Downloads · 30 days
1
4% of all-time downloads
whisper-finetuned-v3_4kstep_augment
tranha1412/whisper-finetuned-v3_4kstep_augment
whisper-finetuned-v3_4kstep_augment is a automatic speech recognition model from tranha1412. Use it when you need speech turned into text. It is set up for transformers. The card lists the license as mit.
should probably proofread and complete it, then remove this comment. --
All-time downloads
24
Public
Parameters
809M
201 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors6.5 GB · 100%
At a glance
- Task
- Automatic Speech Recognition
- Library
- transformers
- License
- mit
- Model type
- whisper
- Access
- Public
- Created
- May 11, 2025
- Updated
- May 13, 2025
- SHA
- 1b140c5c
Base models
Model card
From the Hugging Face model README
<!-- 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-finetuned-v3_4kstep_augment
This model is a fine-tuned version of openai/whisper-large-v3-turbo on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.0494
- Wer: 51.4104
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: 1e-05
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 4
- optimizer: Use 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
- training_steps: 3000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 1.6773 | 0.0850 | 100 | 1.2154 | 235.3315 |
| 1.0178 | 0.1700 | 200 | 0.6814 | 283.2863 |
| 0.4466 | 0.2550 | 300 | 0.2805 | 92.8068 |
| 0.2623 | 0.3400 | 400 | 0.2547 | 79.3018 |
| 0.2228 | 0.4250 | 500 | 0.2125 | 75.8463 |
| 0.1934 | 0.5100 | 600 | 0.1790 | 72.1086 |
| 0.1903 | 0.5950 | 700 | 0.1532 | 71.7913 |
| 0.1488 | 0.6800 | 800 | 0.1339 | 70.5571 |
| 0.137 | 0.7650 | 900 | 0.1205 | 69.2525 |
| 0.1275 | 0.8500 | 1000 | 0.1146 | 66.2200 |
| 0.1158 | 0.9350 | 1100 | 0.1049 | 66.3258 |
| 0.1088 | 1.0195 | 1200 | 0.1017 | 65.8674 |
| 0.0828 | 1.1045 | 1300 | 0.0989 | 62.3413 |
| 0.0716 | 1.1895 | 1400 | 0.0902 | 61.5656 |
| 0.0774 | 1.2745 | 1500 | 0.0867 | 70.3808 |
| 0.0769 | 1.3595 | 1600 | 0.0836 | 60.1904 |
| 0.0676 | 1.4445 | 1700 | 0.0792 | 58.4626 |
| 0.0669 | 1.5295 | 1800 | 0.0778 | 59.5910 |
| 0.0624 | 1.6145 | 1900 | 0.0708 | 57.8279 |
| 0.0585 | 1.6995 | 2000 | 0.0685 | 57.4753 |
| 0.057 | 1.7845 | 2100 | 0.0657 | 55.9238 |
| 0.0587 | 1.8695 | 2200 | 0.0609 | 57.0169 |
| 0.0487 | 1.9545 | 2300 | 0.0593 | 55.3244 |
| 0.0443 | 2.0391 | 2400 | 0.0564 | 52.7151 |
| 0.0287 | 2.1241 | 2500 | 0.0564 | 60.3315 |
| 0.0274 | 2.2091 | 2600 | 0.0554 | 52.1862 |
| 0.0303 | 2.2941 | 2700 | 0.0542 | 54.1961 |
| 0.0315 | 2.3791 | 2800 | 0.0508 | 52.1157 |
| 0.0275 | 2.4641 | 2900 | 0.0493 | 52.6798 |
| 0.0191 | 2.5491 | 3000 | 0.0494 | 51.4104 |
Framework versions
- Transformers 4.51.3
- Pytorch 2.7.0+cu126
- Datasets 3.5.1
- Tokenizers 0.21.1
- Task
- Automatic Speech Recognition
- Library
- transformers
- Type
- whisper
- License
- mit
- Created
- May 11, 2025
- Updated
- May 13, 2025