metadata
language:
- fa
tags:
- automatic-speech-recognition
- common_voice
- generated_from_trainer
datasets:
- common_voice
model-index:
- name: common6
results: []
common6
This model is a fine-tuned version of common6/checkpoint-3500 on the COMMON_VOICE - FA dataset. It achieves the following results on the evaluation set:
- Loss: 0.3502
- Wer: 0.3550
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: 6e-05
- train_batch_size: 32
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 256
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- num_epochs: 100.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 1.0344 | 10.0 | 500 | 0.4043 | 0.4511 |
| 0.9651 | 20.0 | 1000 | 0.3793 | 0.4159 |
| 0.9125 | 30.0 | 1500 | 0.3756 | 0.4046 |
| 0.8831 | 40.0 | 2000 | 0.3650 | 0.3876 |
| 0.8399 | 50.0 | 2500 | 0.3605 | 0.3772 |
| 0.819 | 60.0 | 3000 | 0.3622 | 0.3714 |
| 0.8029 | 70.0 | 3500 | 0.3561 | 0.3664 |
| 0.8045 | 80.0 | 4000 | 0.3607 | 0.3621 |
| 0.7799 | 90.0 | 4500 | 0.3501 | 0.3561 |
| 0.7769 | 100.0 | 5000 | 0.3502 | 0.3550 |
Framework versions
- Transformers 4.17.0.dev0
- Pytorch 1.10.2
- Datasets 1.18.3.dev0
- Tokenizers 0.10.3