MultiBench

该数据集是一个大规模的基准测试，涵盖了15个数据集、10种模态、20个预测任务以及6个研究领域，为多模态深度学习的评估提供了全面的衡量标准。此外，MultiBench还提供了一个自动化的端到端机器学习流程，包括数据加载、实验设置和模型评估。其任务是在各种任务中跨多种模态进行表示学习与评估。

This dataset is a large-scale benchmark that encompasses 15 datasets, 10 modalities, 20 prediction tasks, and 6 research domains, providing a comprehensive evaluation framework for multimodal deep learning. Additionally, MultiBench offers an automated end-to-end machine learning pipeline, including data loading, experimental setup, and model evaluation. The core task of MultiBench is to perform representation learning and evaluation across diverse modalities across a range of tasks.