Hugging Face's logo

Marcusxx
/
CngFSt10sec_model

TensorBoard
Safetensors
Korean
whisper
hf-asr-leaderboard
Generated from Trainer
Model card Files Metrics Community
CngFSt10sec_model / README.md
Marcusxx's picture
Marcusxx
Training in progress, step 1000
2777b5c verified
preview code
|
raw
history blame
1.76 kB
metadata
base_model: openai/whisper-medium
datasets:
  - Marcusxx/CngFSt10sec
language:
  - ko
license: apache-2.0
tags:
  - hf-asr-leaderboard
  - generated_from_trainer
model-index:
  - name: CngFSt10sec_model
    results: []

CngFSt10sec_model

This model is a fine-tuned version of openai/whisper-medium on the Marcusxx/CngFSt10sec dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0508
  • Cer: 73.2002

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: 1e-05
  • train_batch_size: 16
  • 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: 500
  • training_steps: 5000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Cer
0.0079 8.2645 1000 0.0496 79.5377
0.0004 16.5289 2000 0.0442 40.5683
0.0002 24.7934 3000 0.0477 57.3026
0.0001 33.0579 4000 0.0498 72.9351
0.0001 41.3223 5000 0.0508 73.2002

Framework versions

  • Transformers 4.41.2
  • Pytorch 2.2.2+cu121
  • Datasets 2.20.0
  • Tokenizers 0.19.1