metadata

library_name: peft
language:
  - th
license: apache-2.0
base_model: openai/whisper-large-v3
tags:
  - generated_from_trainer
datasets:
  - mozilla-foundation/common_voice_17_0
metrics:
  - wer
model-index:
  - name: Whisper Large v3 Thai Finetuned
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: Common Voice 17.0
          type: mozilla-foundation/common_voice_17_0
          config: th
          split: None
          args: 'config: th, split: train'
        metrics:
          - type: wer
            value: 37.2840522511834
            name: Wer

Whisper Large v3 Thai Finetuned

This model is a fine-tuned version of openai/whisper-large-v3 on the Common Voice 17.0 dataset. It achieves the following results on the evaluation set:

Loss: 0.1197
Cer: 347.2750
Wer: 37.2841

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: 5e-05
train_batch_size: 16
eval_batch_size: 16
seed: 42
optimizer: Use paged_adamw_8bit with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_ratio: 0.01
num_epochs: 5
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer	Wer
0.1468	1.0	2052	0.1265	348.7140	42.3821
0.1225	2.0	4104	0.1178	387.4029	34.6267
0.0914	3.0	6156	0.1157	368.2561	36.4184
0.0779	4.0	8208	0.1182	347.7405	36.8434
0.0624	5.0	10260	0.1197	347.2750	37.2841

Framework versions

PEFT 0.14.1.dev0
Transformers 4.49.0.dev0
Pytorch 2.5.1+cu124
Datasets 3.2.0
Tokenizers 0.21.0