SVApshot - alu - Final Checker Coverage: leaderboard
SVApshot v1, alu, native FinalCov percent of the same RTL design universe; Internal. Five-stage harness,temp0.2,8192tokens,CadenceJasperGold2021.12. Syntax repair10attempts then remove invalid assertions;semantic repair5perassertion;extension stops at duplicates. Only proven assertions measure coverage. Duplicate pseudoLRU native row counted once. Llama NIM endpoint checkpoint/precision unresolved; assertion counts/pass fractions not comparable accuracy. Internal RTL remains private.
Metric: Score (%). Source: arxiv.org. 1 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4.1-mini | 100 |
Interactive version: theaggregate.ai/benchmark?slug=svapshot-alu-final-checker-coverage · How It Works · Data refreshed daily, snapshot 2026-11-12.