metadata
license: mit
base_model: facebook/w2v-bert-2.0
tags:
- generated_from_trainer
datasets:
- common_voice_17_0
metrics:
- wer
model-index:
- name: w2v-bert-2.0-sv
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: common_voice_17_0
type: common_voice_17_0
config: sv-SE
split: test
args: sv-SE
metrics:
- name: Wer
type: wer
value: 0.10046931592103249
language:
- sv
w2v-bert-2.0-sv
This model is a fine-tuned version of facebook/w2v-bert-2.0 on the common_voice_17_0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.1962
- Wer: 0.1005
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: 5e-05
- 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: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 2.075 | 0.7407 | 300 | 0.3441 | 0.3057 |
| 0.2837 | 1.4815 | 600 | 0.2995 | 0.2274 |
| 0.2081 | 2.2222 | 900 | 0.2443 | 0.1768 |
| 0.1579 | 2.9630 | 1200 | 0.2143 | 0.1493 |
| 0.1248 | 3.7037 | 1500 | 0.2165 | 0.1504 |
| 0.0934 | 4.4444 | 1800 | 0.1869 | 0.1284 |
| 0.0719 | 5.1852 | 2100 | 0.2072 | 0.1216 |
| 0.0573 | 5.9259 | 2400 | 0.1949 | 0.1195 |
| 0.0436 | 6.6667 | 2700 | 0.2025 | 0.1142 |
| 0.0317 | 7.4074 | 3000 | 0.2003 | 0.1097 |
| 0.0256 | 8.1481 | 3300 | 0.1942 | 0.1060 |
| 0.0169 | 8.8889 | 3600 | 0.1851 | 0.1030 |
| 0.0121 | 9.6296 | 3900 | 0.1962 | 0.1005 |
Framework versions
- Transformers 4.41.1
- Pytorch 2.1.1+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1