Public demo

Agents & Tools

A source-grounded theme page on Toolformer: language models using APIs, tool calls, self-supervised data, and reliability limits.

Theme overview

This theme covers the move from language models that only generate text to systems that decide when to use external tools. toolformer is an early example of learning tool-use behavior from model-generated API call examples.

Takeaways

• Tool use expands what an LLM can do, but it does not automatically solve reliability.
• Evidence search is a tool; the product value is making tool output inspectable and usable while writing.
• This theme connects directly to the planned Word add-in and right-side evidence panel.

Included papers

• Toolformer: Language Models Can Teach Themselves to Use Tools

Evidence examples

• Toolformer: Language Models Can Teach Themselves to Use Tools: Toolformer enables language models to self-teach tool usage via API calls, overcoming limitations for broad task generality and functional accuracy without degrading performance.

  Therefore, we propose Toolformer, a model that learns to use tools in a novel way, which fulfills the following desiderata:

• Toolformer: Language Models Can Teach Themselves to Use Tools: The ability to ask external tools for help is investigated as model size is varied.

  We investigate how the ability to ask external tools for help affects performance as we vary the size of our LM.