feat: phase 4 — benchmarks, HPC, llms.txt, wiki-currency CI#3
Merged
Conversation
Phase 4 of #1 (final construction phase). Benchmarks: the harness re-ported (deferred from Phase 1 with human approval) + three inference prompt cards — teacher-basic-workflow (gaussian_x1 drift probe), assistant-medium-wrap- likelihood (chi-squared/positional-vector conventions, self-contained inline code), assistant-hard-sne-cosmology (autonomous Pantheon+ fit + evidence-based wCDM comparison; README-mirrored) — with the card/README parity, rubric-total and dataset-existence tests adapted and green (52/52 suite). HPC: sync CLI + conf examples, CPU/GPU SLURM array batch templates (generically named), and an autofit hpc/template.py preserving the parse_fit_args/--use_cpu/--number_of_cores interface the templates depend on; wiki operations/hpc.md updated + re-stamped. llms.txt front door (chat bootstrap, read order, domain-adaptation pitch) and the wiki-currency workflow ported (stack cut to autoconf+autofit; five legs: version/symbols/idioms/provenance/citations; workflow_call ready for PyAutoBuild release wiring, which files as its own follow-up). All legs green: 52/52 tests, audit 0 misses, 77 citations resolve, idioms clean, provenance --strict clean. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The wiki-currency CI run on PR #3 correctly failed its version-drift leg: the baseline pinned the local source stack (2026.7.6.649) while PyPI's latest release is 2026.7.9.1 (the armed nightly). The CI report showed no API-surface hash drift and every content leg green against the newer release, so this is the honest re-pin the af_audit_skill_apis workflow prescribes: baseline rewritten against the released wheel in a clean venv (ambient PYTHONPATH cleared — it had silently aliased the local source), version.txt bumped, page pinned_commits moved to the 2026.7.9.1 tag commits (content stamps remain valid — they hash the body, and the audit re-validated it against the release). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Jammy2211
changed the base branch from
feature/afa-phase-3-domain-adaptation
to
main
July 10, 2026 16:00
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Phase 4 of #1, stacked on the phases-1–3 PR (#2) — retarget to
mainafter #2 merges.What's in it
teacher-basic-workflow(gaussian_x1; the cheap drift probe),assistant-medium-wrap-likelihood(self-contained inline user code; measures the chi²→lnL conversion, validation discipline, and the noise-normalisation conversation),assistant-hard-sne-cosmology(autonomous Pantheon+ fit + evidence-based wCDM comparison; README-mirrored; rewards the correct negative result). Card↔README parity, rubric-total (=100) and dataset-existence tests adapted and green.hpc/syncCLI + conf examples, CPU/GPU SLURM array templates (generic naming, thread pinning,--use_cpuwiring), and an autofithpc/template.pypreserving the batch-interface contract;wiki/core/operations/hpc.mdupdated and re-stamped.pull_request+workflow_dispatch+workflow_call; PyAutoBuild release wiring files as its own follow-up prompt.Gates
--strict0/0.profile.mdreference only · leg 3 (wiki-currency) validated locally leg-by-leg; the workflow itself runs on this PR · leg 4 (chat-surface smoke) is post-publish and human-side (modes/maintainer.md).🤖 Generated with Claude Code