Downloads · 30 days
3
2% of all-time downloads
whisper-finetuned-v3_30e_augment_new_168_v2
tranha/whisper-finetuned-v3_30e_augment_new_168_v2
whisper-finetuned-v3_30e_augment_new_168_v2 is a automatic speech recognition model from tranha. 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
138
Public
Parameters
809M
97.1 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
- Sep 16, 2025
- Updated
- Sep 17, 2025
- SHA
- 8f399ac9
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_30e_augment_new_168_v2
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: 0.1205
- Wer: 52.0677
- Cer: 28.3168
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: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 0.0967 | 0.9997 | 1964 | 0.0914 | 63.7531 | 31.8065 |
| 0.0579 | 2.0 | 3929 | 0.0812 | 60.0877 | 29.8874 |
| 0.0312 | 2.9997 | 5893 | 0.0763 | 60.1190 | 30.3982 |
| 0.033 | 4.0 | 7858 | 0.0802 | 57.7694 | 29.7525 |
| 0.0188 | 4.9997 | 9822 | 0.0815 | 60.0877 | 30.4146 |
| 0.0192 | 6.0 | 11787 | 0.0856 | 56.6103 | 29.8345 |
| 0.0112 | 6.9997 | 13751 | 0.0814 | 55.7644 | 29.6521 |
| 0.0117 | 8.0 | 15716 | 0.0890 | 56.1090 | 29.0684 |
| 0.0067 | 8.9997 | 17680 | 0.0900 | 55.9837 | 29.1760 |
| 0.0072 | 10.0 | 19645 | 0.0990 | 56.0464 | 29.3749 |
| 0.0053 | 10.9997 | 21609 | 0.0968 | 54.9499 | 29.1195 |
| 0.0044 | 12.0 | 23574 | 0.1001 | 55.2005 | 28.6123 |
| 0.003 | 12.9997 | 25538 | 0.0996 | 56.7356 | 29.1596 |
| 0.0035 | 14.0 | 27503 | 0.1014 | 55.0439 | 29.0830 |
| 0.0027 | 14.9997 | 29467 | 0.1040 | 54.6679 | 28.3150 |
| 0.0032 | 16.0 | 31432 | 0.0994 | 54.5426 | 29.0538 |
| 0.0043 | 16.9997 | 33396 | 0.1059 | 53.9160 | 28.9535 |
| 0.0017 | 18.0 | 35361 | 0.1087 | 54.1980 | 28.8385 |
| 0.0018 | 18.9997 | 37325 | 0.1103 | 54.2607 | 28.6543 |
| 0.0007 | 20.0 | 39290 | 0.1105 | 54.5739 | 28.8896 |
| 0.0006 | 20.9997 | 41254 | 0.1121 | 54.0100 | 28.7546 |
| 0.0007 | 22.0 | 43219 | 0.1081 | 53.0075 | 28.8185 |
| 0.001 | 22.9997 | 45183 | 0.1075 | 53.9474 | 28.7948 |
| 0.0002 | 24.0 | 47148 | 0.1150 | 52.9762 | 28.5886 |
| 0.0002 | 24.9997 | 49112 | 0.1164 | 52.6316 | 28.6397 |
| 0.0002 | 26.0 | 51077 | 0.1175 | 52.8195 | 28.7875 |
| 0.0003 | 26.9997 | 53041 | 0.1169 | 52.5063 | 28.6142 |
| 0.0 | 28.0 | 55006 | 0.1201 | 52.2556 | 28.4153 |
| 0.0 | 28.9997 | 56970 | 0.1209 | 52.1303 | 28.3934 |
| 0.0 | 29.9924 | 58920 | 0.1205 | 52.0677 | 28.3168 |
Framework versions
- Transformers 4.44.2
- Pytorch 2.8.0+cu128
- Datasets 4.1.0
- Tokenizers 0.19.1
- Task
- Automatic Speech Recognition
- Library
- transformers
- Type
- whisper
- License
- mit
- Created
- Sep 16, 2025
- Updated
- Sep 17, 2025