NVIDIA and Red Hat host 33B-cap Small Models Hackathon in Raleigh, vLLM adds a bounty

vllm_project · x · 2026-09-11

NVIDIA and Red Hat AI are hosting an in-person Small Models Hack in Raleigh on October 17-18, capping total open-model parameters at 33B. Three tracks: fine-tune a small model to beat a general one on a real task, wire small models into a multi-agent system, and make one model call real tools reliably. vLLM joined with a Best Use of vLLM bounty. Participants get Brev GPU credits; prizes include an RTX 5090 and Jetson Orin Nano kits. 100 spots, solo or teams of 3.

Original post →

More from coding & agent

coding & agent channel →