metadata
license: apache-2.0
tags:
- generated_from_trainer
model-index:
- name: urdumodel
results: []
urdumodel
This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.4889
- Wer: 0.3744
- Cer: 0.1488
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: 24
- eval_batch_size: 24
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 96
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 15
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 2.8998 | 1.0 | 508 | 1.0261 | 0.7004 | 0.2832 |
| 0.7426 | 2.0 | 1016 | 0.6532 | 0.5236 | 0.2026 |
| 0.5694 | 3.0 | 1524 | 0.5495 | 0.4611 | 0.1799 |
| 0.4966 | 4.0 | 2032 | 0.5361 | 0.4350 | 0.1729 |
| 0.4555 | 5.0 | 2540 | 0.5335 | 0.4266 | 0.1684 |
| 0.4203 | 6.0 | 3048 | 0.5040 | 0.4107 | 0.1641 |
| 0.3951 | 7.0 | 3556 | 0.5213 | 0.4037 | 0.1579 |
| 0.3675 | 8.0 | 4064 | 0.4949 | 0.3973 | 0.1563 |
| 0.3555 | 9.0 | 4572 | 0.4968 | 0.3978 | 0.1581 |
| 0.3408 | 10.0 | 5080 | 0.4827 | 0.3925 | 0.1561 |
| 0.3286 | 11.0 | 5588 | 0.5011 | 0.3858 | 0.1524 |
| 0.3156 | 12.0 | 6096 | 0.4871 | 0.3833 | 0.1524 |
| 0.3047 | 13.0 | 6604 | 0.4835 | 0.3774 | 0.1499 |
| 0.2929 | 14.0 | 7112 | 0.4844 | 0.3751 | 0.1489 |
| 0.2912 | 15.0 | 7620 | 0.4889 | 0.3744 | 0.1488 |
Framework versions
- Transformers 4.21.1
- Pytorch 1.12.0
- Datasets 2.4.0
- Tokenizers 0.12.1