Learning Machines to Read and Compose

5min Snip

00:00

Play full episode

Summary

Transcript

Episode notes

I thought this paper was really interesting because I did observe some ideas that seemed to have been fleshed out more in the transformer. In contrast with transformers where you need to provide that explicitly, or at least more or less explicitly with positional embeddings. So I think maybe we can bridge towards some of your more recent work. And I think the next one I wanted to discuss was teaching machines to read and comprehend.

In episode 62 of The Gradient Podcast, Daniel Bashir speaks to Ed Grefenstette.

Ed is Head of Machine Learning at Cohere and an Honorary Professor at University College London. He previously held research scientist positions at Facebook AI Research and DeepMind, following a stint as co-founder and CTO of Dark Blue Labs. Before his time in industry, Ed worked at Oxford’s Department of Computer Science as a lecturer and Fulford Junior Research Fellow at Somerville College. Ed also received his MSc and DPhil from Oxford’s Computer Science Department.

Have suggestions for future podcast guests (or other feedback)? Let us know here!

Subscribe to The Gradient Podcast: Apple Podcasts | Spotify | Pocket Casts | RSSFollow The Gradient on Twitter

Outline:

* (00:00) Intro

* (02:18) The Ed Grefenstette Origin Story

* (08:15) Distributional semantics and Ed’s PhD research

* (14:30) Extending the distributional hypothesis, later Wittgenstein

* (18:00) Recovering parse trees in LMs, can LLMs understand communication and not just bare language?

* (23:15) LMs capture something about pragmatics, proxies for grounding and pragmatics

* (25:00) Human-in-the-loop training and RLHF—what is the essential differentiator?

* (28:15) A convolutional neural network for modeling sentences, relationship to attention

* (34:20) Difficulty of constructing supervised learning datasets, benchmark-driven development