tri competitors audit: the table, audited
168 records: 14 papers entered twice, 141 cite no score, 144 state zero at pass@1.
t27 · tri competitors audit
prints as Alpha-RTL
arXiv:2606.05680 cass_rtl_competitor + cass_rtl_competitor_2 + cass_rtl_competitor_3
prints as CASS-RTL
arXiv:2606.06530 rtlscout_competitor + rtlscout_competitor_2 + rtlscout_competitor_3
prints as RTLScout
arXiv:2606.08976 rtlbenchls_competitor + rtlbenchls_competitor_2
prints as RTL-BenchLS
arXiv:2606.09867 estrtl_competitor + estrtl_competitor_2 + estrtl_competitor_3
prints as EstRTL
arXiv:2606.10285 openrtlset_competitor + openrtlset_competitor_2
prints as OpenRTLSet
arXiv:2606.13706 hiersva_competitor + hiersva_competitor_update
prints as HierSVA
arXiv:2606.15500 llm4rtl_2026_competitor + llm4rtl_2026_competitor_2 + llm4rtl_competitor + llm4rtl
_competitor_2
prints as LLM4RTL / LLM4RTL-2026
A paper is attributed to a record only from the contiguous run of `///`
lines directly above its `pub fn`. A wider window reads the previous
record's citation as this one's: that window reported 16 papers here,
and ten of them were the neighbour's.
`pass_at_1: 0.0` is the value a record takes when there is no published
score to cite -- the struct has no other way to say it. It is in range and
it is what `compare_with_competitor` subtracts, so every one of those 144
records yields our own score as the margin over it at pass@1. Read the
two counts apart: 141 records cite nothing at any metric, and 3 state zero
at pass@1 while citing a nonzero score at some OTHER k -- true citations
that still read as zero to the default comparison.
t27 $ recorded 2026-10-03 16:18 UTC real 2.6 s shown 2.6 s exit codes 0
$ tri competitors audit
Staged: the prompt and the typing. Real: every byte the commands printed, at the time they printed it. Any silence longer than 2 s is shown for 2 s, and the title bar says so while it happens. Edited: home directory shown as ~.
Trinity S³AI
