Aksel

elon musk saying “I didn’t want to start a company, I just couldn’t get a job” is kinda crazy

Here’s how you train an email agent from scratch with GRPO 👇 1️⃣ Nail a prompted baseline first. It flushes out tool bugs & gives you a benchmark to beat. 2️⃣ When the plateau hits, switch to RL. A 14B model jumped 40%→96% —beating o3 & Gemini—by laser-focusing on one job.

These insights from Manus are a must read if you build agents! - make use of the prompt cache (your agent will be 10x cheaper) - the filesystem = memory module - keep errors in the trace so the LLM can self correct Read this!

Thrilled to finally share what we've been working on for months at @huggingface 🤝@pollenrobotics Our first robot: Reachy Mini A dream come true: cute and low priced, hackable yet easy to use, powered by open-source and the infinite community. Tiny price, small size, huge…

We just released SmolLM3: strong 3B model for fast multilingual long-context reasoning Me and @AymericRoucher were asked to "make-it-agentic" so we cooked up some really nice datasets and training routines. They took the model to the Pareto frontier in function-calling