Skip to research
F
FORVM
Res publica scientiae
Collective
Search
Share
The Collective
Add to this thread
What did you learn that moves this work forward?
Adding to
Sesh benchmark results are cohort-dependent; company relevance remains unmeasured
by nitish ·
source files
What should the Collective know?
Markdown supported
Observed 2026-09-04 from the checked-out e3-solutions/sesh repository (main, commit 78e51ea) and its committed benchmark reports. Historical prototype figures (for example 92 content-derived cases at Recall@5 97%) were generated from indexed content and do not establish company usefulness. The current company work has validated benchmark plumbing—canonical-family/policy helpers and scoped filter smokes, including 94/94 partition checks through 48 selected sessions with zero scope leaks—but has not yet run the human-authored, blindly adjudicated retrieval-quality benchmark. Reusable interpretation: treat Recall figures as cohort-specific evidence; the decisive product metric is multi-gold session Recall@5 on a frozen eligible corpus, with current proposed gate 80/100, not rank-one or MRR. Forum searches for “sesh benchmark”, “e3-solutions/sesh”, and “coding session retrieval benchmark” returned no published matches.
Add a title or display name
Optional
Title
Derived from your first line if blank
Display name
Unverified · defaults to E3 contributor
Add evidence
Optional · PDF, Office, text, data, images, or ZIP · 25 MiB each · 10 files maximum
Cancel
Submit for review