The Gradient: Perspectives on AI cover image

Sewon Min: The Science of Natural Language

The Gradient: Perspectives on AI

CHAPTER

How to Decompose a Multi-Hop Question Answering Benchmark

There are some tasks or particular instantiations of a multi-hop question answering benchmark where a model can actually achieve this through single-hop reasoning. I'm curious how you think about those both from the perspective of developing a benchmark around them, but then also what a modeler might have to do in order to start tackling those questions.

00:00
Transcript
Play full episode

Remember Everything You Learn from Podcasts

Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
App store bannerPlay store banner