SentiCap

该数据集被用于训练模型，目的是生成具有特定风格的标题，这些风格包括积极的和消极的情感。此外，该数据集还用于对基于情感的生成标题进行微调，以优化特定风格的LoRAs。这项任务旨在为视觉-语言模型实现风格特定的标题生成。

This dataset is utilized to train models to generate captions with specific emotional styles, including positive and negative sentiments. Moreover, this dataset is also employed to fine-tune emotion-based caption generation workflows to optimize LoRAs tailored for specific styles. This task aims to enable style-specific caption generation for vision-language models.