Shengtao Zhang← All publications

ICML · 2026

Towards Cold-Start Drafting and Continual Refining: A Value-Driven Memory Approach with Application to NPU Kernel Synthesis

Yujie Zheng, Zhuo Li, Shengtao Zhang, Jiaqian Wang, Junjie Sheng, Junchi Yan, Weinan Zhang, Ying Wen, Bo Tang, Muning Wen

AgentsRLMemorySystems

Abstract

EvoKernel

Published at ICML 2026, EvoKernel addresses cold-start code generation in data-scarce accelerator ecosystems. It casts NPU kernel synthesis as a memory-based reinforcement-learning process, learns stage-specific experience values for initial drafting and later latency refinement, and shares useful experience across tasks. The resulting agent accumulates practical optimization knowledge online instead of relying on expensive domain-specific fine-tuning.

Value-guided experience reuse turns sparse NPU feedback into a continual drafting-and-refinement loop.