Hugging Face's logo
Stefano Fiorucci's picture
36 35 117

Stefano Fiorucci PRO

anakin87
juliiiiiiiiiius's profile picture lukecage's profile picture agaffu's profile picture
Ā·

AI & ML interests

Contributing to Haystack LLM framework šŸ—ļø. Language Models: orchestration, post-training, synthetic data...

Recent Activity

updated a collection 1 day ago
šŸ“ Cool LLM papers
liked a dataset 2 days ago
ruggsea/wsdm2024-cot-dataset
updated a collection 2 days ago
šŸ“ Cool LLM papers
View all activity

Articles

Organizations

deepset's profile picture Blog-explorers's profile picture ZeroGPU Explorers's profile picture Hugging Face Discord Community's profile picture

Posts 11

view post
Post
1539
Tulu 3 SFT Mixture by AllenAI is a massive, good, multilingual dataset for fine-tuning Language Models.

Unfortunately, it was missing the "language" column.

I added it using the good old fastText.

Check out the dataset here šŸ‘‰ anakin87/tulu-3-sft-mixture-with-language

view post
Post
368
šŸšŸšŸ š€ š’š°ššš«š¦ šØšŸ š€š šžš§š­š¬ š°š¢š­š” š‹š„ššš¦šš 3.2, š†šš“-4šØ š¦š¢š§š¢ ššš§š š‚š„ššš®ššž 3.5 š’šØš§š§šžš­

š“š‹;šƒš‘: I reimplemented the Swarm concept using Haystack, but made it work with both open and proprietary models šŸ’«

āœļø blog article: https://haystack.deepset.ai/blog/swarm-of-agents
šŸ““ 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:
惻 š‘šØš®š­š¢š§šžš¬: Each agent follows specific šŸ“œ instructions and uses šŸ› ļø tools to execute them.
惻 š‡ššš§ššØšŸšŸš¬ šŸ¤: Agents can transfer control to one another using tool/function calling.


When I first read these ideas, I thought: š˜“š˜Ŗš˜®š˜±š˜­š˜¦ š˜£š˜¶š˜µ š˜±š˜°š˜øš˜¦š˜³š˜§š˜¶š˜­! And they pair well with the recent unified tool support in Haystack.

šŸ§‘ā€šŸ’» So, I decided to re-implement these concepts using Haystack, and in just a few lines of code, I had a working prototype.

šŸ†’ 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:
šŸ Triage Agent - Llama 3.2 running on Ollama
šŸ Sales Agent - Anthropic Claude 3.5 Sonnet
šŸ 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! āœØ
View all posts

Collections 2

šŸ‡®šŸ‡¹ Italian Merges
I tried to merge two of the best Italian LLMs using Mergekit. The results are acceptable, but I could not improve on the best existing model.