|
--- |
|
license: apache-2.0 |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: test1000v2 |
|
results: [] |
|
--- |
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
# test1000v2 |
|
|
|
This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on the None dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.7873 |
|
- Wer: 0.6162 |
|
|
|
## 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: 0.001 |
|
- train_batch_size: 16 |
|
- eval_batch_size: 8 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 2 |
|
- 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: 500 |
|
- num_epochs: 20 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Wer | |
|
|:-------------:|:-----:|:----:|:---------------:|:------:| |
|
| 7.7913 | 3.22 | 100 | 3.3481 | 1.0 | |
|
| 3.3831 | 6.44 | 200 | 3.3229 | 1.0 | |
|
| 3.3778 | 9.67 | 300 | 3.3211 | 1.0 | |
|
| 3.3671 | 12.89 | 400 | 3.2973 | 1.0 | |
|
| 3.3528 | 16.13 | 500 | 3.1349 | 1.0 | |
|
| 1.8611 | 19.35 | 600 | 0.7873 | 0.6162 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.22.2 |
|
- Pytorch 1.10.0+cu102 |
|
- Datasets 1.4.1 |
|
- Tokenizers 0.12.1 |
|
|
