Reproduce “NoTB: Oracle-Free Triage of LLM-Generated RTL via Cross-Model Formal Consensus” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Elisavet Lydia Alvanaki, Je Yang, Biruk Seyoum, Luca P. Carloni
The authors introduce NoTB to triage model-generated RTL before a trusted testbench or golden design is available. It groups candidates only when sequential equivalence checking proves equivalence, then uses the number of model families in the largest group as an empirical correctness signal. On 70 evaluable CVDP specifications, four-family agreement achieved 94.7% precision at 27% coverage, but does not prove specification correctness.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.