metadata
license: mit
tags:
- generated_from_trainer
base_model: facebook/w2v-bert-2.0
datasets:
- common_voice_17_0
metrics:
- wer
model-index:
- name: w2v2-bert-urdu
results:
- task:
type: automatic-speech-recognition
name: Automatic Speech Recognition
dataset:
name: common_voice_17_0
type: common_voice_17_0
config: ur
split: test[:100]
args: ur
metrics:
- type: wer
value: 0.6273224043715847
name: Wer
w2v2-bert-urdu
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: 1.1498
- Wer: 0.6273
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: 2
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3.5968 | 0.1695 | 50 | 3.1737 | 1.0 |
| 3.1414 | 0.3390 | 100 | 2.9666 | 1.0 |
| 2.3694 | 0.5085 | 150 | 1.0788 | 0.6525 |
| 0.7692 | 0.6780 | 200 | 0.5647 | 0.4186 |
| 0.5488 | 0.8475 | 250 | 0.4491 | 0.3486 |
| 0.5568 | 1.0169 | 300 | 0.5883 | 0.7388 |
| 0.7925 | 1.1864 | 350 | 1.0338 | 0.7967 |
| 1.4791 | 1.3559 | 400 | 1.1474 | 0.6251 |
| 1.2758 | 1.5254 | 450 | 1.1359 | 0.6251 |
| 1.2763 | 1.6949 | 500 | 1.1497 | 0.6273 |
| 1.2789 | 1.8644 | 550 | 1.1498 | 0.6273 |
Framework versions
- Transformers 4.40.2
- Pytorch 2.2.1+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1