metadata

base_model: openai/whisper-medium
datasets:
  - Marcusxx/CngFSt10_5
language:
  - ko
license: apache-2.0
tags:
  - hf-asr-leaderboard
  - generated_from_trainer
model-index:
  - name: CngFSt10_5_sec_model
    results: []

CngFSt10_5_sec_model

This model is a fine-tuned version of openai/whisper-medium on the Marcusxx/CngFSt10_5 dataset. It achieves the following results on the evaluation set:

Loss: 0.0075
Cer: 20.8184

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
0.0326	3.6765	1000	0.0403	67.4303
0.0038	7.3529	2000	0.0172	47.8057
0.0035	11.0294	3000	0.0079	11.7720
0.0002	14.7059	4000	0.0074	23.4474
0.0001	18.3824	5000	0.0075	20.8184

Framework versions

Transformers 4.41.2
Pytorch 2.2.2+cu121
Datasets 2.20.0
Tokenizers 0.19.1