav2vec2-large-xls-r-300m-hindi
results: []
wav2vec2-large-xls-r-300m-hindi
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: 5.8345
- Wer: 1.0592
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_steps: 500
- num_epochs: 400
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 6.0837 | 57.14 | 400 | 3.3819 | 1.0081 |
| 0.2254 | 114.29 | 800 | 4.4935 | 1.0255 |
| 0.061 | 171.43 | 1200 | 5.1464 | 1.0870 |
| 0.0282 | 228.57 | 1600 | 5.6430 | 1.0789 |
| 0.0144 | 285.71 | 2000 | 5.7916 | 1.0940 |
| 0.0077 | 342.86 | 2400 | 5.8701 | 1.0580 |
| 0.0041 | 400.0 | 2800 | 5.8345 | 1.0592 |
Framework versions
- Transformers 4.11.3
- Pytorch 1.10.0+cu113
- Datasets 1.13.3
- Tokenizers 0.10.3