ButterChicken98/soyabean_ymv_plus_healthy_rag_short_ablation

该数据集是一个多模态数据集，包含1400个训练样本，每个样本由图像和文本组成。特征包括：图像（image）、单个标题（caption）、多个标题（captions）、标签（label）、随机提示（prompt_random）及其ID、基于检索的提示（prompt_retrieval_k1）及其ID和分数、RAG（检索增强生成）提示（prompt_rag_k2_slm）及其ID列表和分数列表，以及相关元数据如rag_short_class_key和rag_short_source_corpus。数据集设计用于支持图像描述生成、提示工程和检索增强生成任务，可能用于评估或训练NLP和计算机视觉模型。This is a multimodal dataset containing 1400 training samples, where each sample consists of an image and paired text. Its features include: image, single caption, multiple captions, label, random prompt (prompt_random) along with its ID, retrieval-based prompt (prompt_retrieval_k1) along with its ID and score, RAG (Retrieval-Augmented Generation) prompt (prompt_rag_k2_slm) along with its ID list and score list, as well as relevant metadata such as rag_short_class_key and rag_short_source_corpus. This dataset is designed to support tasks including image caption generation, prompt engineering and retrieval-augmented generation, and can be utilized for evaluating or training natural language processing (NLP) and computer vision models.