wav2vec2-large-xls-r-1b-Indonesian
This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on the common_voice dataset.
It achieves the following results on the evaluation set:
- Loss: 0.9550
- Wer: 0.4551
- Cer: 0.1643
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0003
- train_batch_size: 64
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 400
- num_epochs: 50
- mixed_precision_training: Native AMP
Training results
Training Loss |
Epoch |
Step |
Validation Loss |
Wer |
Cer |
3.663 |
7.69 |
200 |
0.7898 |
0.6039 |
0.1848 |
0.7424 |
15.38 |
400 |
1.0215 |
0.5615 |
0.1924 |
0.4494 |
23.08 |
600 |
1.0901 |
0.5249 |
0.1932 |
0.5075 |
30.77 |
800 |
1.1013 |
0.5079 |
0.1935 |
0.4671 |
38.46 |
1000 |
1.1034 |
0.4916 |
0.1827 |
0.1928 |
46.15 |
1200 |
0.9550 |
0.4551 |
0.1643 |
Framework versions
- Transformers 4.17.0.dev0
- Pytorch 1.10.2+cu102
- Datasets 1.18.2.dev0
- Tokenizers 0.11.0