gorilla

Gorilla is a Python project developed to train and evaluate the function and tool calling (tool call) capabilities of large language models. The Berkeley Function-Calling Leaderboard (BFCL) enables testing multistep workflows, agent scenarios, and cost/delay metrics. It also validates actions produced by LLM with the GoEx runtime environment, manages risks and provides retrieval abstractions.

GO

ShishirPatil · Docs & Developer Portals

STARS13.1K
FORKS1.4K
COMMITS470
CONTRIBUTORS140
VERSIONv1.3
UPDATED14 days ago
PythonLLMFunction CallingAPIApache-2.0

Common commercial products in this category: ReadMe, GitBook, Document360

What can it do?
  • You can compare function-calling performance of different major language models with Berkeley Function-Calling Leaderboard.
  • Thanks to the GoEx runtime engine, you can execute and then verify the code and API calls produced by LLM.
  • Using the Agent Arena platform, you can test and rank different agents in tasks such as search, finance, and RAG.
User reviews
  • • NeutralHacker News

    Gorilla Youngsters Seen Dismantling Poachers’ Traps (2012)

    “Intelligence is clearly useful for almost everything. And that's exactly my point. The closer we look the more intelligence we find in nature. But there's a tremendous qualitative gap between human-type intelligence and everything else. When we believed other animals were dumb…”

  • • NeutralHacker News

    Gorilla Youngsters Seen Dismantling Poachers’ Traps (2012)

    “Could the gorillas be the poachers? The story seems consistent with the gorrilas warning the tracker by grunt to stay put, then dismantle their nearby traps to protect the humans. This is also consistent with the fact that the tracker hadn't noticed the second trap. I don't…”

My note

Alternative pages

More in: Docs & Developer Portals