홈 › Papers

CUDA Agent: Large-Scale Agentic RL for High-Performance CUDA Kernel Generation

2026-02-27 · Weinan Dai, Hanlin Wu, Qiying Yu, Huan-ang Gao, Jiahao Li, Chengquan Jiang, Weiqiang Lou, Yufan Song, Hongli Yu, Jiaze Chen, Wei-Ying Ma, Ya-Qin Zhang, Jingjing Liu, Mingxuan Wang, Xin Liu, Hao Zhou arxiv

GPU kernel optimization is fundamental to modern deep learning but remains a highly specialized task requiring deep hardware expertise. Despite strong performance in general programming, large language models (LLMs) remain uncompetitive with compiler-based systems such as torch.compile for CUDA kernel generation. Existing CUDA code generation approaches either rely on training-free refinement or fine-tune models within fixed multi-turn execution-feedback loops, but both paradigms fail to fundamentally improve the model's intrinsic CUDA optimization ability, resulting in limited performance gains. We present CUDA Agent, a large-scale agentic reinforcement learning system that develops CUDA kernel expertise through three components: a scalable data synthesis pipeline, a skill-augmented CUDA development environment with automated verification and profiling to provide reliable reward signals, and reinforcement learning algorithmic techniques enabling stable training. CUDA Agent achieves state-of-the-art results on KernelBench, delivering 100\%, 100\%, and 92\% faster rate over torch.compile on KernelBench Level-1, Level-2, and Level-3 splits, outperforming the strongest proprietary models such as Claude Opus 4.5 and Gemini 3 Pro by about 40\% on the hardest Level-3 setting.

📄 PDF Abstract BibTeX arXiv:2602.24286

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning Code Generation

KernelBlaster: Continual Cross-Task CUDA Optimization via Memory-Augmented In-Context Reinforcement Learning

2026-02-15 · Kris Shengjun Dong, Sahil Modi, Dima Nikiforov, Sana Damani 외 arxiv

Optimizing CUDA code across multiple generations of GPU architectures is challenging, as achieving peak performance requires an extensive exploration of an increasingly complex, hardware-specific optimization space. Trad…

Reinforcement Learning

CudaForge: An Agent Framework with Hardware Feedback for CUDA Kernel Optimization

2025-10-23 · Zijian Zhang, Rong Wang, Shiyang Li, Yuebo Luo 외 arxiv

Developing efficient CUDA kernels is increasingly critical for AI applications such as large-scale LLM training. However, manual kernel design is both costly and time-consuming, motivating automatic approaches that lever…

Code Generation

Towards Robust Agentic CUDA Kernel Benchmarking, Verification, and Optimization

2025-09-16 · Robert Tjarko Lange, Qi Sun, Aaditya Prasad, Maxence Faldor 외 arxiv

Recent advances in large language models (LLMs) demonstrate their effectiveness in scaling test-time compute for software engineering tasks. However, these approaches often focus on high-level solutions, with limited att…

From Tokens to Regions: CUDA-Sensitive Instruction Tuning for GPU Kernel Generation

2026-06-15 · Wentao Chen, Jiace Zhu, Xing Zhe Chai, Zeng Qu 외 arxiv

High-performance CUDA kernels are essential for scalable AI systems, while Large Language Models (LLMs) still struggle to generate correct kernels due to strict and implicit execution constraints. Existing LLM-based appr…

StitchCUDA: An Automated Multi-Agents End-to-End GPU Programing Framework with Rubric-based Agentic Reinforcement Learning

2026-03-03 · Shiyang Li, Zijian Zhang, Winson Chen, Yuebo Luo 외 arxiv

Modern machine learning (ML) workloads increasingly rely on GPUs, yet achieving high end-to-end performance remains challenging due to dependencies on both GPU kernel efficiency and host-side settings. Although LLM-based…