Annmaria Antony(@AnnmariaKAntony):LLMs are good at CUDA because the internet is full of it. But a model that gives you highly optimized CUDA may still struggle to write compilable HIP. We built a synthetic data pipeline with multi-agent search and post-trained a 14B open-source model with SFT + GRPO RL, leading to substantially better HIP compilation + correctness rates on AMD MI350X GPUs.

Annmaria Antony

@AnnmariaKAntony

CS/EE @stanford, prev @nvidia

参加 February 2025

192 フォロー中 289 ファン

Annmaria Antony@AnnmariaKAntony

2026.07.02 21:38

LLMs are good at CUDA because the internet is full of it. But a model that gives you highly optimized CUDA may still struggle to write compilable HIP. We built a synthetic data pipeline with multi-agent search and post-trained a 14B open-source model with SFT + GRPO RL, leading to substantially better HIP compilation + correctness rates on AMD MI350X GPUs.