Wrong Code, Right Structure: Learning Netlist Representations from Imperfect LLM-Generated RTL

Siyang Cai, Cangyuan Li, Haoyu Gao +3

Learning effective netlist representations is fundamentally constrained by the scarcity of labeled datasets, as real designs are protected by Intellectual Property (IP) and costly…