Skip to content
ModelListedExternal
Download

Korean Kws

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: 0.2861 Accuracy: 1.0

Sourcehuggingface.co/bsb0107/korean_kws/tree/30886d5826f3db1a1d022034e9171f5d669b2f70
KindOtherBase modelKkonjeong/wav2vec2-base-koreanVersionRevision 30886d5

Model source

Kind
Other
File size
361 MB
Base model
Kkonjeong/wav2vec2-base-korean
Version
vRevision 30886d5
Updated
Sep 19, 2026
Downloads
9
Source
Hugging Face

Source description

korean_kws

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: 0.2861
  • 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
  • mixed_precision_training: Native AMP

Training results

Training LossEpochStepValidation LossAccuracy
No log1.021.92040.0769
No log2.041.84850.1538
No log3.061.71390.0769
No log4.081.64310.3077
No log5.0101.47320.5385
No log6.0121.29831.0
No log7.0141.21260.8462
No log8.0161.08951.0
No log9.0180.98801.0
No log10.0200.88461.0
No log11.0220.81911.0
No log12.0240.74761.0
No log13.0260.68571.0
No log14.0280.62311.0
No log15.0300.57121.0
No log16.0320.52181.0
No log17.0340.48331.0
No log18.0360.44451.0
No log19.0380.41841.0
No log20.0400.39061.0
No log21.0420.37171.0
No log22.0440.35071.0
No log23.0460.33271.0
No log24.0480.32031.0
No log25.0500.30971.0
No log26.0520.30071.0
No log27.0540.29451.0
No log28.0560.29011.0
No log29.0580.28741.0
No log30.0600.28611.0

Framework versions

  • Transformers 4.57.3
  • Pytorch 2.9.0+cu126
  • Datasets 4.0.0
  • Tokenizers 0.22.1

Sources

1 source
Hugging FacePrimaryVersion 30886d5826f3db1a1d022034e9171f5d669b2f70
huggingface.co/bsb0107/korean_kws/tree/30886d5826f3db1a1d022034e9171f5d669b2f70

Verified Sep 19

Model artifacts

1 artifact

model.safetensors

safetensors · 361 MB · SHA-256 22d81aae9483…c864 · Hugging Face

Download

Source excerpts

2 excerpts
ReadmeSep 19huggingface.co/bsb0107/korean_kws

--- library_name: transformers base_model: Kkonjeong/wav2vec2-base-korean tags: - generated_from_trainer metrics: - accuracy model-index: - name: korean_kws 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_kws 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: 0.2861 - 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 - mixed_precision_training: Native AMP ### Training results | Training Loss | Epoch | Step | Validation Loss | Accuracy | |:-------------:|:-----:|:----:|:---------------:|:--------:| | No log | 1.0 | 2 | 1.9204 | 0.0769 | | No log | 2.0 | 4 | 1.8485 | 0.1538 | | No log | 3.0 | 6 | 1.7139 | 0.0769 | | No log | 4.0 | 8 | 1.6431 | 0.3077 | | No log | 5.0 | 10 | 1.4732 | 0.5385 | | No log | 6.0 | 12 | 1.2983 | 1.0 | | No log | 7.0 | 14 | 1.2126 | 0.8462 | | No log | 8.0 | 16 | 1.0895 | 1.0 | | No log | 9.0 | 18 | 0.9880 | 1.0 | | No log | 10.0 | 20 | 0.8846 | 1.0 | | No log | 11.0 | 22 | 0.8191 | 1.0 | | No log | 12.0 | 24 | 0.7476 | 1.0 | | No log | 13.0 | 26 | 0.6857 | 1.0 | | No log | 14.0 | 28 | 0.6231 | 1.0 | | No log | 15.0 | 30 | 0.5712 | 1.0 | | No log | 16.0 | 32 | 0.5218 | 1.0 | | No log | 17.0 | 34 | 0.4833 | 1.0 | | No log | 18.0 | 36 | 0.4445 | 1.0 | | No log | 19.0 | 38 | 0.4184 | 1.0 | | No log | 20.0 | 40 | 0.3906 | 1.0 | | No log | 21.0 | 42 | 0.3717 | 1.0 | | No log | 22.0 | 44 | 0.3507 | 1.0 | | No log | 23.0 | 46 | 0.3327 | 1.0 | | No log | 24.0 | 48 | 0.3203 | 1.0 | | No log | 25.0 | 50 | 0.3097 | 1.0 | | No log | 26.0 | 52 | 0.3007 | 1.0 | | No...

NoteSep 19huggingface.co/bsb0107/korean_kws

Source context: 9 downloads · 0 likes · Pipeline audio-classification · Library transformers · Repo bsb0107/korean_kws