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 dataset is a multimodal dataset comprising 1400 training examples, each consisting of images and text. Features include: image, caption, captions, label, prompt_random and its ID, retrieval-based prompt (prompt_retrieval_k1) with its ID and score, RAG (Retrieval-Augmented Generation) prompt (prompt_rag_k2_slm) with its ID list and score list, and related metadata such as rag_short_class_key and rag_short_source_corpus. The dataset is designed to support image captioning, prompt engineering, and retrieval-augmented generation tasks, likely for evaluating or training NLP and computer vision models.