metadata
license: apache-2.0
base_model: facebook/wav2vec2-lv-60-espeak-cv-ft
tags:
- generated_from_trainer
datasets:
- voxpopuli
metrics:
- wer
model-index:
- name: cs2fi_wav2vec2-large-xls-r-300m-czech-colab
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: voxpopuli
type: voxpopuli
config: fi
split: test
args: fi
metrics:
- name: Wer
type: wer
value: 1.0754716981132075
cs2fi_wav2vec2-large-xls-r-300m-czech-colab
This model is a fine-tuned version of facebook/wav2vec2-lv-60-espeak-cv-ft on the voxpopuli dataset. It achieves the following results on the evaluation set:
- Loss: 485.7458
- Wer: 1.0755
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: 8
- 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: 50
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3007.5297 | 3.51 | 100 | 523.8923 | 0.9706 |
| 353.1859 | 7.02 | 200 | 270.8087 | 0.9665 |
| 207.1084 | 10.53 | 300 | 215.3542 | 0.9350 |
| 186.3063 | 14.04 | 400 | 210.1422 | 0.9119 |
| 171.7259 | 17.54 | 500 | 291.5182 | 1.0629 |
| 142.6091 | 21.05 | 600 | 219.2806 | 0.9602 |
| 118.6791 | 24.56 | 700 | 312.2755 | 1.1132 |
| 96.153 | 28.07 | 800 | 320.7119 | 1.0545 |
| 82.968 | 31.58 | 900 | 357.5117 | 1.0629 |
| 71.2426 | 35.09 | 1000 | 421.3889 | 0.9916 |
| 58.8083 | 38.6 | 1100 | 433.8375 | 1.1048 |
| 54.5225 | 42.11 | 1200 | 482.5988 | 1.0566 |
| 48.12 | 45.61 | 1300 | 479.3787 | 1.0860 |
| 43.3324 | 49.12 | 1400 | 485.7458 | 1.0755 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu118
- Datasets 2.15.0
- Tokenizers 0.15.0