LLMs can be programmed by backprop 🔎

In our new preprint, we show they can act as fuzzy program interpreters and databases. After being ‘programmed’ with next-token prediction, they can retrieve, evaluate, and even *compose* programs at test time, without seeing I/O examples. In a range of experimental settings, we discover that LLMs have some ability to evaluate programs for inputs when trained on their source code alone (no I/O examples). In this thread: my 3 favourite findings, and where I think this line of work can go.

How do LLMs learn to reason from data? Are they ~retrieving the answers from parametric knowledge🦜? In our new preprint, we look at the pretraining data and find evidence against this:

Procedural knowledge in pretraining drives LLM reasoning ⚙️🔢

🧵⬇️ Since LLMs entered the stage, there has been a hypothesis prevalent:

When LLMs are reasoning, they are doing some form of approximate retrieval where they “retrieve” the answer to intermediate reasoning steps from parametric knowledge, as opposed to doing “genuine” reasoning.

I simultaneously have chatGPT-fatigue and also think about it every day - it truly changed the discourse around LLMs. Plus, chatty-P provided us with _so_ many memes. The following is my current mental model of chatGPT (now popularly synonymous to LLMs) and what might be missing. Contrary to what we sometimes read on twitter, language is meaningfully different from other types of communication; it's a complex system of symbolic reference.

Lots of people have been sending me implicatures we used as examples in our paper that #chatGPT understands (i.e. explains well when prompted). So cool! Happy to see people interested in this. I wanted to write a short thread about my thoughts of what this means. ⬇️ E.g. left is chatGPT, right is Davinci-2

Some of the best #advice I got early in my PhD was from @rockt urging me to set up a good note-keeping system. 1 year in my notes have helped me in so many situations! It took a few iterations, and I'll share the things that were most helpful for me personally here ⬇️🐊 1/13 Everything below I use @obsdmd for. I love how it's simply a markdown editor and file-structure visualiser at first but you can add all the plugins you need to make it ever more complex.

Before giving up on Twitter, check out our recent finding 🔥 LLMs are *not* zero-shot communicators!

We show a limitation of #LLMs in interpreting language given the context of shared experience of the world, a ubiquitous part of communication ⬇️🧵

arxiv.org/abs/2210.14986

1/22 Work with awesome collaborators @akbirkhan @BlancheMinerva @sarahookr @_rockt @egrefen

If this thread is long, it’s only because I had to keep the meme-to-information ratio at a respectable level (and because there’s a lot to unpack)

lauraruis.github.io/2022/09/29/com…

2/22