Kevin-32B – how to do multi-turn RL on writing CUDA kernels

Hey – we just published a blog post about Kevin-32B = K(ernel D)evin.It's to our knowledge the first open-source model that's RL-trained on CUDA kernels. Our goal was to demonstrate multi-turn RL using GRPO. We used 180 Python->CUDA conversion tasks from the KernelBench dataset. The results were surprisingly strong! We were able to outperform top reasoning model like o3 & o4-mini. We're sharing our training setup and learnings in the blogpost. Also the model is on HuggingFace:huggingface.co/cognition-ai/Kevin-32B