metadata
language:
- zh-HK
license: apache-2.0
tags:
- automatic-speech-recognition
- mozilla-foundation/common_voice_8_0
- generated_from_trainer
datasets:
- common_voice
model-index:
- name: ''
results: []
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the MOZILLA-FOUNDATION/COMMON_VOICE_8_0 - ZH-HK dataset. It achieves the following results on the evaluation set:
- Loss: 40.6968
- Wer: 1.0
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: 0.0003
- train_batch_size: 8
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 1.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| No log | 0.05 | 10 | 239.0014 | 1.0 |
| No log | 0.1 | 20 | 235.8207 | 1.0 |
| No log | 0.15 | 30 | 226.9009 | 1.0 |
| No log | 0.21 | 40 | 198.0769 | 1.0 |
| No log | 0.26 | 50 | 166.6728 | 1.0 |
| No log | 0.31 | 60 | 149.1445 | 1.0 |
| No log | 0.36 | 70 | 138.4403 | 1.0 |
| No log | 0.41 | 80 | 131.7249 | 1.0 |
| No log | 0.46 | 90 | 125.5583 | 1.0 |
| No log | 0.51 | 100 | 119.7515 | 1.0 |
| No log | 0.56 | 110 | 113.7283 | 1.0 |
| No log | 0.62 | 120 | 107.2455 | 1.0 |
| No log | 0.67 | 130 | 100.2172 | 1.0 |
| No log | 0.72 | 140 | 92.5585 | 1.0 |
| No log | 0.77 | 150 | 84.2573 | 1.0 |
| No log | 0.82 | 160 | 75.2953 | 1.0 |
| No log | 0.87 | 170 | 65.6953 | 1.0 |
| No log | 0.92 | 180 | 55.7544 | 1.0 |
| No log | 0.97 | 190 | 45.7297 | 1.0 |
Framework versions
- Transformers 4.17.0.dev0
- Pytorch 1.10.2+cu102
- Datasets 1.18.3
- Tokenizers 0.11.0