File size: 3,380 Bytes
3cc2f5b 98e3dca 3cc2f5b 98e3dca 3cc2f5b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 |
---
library_name: transformers
language:
- ar
license: apache-2.0
base_model: openai/whisper-base
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: Whisper base AR - BA
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# Whisper base AR - BA
This model is a fine-tuned version of [openai/whisper-base](https://huggingface.co/openai/whisper-base) on the quran-ayat-speech-to-text dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0036
- Wer: 0.0518
- Cer: 0.0212
## 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: 0.0001
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 25
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|:-------------:|:-------:|:----:|:---------------:|:------:|:------:|
| 0.0051 | 1.0 | 282 | 0.0036 | 0.0472 | 0.0188 |
| 0.0033 | 2.0 | 564 | 0.0042 | 0.0547 | 0.0200 |
| 0.0018 | 3.0 | 846 | 0.0050 | 0.0599 | 0.0217 |
| 0.0012 | 4.0 | 1128 | 0.0054 | 0.0536 | 0.0195 |
| 0.0007 | 5.0 | 1410 | 0.0059 | 0.0599 | 0.0203 |
| 0.0006 | 6.0 | 1692 | 0.0069 | 0.0632 | 0.0238 |
| 0.0006 | 7.0 | 1974 | 0.0076 | 0.0673 | 0.0239 |
| 0.0004 | 8.0 | 2256 | 0.0076 | 0.0644 | 0.0235 |
| 0.0004 | 9.0 | 2538 | 0.0087 | 0.0603 | 0.0225 |
| 0.0003 | 10.0 | 2820 | 0.0076 | 0.0601 | 0.0220 |
| 0.0003 | 11.0 | 3102 | 0.0091 | 0.0615 | 0.0213 |
| 0.0002 | 12.0 | 3384 | 0.0094 | 0.0572 | 0.0211 |
| 0.0002 | 13.0 | 3666 | 0.0100 | 0.0686 | 0.0284 |
| 0.0001 | 14.0 | 3948 | 0.0102 | 0.0608 | 0.0221 |
| 0.0001 | 15.0 | 4230 | 0.0115 | 0.0552 | 0.0196 |
| 0.0001 | 16.0 | 4512 | 0.0119 | 0.0563 | 0.0208 |
| 0.0001 | 17.0 | 4794 | 0.0121 | 0.0614 | 0.0228 |
| 0.0 | 18.0 | 5076 | 0.0132 | 0.0559 | 0.0216 |
| 0.0 | 19.0 | 5358 | 0.0135 | 0.0568 | 0.0211 |
| 0.0 | 20.0 | 5640 | 0.0139 | 0.0559 | 0.0213 |
| 0.0 | 21.0 | 5922 | 0.0137 | 0.0552 | 0.0206 |
| 0.0 | 22.0 | 6204 | 0.0136 | 0.0541 | 0.0204 |
| 0.0 | 23.0 | 6486 | 0.0136 | 0.0538 | 0.0200 |
| 0.0 | 24.0 | 6768 | 0.0132 | 0.0608 | 0.0244 |
| 0.0 | 24.9130 | 7025 | 0.0136 | 0.0539 | 0.0201 |
### Framework versions
- Transformers 4.51.1
- Pytorch 2.5.1+cu124
- Datasets 3.5.0
- Tokenizers 0.21.0
|