Downloads · 30 days
9
7% of all-time downloads
ft_0122_korean
yoon1000/ft_0122_korean
ft_0122_korean is a automatic speech recognition model from yoon1000. Use it when you need speech turned into text. It is set up for transformers. The card lists the license as apache-2.0.
should probably proofread and complete it, then remove this comment. --
All-time downloads
129
Public
Parameters
317M
63.4 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors1.3 GB · 100%
At a glance
- Task
- Automatic Speech Recognition
- Library
- transformers
- License
- apache-2.0
- Model type
- wav2vec2
- Access
- Public
- Created
- Jan 22, 2024
- Updated
- Jan 22, 2024
- SHA
- 95fe3701
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. -->
ft_0122_korean
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.3680
- Cer: 0.3009
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: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- num_epochs: 4
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 46.2918 | 0.08 | 100 | 50.8931 | 1.0 |
| 15.0805 | 0.16 | 200 | 8.0154 | 1.0 |
| 5.0807 | 0.24 | 300 | 5.0733 | 1.0 |
| 4.7729 | 0.32 | 400 | 5.0188 | 1.0 |
| 4.7749 | 0.4 | 500 | 4.9321 | 1.0 |
| 4.7657 | 0.48 | 600 | 4.8467 | 1.0 |
| 4.7006 | 0.56 | 700 | 4.8097 | 1.0 |
| 4.6775 | 0.64 | 800 | 4.8150 | 1.0 |
| 4.6484 | 0.72 | 900 | 4.7064 | 1.0 |
| 4.6009 | 0.8 | 1000 | 4.7248 | 1.0 |
| 4.5753 | 0.88 | 1100 | 4.6097 | 0.9991 |
| 4.533 | 0.96 | 1200 | 4.5760 | 1.0 |
| 4.4915 | 1.04 | 1300 | 4.5454 | 0.9845 |
| 4.436 | 1.12 | 1400 | 4.4345 | 0.9842 |
| 4.3636 | 1.2 | 1500 | 4.3722 | 0.9363 |
| 4.3186 | 1.28 | 1600 | 4.2747 | 0.9566 |
| 4.2326 | 1.36 | 1700 | 4.0707 | 0.9416 |
| 4.0018 | 1.44 | 1800 | 3.6969 | 0.7459 |
| 3.6566 | 1.52 | 1900 | 3.2351 | 0.5922 |
| 3.3079 | 1.6 | 2000 | 2.9201 | 0.5560 |
| 3.1056 | 1.68 | 2100 | 2.6958 | 0.5182 |
| 2.9577 | 1.76 | 2200 | 2.5580 | 0.4942 |
| 2.7556 | 1.84 | 2300 | 2.3948 | 0.4671 |
| 2.6682 | 1.92 | 2400 | 2.2875 | 0.4521 |
| 2.5461 | 2.0 | 2500 | 2.1791 | 0.4386 |
| 2.4496 | 2.08 | 2600 | 2.0957 | 0.4230 |
| 2.3548 | 2.16 | 2700 | 2.0157 | 0.4132 |
| 2.2973 | 2.24 | 2800 | 1.9433 | 0.4010 |
| 2.2631 | 2.32 | 2900 | 1.8848 | 0.3926 |
| 2.1837 | 2.4 | 3000 | 1.8301 | 0.3827 |
| 2.1427 | 2.48 | 3100 | 1.7783 | 0.3738 |
| 2.0955 | 2.56 | 3200 | 1.7377 | 0.3678 |
| 2.0393 | 2.64 | 3300 | 1.6954 | 0.3616 |
| 2.0162 | 2.72 | 3400 | 1.6446 | 0.3576 |
| 1.9577 | 2.8 | 3500 | 1.6098 | 0.3507 |
| 1.9501 | 2.88 | 3600 | 1.5772 | 0.3427 |
| 1.8684 | 2.96 | 3700 | 1.5502 | 0.3361 |
| 1.8625 | 3.04 | 3800 | 1.5222 | 0.3290 |
| 1.8252 | 3.12 | 3900 | 1.5000 | 0.3265 |
| 1.8299 | 3.2 | 4000 | 1.4743 | 0.3201 |
| 1.758 | 3.28 | 4100 | 1.4529 | 0.3156 |
| 1.7731 | 3.36 | 4200 | 1.4388 | 0.3137 |
| 1.7579 | 3.44 | 4300 | 1.4195 | 0.3076 |
| 1.7205 | 3.52 | 4400 | 1.4120 | 0.3084 |
| 1.6997 | 3.6 | 4500 | 1.3986 | 0.3055 |
| 1.7536 | 3.68 | 4600 | 1.3830 | 0.3039 |
| 1.6927 | 3.76 | 4700 | 1.3755 | 0.3025 |
| 1.6604 | 3.84 | 4800 | 1.3701 | 0.3010 |
| 1.6599 | 3.92 | 4900 | 1.3686 | 0.3002 |
| 1.6567 | 4.0 | 5000 | 1.3680 | 0.3009 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.1+cu121
- Datasets 2.13.0
- Tokenizers 0.15.0
- Task
- Automatic Speech Recognition
- Library
- transformers
- Type
- wav2vec2
- License
- apache-2.0
- Created
- Jan 22, 2024
- Updated
- Jan 22, 2024