Your Daily AI Research tl;dr — 2022–10–28 🧠

I was featured in the “Who’s Who in artificial intelligence” report! (& very cool large language models papers)

Welcome to your official daily AI research tl;dr (often with code and news) for AI professionals where I share the most exciting papers I find daily, along with a one-liner summary to help you quickly determine if the article (and code) is worth investigating.

Receive your daily update right in your inbox ⬇️

1️⃣ Large language models are not zero-shot communicators

Humans intuitively understand the response “I wore gloves” to the question “Did you leave fingerprints?” as meaning “No”. They design a simple task and evaluate widely used state-of-the-art models to figure out if large language models can understand implicatures (the type of inference mentioned above).

Link to the paper: https://arxiv.org/abs/2210.14986

2️⃣ What Language Model to Train if You Have One Million GPU Hours?

“We perform an ablation study at the billion-parameter scale comparing different modeling practices and their impact on zero-shot generalization.”