Protein Representation Alignment Dataset

该数据集旨在评估大型语言模型（LLM）与几何深度模型（GDMs）在蛋白质领域中的多模态表示对齐情况，重点关注蛋白质对和模型对的配对关系。数据集包含了使用余弦相似度计算的项目图与文本表示之间的对齐分数，这有助于评估不同模型对的配对及其对齐性能。该任务的目标是进行多模态对齐评估。

This dataset is designed to evaluate the multimodal representation alignment between Large Language Models (LLMs) and Geometric Deep Models (GDMs) in the protein domain, with a focus on the pairing relationships between protein pairs and model pairs. The dataset contains alignment scores between graphical representations and text representations calculated via cosine similarity, which assists in assessing the pairing and alignment performance of different model pairs. The objective of this task is to conduct multimodal alignment evaluation.