|
--- |
|
library_name: peft |
|
language: |
|
- ko |
|
license: mit |
|
base_model: openai/whisper-large-v3-turbo |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: Whisper Small ko |
|
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 Small ko |
|
|
|
This model is a fine-tuned version of [openai/whisper-large-v3-turbo](https://huggingface.co/openai/whisper-large-v3-turbo) on the custom dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.2388 |
|
|
|
## 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: 64 |
|
- eval_batch_size: 256 |
|
- seed: 42 |
|
- 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: 50 |
|
- training_steps: 500 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:------:|:----:|:---------------:| |
|
| 0.8865 | 0.0901 | 10 | 1.5498 | |
|
| 0.8668 | 0.1802 | 20 | 1.4739 | |
|
| 0.7179 | 0.2703 | 30 | 1.2411 | |
|
| 0.3916 | 0.3604 | 40 | 0.8666 | |
|
| 0.219 | 0.4505 | 50 | 0.7558 | |
|
| 0.1545 | 0.5405 | 60 | 0.6752 | |
|
| 0.1278 | 0.6306 | 70 | 0.5819 | |
|
| 0.0983 | 0.7207 | 80 | 0.5394 | |
|
| 0.0908 | 0.8108 | 90 | 0.5013 | |
|
| 0.0718 | 0.9009 | 100 | 0.4740 | |
|
| 0.0773 | 0.9910 | 110 | 0.4579 | |
|
| 0.0665 | 1.0811 | 120 | 0.4430 | |
|
| 0.0608 | 1.1712 | 130 | 0.4284 | |
|
| 0.0612 | 1.2613 | 140 | 0.4136 | |
|
| 0.0605 | 1.3514 | 150 | 0.4104 | |
|
| 0.0632 | 1.4414 | 160 | 0.3857 | |
|
| 0.0534 | 1.5315 | 170 | 0.3678 | |
|
| 0.0527 | 1.6216 | 180 | 0.3584 | |
|
| 0.0516 | 1.7117 | 190 | 0.3458 | |
|
| 0.0467 | 1.8018 | 200 | 0.3373 | |
|
| 0.0526 | 1.8919 | 210 | 0.3299 | |
|
| 0.0363 | 1.9820 | 220 | 0.3280 | |
|
| 0.0468 | 2.0721 | 230 | 0.3202 | |
|
| 0.0473 | 2.1622 | 240 | 0.3152 | |
|
| 0.0394 | 2.2523 | 250 | 0.3065 | |
|
| 0.0356 | 2.3423 | 260 | 0.3009 | |
|
| 0.042 | 2.4324 | 270 | 0.2934 | |
|
| 0.042 | 2.5225 | 280 | 0.2911 | |
|
| 0.0314 | 2.6126 | 290 | 0.2899 | |
|
| 0.0397 | 2.7027 | 300 | 0.2817 | |
|
| 0.0377 | 2.7928 | 310 | 0.2743 | |
|
| 0.0412 | 2.8829 | 320 | 0.2695 | |
|
| 0.0362 | 2.9730 | 330 | 0.2649 | |
|
| 0.0321 | 3.0631 | 340 | 0.2589 | |
|
| 0.0406 | 3.1532 | 350 | 0.2572 | |
|
| 0.028 | 3.2432 | 360 | 0.2568 | |
|
| 0.0345 | 3.3333 | 370 | 0.2568 | |
|
| 0.0346 | 3.4234 | 380 | 0.2544 | |
|
| 0.0391 | 3.5135 | 390 | 0.2513 | |
|
| 0.0362 | 3.6036 | 400 | 0.2468 | |
|
| 0.0304 | 3.6937 | 410 | 0.2446 | |
|
| 0.032 | 3.7838 | 420 | 0.2428 | |
|
| 0.0307 | 3.8739 | 430 | 0.2422 | |
|
| 0.0301 | 3.9640 | 440 | 0.2414 | |
|
| 0.0315 | 4.0541 | 450 | 0.2396 | |
|
| 0.0336 | 4.1441 | 460 | 0.2396 | |
|
| 0.024 | 4.2342 | 470 | 0.2396 | |
|
| 0.0286 | 4.3243 | 480 | 0.2391 | |
|
| 0.0289 | 4.4144 | 490 | 0.2389 | |
|
| 0.0323 | 4.5045 | 500 | 0.2388 | |
|
|
|
|
|
### Framework versions |
|
|
|
- PEFT 0.14.0 |
|
- Transformers 4.47.1 |
|
- Pytorch 2.5.1+cu124 |
|
- Datasets 3.2.0 |
|
- Tokenizers 0.21.0 |
