Is Grounding Required to Understand Language?

3min Snip

00:00

Play full episode

Summary

Transcript

Episode notes

The study was based on a psychological study that produced a data set testing, whether or not particular individuals had trouble understanding implicature. And we turned it into binary classification tasks that allowed us without further training to just see whether models can sort of capture this. And we noticed a fairly interesting result. When you start looking at open eyes, da Vinci models, and now we're looking at other sort of models in a similar class, the models that are trained to follow instructions, you notice that they don't necessarily operate at human performance. But they they're performing significantly better than randomly. They're close to human performance, if you look at their, at how well they're correlating

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