Seonglae ChoFunction Calling, Tool Calling, AI Tooling
Bash is all you need. Bash is almost you need
Considering from the training stage, there are fewer hallucinations but less robustness.
Mainly implemented through prompt engineering, with structured text generation following the instruction specified in the system prompt.
e.g.
search(query='latest news from New York')u ‣Tool learning Notion
Tool learning Extensions
Mastery Dividend Period and..
We’ve crossed a threshold where the metagame has fundamentally changed. Mastery of tools no longer pays dividends for the rest of our lifetime—we’re lucky if a new AI tool stays relevant for 12 months! It’s incredibly disorienting when the “sharpen the axe” strategy that worked for thousands of years suddenly stops working.
Berkeley Function-Calling Leaderboard (Gorilla LLM )
Berkeley Function Calling Leaderboard V2 Live(aka Berkeley Tool Calling Leaderboard V2 Live)
Explore The Berkeley Function Calling Leaderboard (also called The Berkeley Tool Calling Leaderboard) to see the LLM's ability to call functions (aka tools) accurately.
https://gorilla.cs.berkeley.edu/leaderboard_live.html
Tool Calling with LangChain
TLDR: We are introducing a new tool_calls attribute on AIMessage. More and more LLM providers are exposing API’s for reliable tool calling. The goal with the new attribute is to provide a standard interface for interacting with tool invocations. This is fully backwards compatible and is supported on
https://blog.langchain.dev/tool-calling-with-langchain/
arxiv.org
https://arxiv.org/pdf/2402.04253.pdf