wav2vec2-large-xls-r-300m-slovenian
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: 0.4462
- Wer: 0.3271
Training procedure
Training Hyper-parameters
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: 20
Training Results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 4.3681 | 4.93 | 400 | 0.7067 | 0.6486 |
| 0.2311 | 9.87 | 800 | 0.5155 | 0.4341 |
| 0.0833 | 14.81 | 1200 | 0.4996 | 0.3799 |
| 0.0455 | 19.75 | 1600 | 0.4462 | 0.3271 |
Framework Versions
- Transformers 4.11.3
- Pytorch 1.10.0+cu113
- Datasets 1.18.3
- Tokenizers 0.10.3