metadata
language:
- ug
license: apache-2.0
tags:
- automatic-speech-recognition
- mozilla-foundation/common_voice_7_0
- generated_from_trainer
datasets:
- common_voice
model-index:
- name: uyghur
results: []
uyghur
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the MOZILLA-FOUNDATION/COMMON_VOICE_7_0 - UG dataset. It achieves the following results on the evaluation set:
- Loss: 0.2266
- Wer: 0.3655
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: 7.5e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- 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: 2000
- num_epochs: 50.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3.6863 | 2.73 | 500 | 3.5362 | 1.0 |
| 3.1409 | 5.46 | 1000 | 3.1328 | 1.0 |
| 1.8979 | 8.2 | 1500 | 0.9715 | 0.8864 |
| 1.4859 | 10.93 | 2000 | 0.5234 | 0.7063 |
| 1.3388 | 13.66 | 2500 | 0.4094 | 0.6203 |
| 1.2531 | 16.39 | 3000 | 0.3596 | 0.5185 |
| 1.1992 | 19.13 | 3500 | 0.3221 | 0.4854 |
| 1.1589 | 21.86 | 4000 | 0.3040 | 0.4610 |
| 1.1345 | 24.59 | 4500 | 0.2907 | 0.4450 |
| 1.086 | 27.32 | 5000 | 0.2744 | 0.4299 |
| 1.0697 | 30.05 | 5500 | 0.2617 | 0.4148 |
| 1.0518 | 32.79 | 6000 | 0.2563 | 0.4033 |
| 1.0101 | 35.52 | 6500 | 0.2480 | 0.3934 |
| 1.0013 | 38.25 | 7000 | 0.2412 | 0.3855 |
| 0.9845 | 40.98 | 7500 | 0.2397 | 0.3771 |
| 0.9739 | 43.72 | 8000 | 0.2303 | 0.3726 |
| 0.9636 | 46.45 | 8500 | 0.2285 | 0.3687 |
| 0.9466 | 49.18 | 9000 | 0.2261 | 0.3648 |
Framework versions
- Transformers 4.16.0.dev0
- Pytorch 1.10.1+cu102
- Datasets 1.18.2.dev0
- Tokenizers 0.11.0