TY  - RPRT
TI  - Step-TP: A Grounded, Step-Level Dataset with Chain-of-Thought Reasoning for LLM-Guided Tensor Program Optimization
AU  - Mengfan Liu
AU  - Da Zheng
AU  - Junwei Su
AU  - Chuan Wu
PY  - 2026
UR  - https://arxiv.org/abs/2605.25954
ID  - 2605.25954
ER  -