arXiv cs.AI
· Papers
Wrong Code, Right Structure: Learning Netlist Representations from Imperfect LLM-Generated RTL
arXiv:2603.09161v2 Announce Type: replace-cross Abstract: Learning effective netlist representations is fundamentally constrained by the scarcity of labeled datasets, as real designs are protected by Intellectual Property (IP) and costly to annotate. Existing work therefore focuses on small-scale circuits with clean la