sft_data_mixed.json: The supervised finetuning data containing all style-related tasks.
style_profile: Raw data for profiles of different styles.
style_transfer: Raw data for the style transfer task.
stylized_dialogue: Raw data for stylized dialogue tasks.
eval: Evaluation Datasets
test.json: The overall test dataset.
test_multi_model.json: Data used to benchmark different models.
test_multi_turn_seed_prompts.json:… See the full description on the dataset page: https://huggingface.co/datasets/JustinZekai/StyleEval.
