works on
reinforcement learning 2algorithm optimization 1approximate nearest neighbor search 1code generation 1contrastive learning 1cuda optimization 1gpu performance 1large language models 1retrieval-augmented generation 1

From the 2 of 9 linked papers with an AI index.

activity
20242026
most citedCUDA-L1: Improving CUDA Optimization via Contrastive Reinforcement Learning

1 citations · 1 across the 3 of their papers we have counts for

collaborators
Showing cs.AIShow all

1 paper · 1 filter

cs.AI2026★ 1 cited

CUDA-L1: Improving CUDA Optimization via Contrastive Reinforcement Learning

Xiaoya Li, Albert Wang, Guoyin Wang +2

The paper presents CUDA-L1, a reinforcement learning system that automatically optimizes CUDA code using a contrastive RL algorithm, achieving large speedups on GPU kernels without…