lmms-lab/LiveBenchDetailedResults

该数据集是一个多模态数据集，包含图像和文本信息。它主要用于评估模型在图像理解和文本生成任务上的性能。数据集中的每个样本都包括一个唯一标识符、图像、问题、真实答案、评估标准、子任务、模型响应、得分和得分原因等字段。数据集分为多个子集，每个子集对应不同的模型配置和评估结果。

This dataset is a multimodal dataset containing both images and text. It is primarily used for evaluating model performance on image understanding and text generation tasks. Each sample in the dataset includes a unique identifier, image, question, ground truth, evaluation criteria, subtask, model response, score, and reason for the score. The dataset is divided into multiple subsets, each corresponding to different model configurations and evaluation results.