metadata
license: apache-2.0
tags:
- generated_from_trainer
model-index:
- name: working
results: []
working
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: 0.2584
- Wer: 0.6024
- Cer: 0.0723
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: 3e-05
- train_batch_size: 2
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 4
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 2000
- training_steps: 70000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 2.0401 | 1.49 | 1000 | 1.3913 | 0.9911 | 0.4127 |
| 1.2772 | 2.98 | 2000 | 0.4117 | 0.7644 | 0.1036 |
| 1.0861 | 4.46 | 3000 | 0.3281 | 0.6962 | 0.0868 |
| 1.0803 | 5.95 | 4000 | 0.2970 | 0.6645 | 0.0796 |
| 1.0256 | 7.44 | 5000 | 0.2986 | 0.6556 | 0.0820 |
| 0.9536 | 8.93 | 6000 | 0.2873 | 0.6418 | 0.0767 |
| 0.9154 | 10.42 | 7000 | 0.3896 | 0.6450 | 0.0812 |
| 0.9187 | 11.9 | 8000 | 0.2946 | 0.6239 | 0.0771 |
| 0.8693 | 13.39 | 9000 | 0.2655 | 0.6093 | 0.0746 |
| 0.8335 | 14.88 | 10000 | 0.2797 | 0.6052 | 0.0764 |
| 0.8461 | 16.37 | 11000 | 0.2879 | 0.6231 | 0.0766 |
| 0.8363 | 17.86 | 12000 | 0.2616 | 0.6052 | 0.0726 |
| 0.796 | 19.35 | 13000 | 0.2656 | 0.6109 | 0.0740 |
| 0.8136 | 20.83 | 14000 | 0.2773 | 0.6255 | 0.0747 |
| 0.7319 | 22.32 | 15000 | 0.2770 | 0.6214 | 0.0748 |
| 0.7428 | 23.81 | 16000 | 0.2697 | 0.6052 | 0.0746 |
| 0.7264 | 25.3 | 17000 | 0.2716 | 0.5971 | 0.0733 |
Framework versions
- Transformers 4.17.0
- Pytorch 2.4.0
- Datasets 3.0.1
- Tokenizers 0.20.0