metadata
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- common_voice_8_0
model-index:
- name: xlsr_ur_training
results: []
xlsr_ur_training
This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the common_voice_8_0 dataset. It achieves the following results on the evaluation set:
- Loss: 1.2610
- Wer: 0.7325
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: 100
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 14.5044 | 1.69 | 100 | 3.9173 | 1.0 |
| 3.3645 | 3.39 | 200 | 3.2475 | 1.0 |
| 3.2318 | 5.08 | 300 | 3.2143 | 1.0 |
| 3.1887 | 6.78 | 400 | 3.1672 | 1.0 |
| 3.1233 | 8.47 | 500 | 3.0927 | 1.0 |
| 3.0938 | 10.17 | 600 | 3.0836 | 0.9970 |
| 3.0706 | 11.86 | 700 | 3.0319 | 0.9996 |
| 2.9622 | 13.56 | 800 | 2.7973 | 0.9985 |
| 2.6267 | 15.25 | 900 | 2.2553 | 0.9974 |
| 1.9748 | 16.95 | 1000 | 1.6858 | 0.9170 |
| 1.4739 | 18.64 | 1100 | 1.4620 | 0.8125 |
| 1.2102 | 20.34 | 1200 | 1.3890 | 0.7779 |
| 1.036 | 22.03 | 1300 | 1.3347 | 0.7672 |
| 0.9462 | 23.73 | 1400 | 1.2970 | 0.7476 |
| 0.8725 | 25.42 | 1500 | 1.2792 | 0.7461 |
| 0.8374 | 27.12 | 1600 | 1.2574 | 0.7384 |
| 0.7976 | 28.81 | 1700 | 1.2610 | 0.7325 |
Framework versions
- Transformers 4.21.0
- Pytorch 1.11.0+cu113
- Datasets 2.4.0
- Tokenizers 0.12.1