Rapidata/xAI_Aurora_t2i_human_preferences

这是一个包含超过40万人类评价的文本到图像评估数据集，用于评估Aurora模型与其他六个模型在偏好、连贯性和对齐性三个方面的表现。数据集通过Rapidata Python API在两天内收集，包含86,000个独立注释者的反馈。This is a text-to-image evaluation dataset containing over 400,000 human evaluations, assessing the performance of the Aurora model and six other models (Imagen-3, Flux-1.1-pro, Flux-1-pro, DALL-E 3, Midjourney-5.2, Stable Diffusion 3) in terms of preference, coherence, and alignment. The dataset was collected in just two days using the Rapidata Python API, with feedback from 86,000 independent annotators.

搜集汇总

背景与挑战

背景概述

该数据集包含超过40万条人类反馈，用于评估xAI的Aurora文本到图像生成模型，与Imagen-3、Flux-1.1-pro、Flux-1-pro、DALL-E 3、Midjourney-5.2和Stable Diffusion 3六个模型进行1v1对比，涵盖偏好、连贯性和对齐三个维度，数据由来自全球的86,000多名标注者收集，大小为3.48 GB。