SLLM-multi-hop/GenderQA

SAKURA-GenderQA数据集包含了来自SAKURA基准的性别轨道的音频文件以及单跳和多跳问题及其答案。该数据集用于研究基于语音和音频信息的大型音频-语言模型的multi-hop推理。数据集中的字段包括音频文件名、音频记录、属性标签（说话者的性别）、单跳问题（指令）、单跳答案、多跳问题（指令）和多跳答案。

The SAKURA-GenderQA dataset contains the audio files and the single/multi-hop questions/answers from the gender track of the SAKURA benchmark. It is used for studying the multi-hop reasoning of large audio-language models based on speech and audio information. The fields in the dataset include the filename of the audio files, the audio recordings, the attribute labels (i.e., gender of the speakers), single-hop questions (instructions), single-hop answers, multi-hop questions (instructions), and multi-hop answers.