metadata
license: apache-2.0
tags:
- generated_from_trainer
model-index:
- name: wav2vec2-large-xls-r-300m-korean-s4
results: []
wav2vec2-large-xls-r-300m-korean-s4
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0378
- Cer: 0.0048
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: 0.0001
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| No log | 0.37 | 300 | 4.6810 | 1.0 |
| 5.541 | 0.74 | 600 | 3.2272 | 1.0 |
| 5.541 | 1.12 | 900 | 2.9931 | 0.9389 |
| 2.8308 | 1.49 | 1200 | 0.3785 | 0.0922 |
| 0.4651 | 1.86 | 1500 | 0.1628 | 0.0385 |
| 0.4651 | 2.23 | 1800 | 0.0769 | 0.0139 |
| 0.1628 | 2.6 | 2100 | 0.0475 | 0.0069 |
| 0.1628 | 2.97 | 2400 | 0.0378 | 0.0048 |
Framework versions
- Transformers 4.17.0
- Pytorch 1.10.0+cu113
- Datasets 2.6.1
- Tokenizers 0.13.1