FM-IQA 多语种问答数据集

FM-IQA 全称 Freestyle Multilingual Image Question Answering，是一个问答数据集，包含 15 万张图像和 31 万个自由式中文问答对及其英文翻译。该数据集可用于训练和评估 mQA 模型。该模型可用于回答关于图像内容的问题，答案可以是一个句子、一个短语或一个单词。

FM-IQA, whose full name is Freestyle Multilingual Image Question Answering, is a question answering dataset. It comprises 150,000 images and 310,000 freestyle Chinese question-answer pairs along with their English translations. This dataset can be utilized for training and evaluating mQA models. These models are capable of answering questions regarding image content, and their answers can be sentences, phrases, or single words.

搜集汇总

背景与挑战

背景概述

FM-IQA 多语种问答数据集包含15万张图像和31万个中英文问答对，支持自由式问答，适用于多语种问答模型的训练和评估。