Deep Papers cover image

Toolformer: Training LLMs To Use Tools

Deep Papers

CHAPTER

Using Pre-Trained Language Models to Generate API Calls

Teemo: One of the problems with this is getting a huge annotated dataset. And part of the problem is because the scale of these text datasets is so large, like common crawl, you can't just pay human labelers to do it. We wanted to avoid having huge costs from humans having to annotate data. So we use a language model to sample a massive amount of different API calls. Then we give it three examples of an input and the corresponding output would be the same text with API calls inserted. And then we ask the model to do the same for a fourth text.

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