How2: A Large-scale Dataset for Multimodal Language Understanding

Code

• github.com/srvk/how2-dataset
  OfficialIn paperpytorch★ 0
• github.com/jasonppy/promptingwhisper
  pytorch★ 150

Abstract

In this paper, we introduce How2, a multimodal collection of instructional videos with English subtitles and crowdsourced Portuguese translations. We also present integrated sequence-to-sequence baselines for machine translation, automatic speech recognition, spoken language translation, and multimodal summarization. By making available data and code for several multimodal natural language tasks, we hope to stimulate more research on these and similar challenges, to obtain a deeper understanding of multimodality in language processing.

Tasks

Reproductions