Toolformer: Language Models Can Teach Themselves to Use Tools

TL;DR

Toolformer enables language models to self-teach the use of external tools via APIs, significantly enhancing zero-shot task performance without additional training data.

Contribution

It introduces a self-supervised training method for LMs to learn when and how to use external APIs, improving their versatility and accuracy.

Findings

Enhanced zero-shot performance across multiple tasks

Competitive results with larger models

Effective integration of diverse external tools

Abstract

Language models (LMs) exhibit remarkable abilities to solve new tasks from just a few examples or textual instructions, especially at scale. They also, paradoxically, struggle with basic functionality, such as arithmetic or factual lookup, where much simpler and smaller models excel. In this paper, we show that LMs can teach themselves to use external tools via simple APIs and achieve the best of both worlds. We introduce Toolformer, a model trained to decide which APIs to call, when to call them, what arguments to pass, and how to best incorporate the results into future token prediction. This is done in a self-supervised way, requiring nothing more than a handful of demonstrations for each API. We incorporate a range of tools, including a calculator, a Q\&A system, two different search engines, a translation system, and a calendar. Toolformer achieves substantially improved zero-shot performance across a variety of downstream tasks, often competitive with much larger models, without sacrificing its core language modeling abilities.