Join the conversation

Join the community of Machine Learners and AI enthusiasts.

Sign Up
anakin87ย 
posted an update 8 days ago
Post
288
๐Ÿ๐Ÿ๐Ÿ ๐€ ๐’๐ฐ๐š๐ซ๐ฆ ๐จ๐Ÿ ๐€๐ ๐ž๐ง๐ญ๐ฌ ๐ฐ๐ข๐ญ๐ก ๐‹๐ฅ๐š๐ฆ๐š 3.2, ๐†๐๐“-4๐จ ๐ฆ๐ข๐ง๐ข ๐š๐ง๐ ๐‚๐ฅ๐š๐ฎ๐๐ž 3.5 ๐’๐จ๐ง๐ง๐ž๐ญ

๐“๐‹;๐ƒ๐‘: I reimplemented the Swarm concept using Haystack, but made it work with both open and proprietary models ๐Ÿ’ซ

โœ๏ธ blog article: https://haystack.deepset.ai/blog/swarm-of-agents
๐Ÿ““ notebook: https://haystack.deepset.ai/cookbook/swarm


Some time ago OpenAI published Swarm: an educational framework for building multi-agent systems.

Their approach focuses on two main concepts:
ใƒป ๐‘๐จ๐ฎ๐ญ๐ข๐ง๐ž๐ฌ: Each agent follows specific ๐Ÿ“œ instructions and uses ๐Ÿ› ๏ธ tools to execute them.
ใƒป ๐‡๐š๐ง๐๐จ๐Ÿ๐Ÿ๐ฌ ๐Ÿค: Agents can transfer control to one another using tool/function calling.


When I first read these ideas, I thought: ๐˜ด๐˜ช๐˜ฎ๐˜ฑ๐˜ญ๐˜ฆ ๐˜ฃ๐˜ถ๐˜ต ๐˜ฑ๐˜ฐ๐˜ธ๐˜ฆ๐˜ณ๐˜ง๐˜ถ๐˜ญ! And they pair well with the recent unified tool support in Haystack.

๐Ÿง‘โ€๐Ÿ’ป So, I decided to re-implement these concepts using Haystack, and in just a few lines of code, I had a working prototype.

๐Ÿ†’ Bonus feature: this implementation isn't tied to a single model provider - different agents can be powered by different models!

I replicated the ACME customer service example from the original article, with 3 Agents:
๐Ÿ Triage Agent - Llama 3.2 running on Ollama
๐Ÿ Sales Agent - Anthropic Claude 3.5 Sonnet
๐Ÿ Issues and Repairs Agent - OpenAI GPT-4o mini


Want to see the full implementation and give it a try? Check out the blog post and notebook! โœจ
In this post