Nemotron 3 Super is here — 120B total / 12B active, Hybrid SSM Latent MoE, designed for Blackwell.
Truly open: permissive license, open data, open training infra. See analysis on @ArtificialAnlys
Details in thread 🧵below:
Check out our latest work on reinforcement learning for generative Best-of-N selection! 🧪
TLDR; we use targeted RL and synthesized math and coding data to outperform majority-voting and prompting baselines of larger models at selection.
Paper ✍: https://t.co/KB3iIqs8Io
Thank you @SchmidhuberAI for speaking in front of a packed room at ZurichAI in the @ETH_AI_Center yesterday! It's the biggest event so far, by far.
Thanks everyone for coming; we're sorry for anyone who couldn't didn't get a spot. More and bigger things are planned!
Our team has just achieved gold medal at the International Olympiad in Informatics (IOI) with open-weight models for the first time! 🏅
This marks a pivotal step in bridging the gap between open and closed-source AI, see @mehrzadsamadi's post for more details. 💥
(1/5) 🚀 For the first time, an open-weight model has achieved a gold medal at the International Olympiad in Informatics (IOI). We did it with gpt-oss-120b and our new framework, GenCluster!