X. Zhu, “QiMeng-Kernel: Macro-Thinking Micro-Coding Paradigm for LLM-Based High-Performance GPU Kernel Generation”, AAAI, vol. 40, no. 34, pp. 29168–29176, Mar. 2026.