🎉BIG ANNOUNCEMENT Your Italian boy is now a Microsoft MVP🎉🎉🎉 #microsoft #community #mvpbuzz

9 new multilingual open LLMs released! Qwen 2.5 is the next iteration of Qwen 2 and comes with 2 new sizes and 5-70% better performance than Qwen2! Qwen 2.5 72B outperforms AI at Meta Llama 3.1 70B and matches 405B! Qwen 2.5 32B outperforms OpenAI GPT-4o mini. 🤯 TL;DR;

I'm thrilled to share I've joined OpenAI to work with Startups in EMEA. EMEA is home to world-class innovators, visionary startups & investors who back them. I am very excited to partner with them as they build the future with OpenAI And I get to do it with the best team ever

🎉SetFit v1.1.0 is out! Training efficient classifiers on CPU or GPU now uses the Sentence Transformers Trainer, and we resolved a lot of issues caused by updates of third-party libraries (like Transformers). Full release notes: github.com/huggingface/se… More details in 🧵

How can we build an Al Virtual Cell 🔮🧬 that simulates all functions and interactions of a cell? How will it transform research and drive breakthroughs in programmable biology, drug discovery and personalized medicine? 🚀 Take a look at our Perspective! arxiv.org/pdf/2409.11654

One of THE largest personas datasets dropped in Hugging Face 🤯 21 Million rows and 142GB 🤯 FinePersonas contains detailed personas for creating customized, realistic synthetic data. So now you can integrate unique persona traits into text generation apps. ------- What's a

Managing up as a PM is a weird skill to learn This list might be overkill, and having it doesn't magically make you do all this. But at the core, it's about getting out of your way, keeping momentum & learning to work with others.

OpenAI’s ability to shrink its reasoning model without losing many capabilities might be as big a deal as the reasoning power itself. theinformation.com/articles/opena…

Today, 30 years ago, Andrew Wiles announced the final insight to his proof of Fermat's Last Theorem! In this short note, I've tried to explain this monumental achievement in a simple way that everyone can understand. Dive in to discover how a 350-year-old mathematical mystery was

Aggregation & refinement improve LLM reasoning, but aggregation saturates, while refinement has 3 issues: 1) over-correction for easy problems 2) fails to localize+fix its own errors 3) insufficient number of refinement iteration for hard problems 🚨Multi-Agent, Iterative,

I've condensed my advice about how to think about LLM evals into this 15 min talk. I follow the same process on every project to improve AI products. We show how we use evals to create a data flywheel to move from demo to production-ready products.

Representation theory has become a pretty important part of mathematics, mostly when you consider its importance within the frame work of Quantum mechanics. Thankfully, Teleman's course notes on the subject are available for free in pdf format Check the comments for the link

updated figure

SEAL Leaderboard Update 🥇 Today, we're launching 2 new Agent leaderboards: - Agentic Tool Use (Chat) - Agentic Tool Use Enterprise) We are also adding OpenAI o1 to the Agentic Tool Use leaderboards, ranking at: #1 on Enterprise Tool Use #3 on Chat Tool Use 🧵(read on)

In the 1970s, scientists chopped up hydra (freshwater organisms) in a blender. After 3-5 days, the hydra cells "spontaneously" reaggregate, regrowing their tentacles & becoming "normal animals" again. Nobody asked, but this is one of my favorite research papers. PDF below.🔻

🚨Below is the open letter signed by Meta, Spotify, and many others, asking for regulatory certainty on AI in Europe, followed by my comments. [If you are interested in AI, data protection, and want to understand what this letter is really about, this long post is for you]

.OpenAI is hiring ML engineers for a new multi-agent research team! We view multi-agent as a path to even better AI reasoning. Prior multi-agent experience isn't needed. If you'd like to research this area with Kevin Stone and me fill out this form: jobs.ashbyhq.com/openai/form/oa…

At Databricks, we want to help customers build more #inference-friendly #llms. With MixAttention architecture, you can maintain model quality while improving inference speed and reducing memory footprint: databricks.com/blog/mixattent…

rocking my Nous swag at a Michelin star tonight