metadata
language:
- et
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 - ET dataset. It achieves the following results on the evaluation set:
- Loss: 0.4927
- Wer: 0.3536
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: 72
- eval_batch_size: 72
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 144
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 500
- num_epochs: 100.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.3442 | 12.5 | 500 | 0.3825 | 0.4763 |
| 0.1934 | 25.0 | 1000 | 0.4236 | 0.4414 |
| 0.149 | 37.5 | 1500 | 0.4503 | 0.4190 |
| 0.1253 | 50.0 | 2000 | 0.4674 | 0.4143 |
| 0.0966 | 62.5 | 2500 | 0.4847 | 0.3925 |
| 0.0741 | 75.0 | 3000 | 0.4745 | 0.3704 |
| 0.0608 | 87.5 | 3500 | 0.4807 | 0.3568 |
| 0.0541 | 100.0 | 4000 | 0.4927 | 0.3536 |
Framework versions
- Transformers 4.16.0.dev0
- Pytorch 1.10.1+cu102
- Datasets 1.18.1.dev0
- Tokenizers 0.11.0