Ir para o conteúdo
ModelListadosExterno
Baixar

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: 0.3043 Accuracy: 1.0

Fontehuggingface.co/rewining/korean_kws2/tree/31409b162e3f3420e3a3b3e266fac85add48daae
TipoOtherModelo baseKkonjeong/wav2vec2-base-koreanVersãoRevision 31409b1Cnota

Fonte do modelo

Tipo
Other
Tamanho do arquivo
361 MB
Modelo base
Kkonjeong/wav2vec2-base-korean
Versão
vRevision 31409b1
Atualizado
17 de out. de 2026
Downloads
4
Fonte
Hugging Face

Descrição da fonte

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: 0.3043
  • 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 log1.021.88860.2105
No log2.041.78590.5263
No log3.061.73170.4211
No log4.081.64820.6842
No log5.0101.49180.6316
No log6.0121.41120.7368
No log7.0141.28460.8947
No log8.0161.14070.8947
No log9.0181.00680.8947
No log10.0200.91920.9474
No log11.0220.84460.8947
No log12.0240.77040.9474
No log13.0260.68061.0
No log14.0280.62371.0
No log15.0300.57771.0
No log16.0320.52501.0
No log17.0340.48611.0
No log18.0360.46731.0
No log19.0380.44151.0
No log20.0400.41241.0
No log21.0420.38511.0
No log22.0440.36431.0
No log23.0460.34981.0
No log24.0480.33961.0
No log25.0500.33081.0
No log26.0520.32231.0
No log27.0540.31491.0
No log28.0560.30971.0
No log29.0580.30601.0
No log30.0600.30431.0

Framework versions

  • Transformers 5.0.0
  • Pytorch 2.10.0+cu128
  • Datasets 4.0.0
  • Tokenizers 0.22.2

Fontes

1 fonte
Hugging FacePrimárioVersão 31409b162e3f3420e3a3b3e266fac85add48daae
huggingface.co/rewining/korean_kws2/tree/31409b162e3f3420e3a3b3e266fac85add48daae

Verificado 29 de ago.

Artefatos de modelo

1 artefato

model.safetensors

safetensors · 361 MB · SHA-256 e827a7531e2d…56aa · Hugging Face

Download file

Trechos de fonte

2 trechos
Readme29 de ago.huggingface.co/rewining/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: 0.3043 - 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 | 1.8886 | 0.2105 | | No log | 2.0 | 4 | 1.7859 | 0.5263 | | No log | 3.0 | 6 | 1.7317 | 0.4211 | | No log | 4.0 | 8 | 1.6482 | 0.6842 | | No log | 5.0 | 10 | 1.4918 | 0.6316 | | No log | 6.0 | 12 | 1.4112 | 0.7368 | | No log | 7.0 | 14 | 1.2846 | 0.8947 | | No log | 8.0 | 16 | 1.1407 | 0.8947 | | No log | 9.0 | 18 | 1.0068 | 0.8947 | | No log | 10.0 | 20 | 0.9192 | 0.9474 | | No log | 11.0 | 22 | 0.8446 | 0.8947 | | No log | 12.0 | 24 | 0.7704 | 0.9474 | | No log | 13.0 | 26 | 0.6806 | 1.0 | | No log | 14.0 | 28 | 0.6237 | 1.0 | | No log | 15.0 | 30 | 0.5777 | 1.0 | | No log | 16.0 | 32 | 0.5250 | 1.0 | | No log | 17.0 | 34 | 0.4861 | 1.0 | | No log | 18.0 | 36 | 0.4673 | 1.0 | | No log | 19.0 | 38 | 0.4415 | 1.0 | | No log | 20.0 | 40 | 0.4124 | 1.0 | | No log | 21.0 | 42 | 0.3851 | 1.0 | | No log | 22.0 | 44 | 0.3643 | 1.0 | | No log | 23.0 | 46 | 0.3498 | 1.0 | | No log | 24.0 | 48 | 0.3396 | 1.0 | | No log | 25.0 | 50 | 0.3308 | 1.0 | | No log | 26.0 | 52 |...

Note29 de ago.huggingface.co/rewining/korean_kws2

Source context: 4 downloads · 0 likes · Pipeline audio-classification · Library transformers · Repo rewining/korean_kws2