SESKStuff Everyone Should Know
SESK: Stuff Everyone Should Know · sesk.info/ai-toolformer-tools-for-weak-spots

Toolformer taught a language model to call tools. Which weak spots of LMs did that target?

Arithmetic and factual lookup, where tiny tools beat huge models.

Start playing free

  1. Writing poems and stories
  2. Translating common sentences
  3. Arithmetic and looking up facts ✔

Why: Big language models struggle with exact math and fresh facts, while a calculator or search engine nails them. Toolformer learned when to call APIs like a calculator, search and a calendar, and how to use the results.

Huh, didn't know that: Toolformer learned tool use in a self-supervised way from just a handful of demonstrations per API.

The dad joke
Even a giant AI needs a calculator. I feel so much better about my math now.
Like this card?
Play the game

References

Answer: Arithmetic and looking up facts. Big language models struggle with exact math and fresh facts, while a calculator or search engine nails them. Toolformer learned when to call APIs like a calculator, search and a calendar, and how to use the results.

  1. Schick et al., Toolformer: Language Models Can Teach Themselves to Use Tools (arXiv 2302.04761) arxiv.org/abs/2302.04761
    They also, paradoxically, struggle with basic functionality, such as arithmetic or factual lookup, where much simpler and smaller models excel.
    Checked 2026-10-10.

Think this answer is wrong? Tap "Challenge this answer" on the card in the game and tell us why.

Test yourself in the game

Artificial IntelligenceWith function calling, who actually runs a tool that the developer defined?Artificial IntelligenceWhat does the ReAct method have a language model do?Artificial IntelligenceIn reinforcement learning, what tells the agent how well an action went?

That's stuff everyone should know.