Skip to content
ModelListedExternal
Download

Korean Kws2

This model is a fine-tuned version of Kkonjeong/wav2vec2-base-korean on the None dataset. It achieves the following results on the evaluation set: Loss: 0.2259 Accuracy: 1.0

Sourcehuggingface.co/devtaco/korean_kws2/tree/65c1978357ca169b00dd870f60068f8bc5c03e2d
KindOtherBase modelKkonjeong/wav2vec2-base-koreanVersionRevision 65c1978

Model source

Kind
Other
File size
361 MB
Base model
Kkonjeong/wav2vec2-base-korean
Version
vRevision 65c1978
Updated
Sep 13, 2026
Downloads
2
Source
Hugging Face

Source description

korean_kws2

This model is a fine-tuned version of Kkonjeong/wav2vec2-base-korean on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2259
  • Accuracy: 1.0

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
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 32
  • optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • num_epochs: 30

Training results

Training LossEpochStepValidation LossAccuracy
No log1.021.92120.2632
No log2.041.85810.2105
No log3.061.69240.5263
No log4.081.51700.5789
No log5.0101.34860.8421
No log6.0121.18421.0
No log7.0141.05890.9474
No log8.0160.99190.8947
No log9.0180.79781.0
No log10.0200.72611.0
No log11.0220.64401.0
No log12.0240.60201.0
No log13.0260.52781.0
No log14.0280.48311.0
No log15.0300.42551.0
No log16.0320.38441.0
No log17.0340.35041.0
No log18.0360.32771.0
No log19.0380.30661.0
No log20.0400.29191.0
No log21.0420.27931.0
No log22.0440.26681.0
No log23.0460.25621.0
No log24.0480.24791.0
No log25.0500.24101.0
No log26.0520.23551.0
No log27.0540.23131.0
No log28.0560.22831.0
No log29.0580.22671.0
No log30.0600.22591.0

Framework versions

  • Transformers 4.57.6
  • Pytorch 2.10.0
  • Datasets 4.6.0
  • Tokenizers 0.22.2

Sources

1 source
Hugging FacePrimaryVersion 65c1978357ca169b00dd870f60068f8bc5c03e2d
huggingface.co/devtaco/korean_kws2/tree/65c1978357ca169b00dd870f60068f8bc5c03e2d

Verified Sep 13

Model artifacts

1 artifact

model.safetensors

safetensors · 361 MB · SHA-256 81080d6f7757…b2f5 · Hugging Face

Download

Source excerpts

2 excerpts
ReadmeSep 13huggingface.co/devtaco/korean_kws2

--- library_name: transformers base_model: Kkonjeong/wav2vec2-base-korean tags: - generated_from_trainer metrics: - accuracy model-index: - name: korean_kws2 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. --> # korean_kws2 This model is a fine-tuned version of [Kkonjeong/wav2vec2-base-korean](https://huggingface.co/Kkonjeong/wav2vec2-base-korean) on the None dataset. It achieves the following results on the evaluation set: - Loss: 0.2259 - Accuracy: 1.0 ## 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 - gradient_accumulation_steps: 4 - total_train_batch_size: 32 - optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments - lr_scheduler_type: linear - num_epochs: 30 ### Training results | Training Loss | Epoch | Step | Validation Loss | Accuracy | |:-------------:|:-----:|:----:|:---------------:|:--------:| | No log | 1.0 | 2 | 1.9212 | 0.2632 | | No log | 2.0 | 4 | 1.8581 | 0.2105 | | No log | 3.0 | 6 | 1.6924 | 0.5263 | | No log | 4.0 | 8 | 1.5170 | 0.5789 | | No log | 5.0 | 10 | 1.3486 | 0.8421 | | No log | 6.0 | 12 | 1.1842 | 1.0 | | No log | 7.0 | 14 | 1.0589 | 0.9474 | | No log | 8.0 | 16 | 0.9919 | 0.8947 | | No log | 9.0 | 18 | 0.7978 | 1.0 | | No log | 10.0 | 20 | 0.7261 | 1.0 | | No log | 11.0 | 22 | 0.6440 | 1.0 | | No log | 12.0 | 24 | 0.6020 | 1.0 | | No log | 13.0 | 26 | 0.5278 | 1.0 | | No log | 14.0 | 28 | 0.4831 | 1.0 | | No log | 15.0 | 30 | 0.4255 | 1.0 | | No log | 16.0 | 32 | 0.3844 | 1.0 | | No log | 17.0 | 34 | 0.3504 | 1.0 | | No log | 18.0 | 36 | 0.3277 | 1.0 | | No log | 19.0 | 38 | 0.3066 | 1.0 | | No log | 20.0 | 40 | 0.2919 | 1.0 | | No log | 21.0 | 42 | 0.2793 | 1.0 | | No log | 22.0 | 44 | 0.2668 | 1.0 | | No log | 23.0 | 46 | 0.2562 | 1.0 | | No log | 24.0 | 48 | 0.2479 | 1.0 | | No log | 25.0 | 50 | 0.2410 | 1.0 | | No log | 26.0 | 52 | 0.2355 | 1.0 | | No log | 27.0 | 54 | 0.2313 | 1.0 |...

NoteSep 13huggingface.co/devtaco/korean_kws2

Source context: 2 downloads · 0 likes · Pipeline audio-classification · Library transformers · Repo devtaco/korean_kws2