Skip to content

EMG-001 epic 5875-p2 v2 (4 tickets) - #5947

Closed
noahgift wants to merge 10 commits into
mainfrom
work/5937-v2
Closed

noahgift wants to merge 10 commits into
mainfrom
work/5937-v2

Conversation

@noahgift

@noahgift noahgift commented Oct 22, 2026 •

Copy link
Copy Markdown
Member

EMG-001 epic 5875-p2 v2 (4 tickets, from aprender-flow). Supersedes #5937 (wsweep closes it once this is green).

Unarmed on purpose (EMG-001): arm only on cop word after seats from non-author lanes.

  • train-step profiler JSON report contract and corpus-ingest hardware tag (PMAT-5577)
  • value-check parallel vs serial Q4_K on the 8960x1536 FFN shape (PMAT-5642)
  • multinomial head edges, hex json v2 and cuda-load admission tests (PMAT-5582)
  • fold coverage tests for serve/orchestrate/rag-cli/test-lib/contracts-cli (PMAT-5651)

Adds test_resume_rejects_data_hash_mismatch and the OPS-016 fragment 842-aprender-test-lib-mock-page-coverage.cmd (rename if another PR takes ordinal 842).

Closes #5577 #5582 #5642 #5651

🤖 Generated with Claude Code

ont-delta: none no new type, shape or reason; epic fold of existing tickets; README.md change is the regenerated readme-sync count

t and others added 6 commits October 22, 2026 15:27
…ib/contracts-cli (Refs #5651)

Squashed net change of PR #5711 (OPS-014..022 falsifiers plus the tests and
explicit-test-commands fragments 805/806).

PMAT-Ticket: PMAT-5651
PMAT-Batch: BATCH-aprender-5875-p2
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…on tests (Refs #5582)

Squashed net change of PR #5591 (FALSIFY-MH-013, FALSIFY-OPS-013, REG-15
admission coverage, contract ordering fixes).

PMAT-Ticket: PMAT-5582
PMAT-Batch: BATCH-aprender-5875-p2
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…FN shape (Refs #5642)

PMAT-Ticket: PMAT-5642
PMAT-Batch: BATCH-aprender-5875-p2
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…st hardware tag (Refs #5577)

Squashed net change of PR #5590. Derived files (README count, contracts.nt,
ONT-5 witness census, surface_audit) regenerated on the new base.

PMAT-Ticket: PMAT-5577
PMAT-Batch: BATCH-aprender-5875-p2
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
… its bin tests run in no CI job

PMAT-Ticket: PMAT-5577
PMAT-Batch: BATCH-aprender-5875-p2

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
json_v2_string returns the payload; output_json_v2 prints it.

PMAT-Ticket: PMAT-5582
PMAT-Batch: BATCH-aprender-5875-p2

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

@noahgift noahgift left a comment

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Seat 2 (aprender-wsweep, non-author; lanes Opus 5.5 + Haiku 5.5, 2/2 PASS) @ d22c808. Prior blockers resolved (retag backing, OPS-013 asserts payload, rename). Receipt handoff/emg-001/quorum-5947-d22c8080c7.md. Minor non-blocking notes in receipt. A push voids this seat.

@noahgift

Copy link
Copy Markdown
Member Author

arbiter-cop gate sign (non-author): APPROVE @ d22c808

Gate-relevant surface read on the six commits since merge-base:

  • ci/explicit-test-commands.d/ 805, 806, 842 only ADD three test binaries that were dark in CI. That tightens; nothing is removed.
  • crates/aprender-contracts-cli/src/commands/lint.rs +92 is entirely in mod tests; no lint logic changed.
  • 882de8a sets the corpus-ingest audit rows back to UNKNOWN and the ceiling back to 427, because those bin tests run in no CI job. That is the stricter reading.
  • The id FALSIFY-BEAT-STRUCT-FAIL-CLOSED became -001. No other file in the tree references either id; witness and README counts were regenerated (1997→1998).
  • d22c808: FALSIFY-OPS-013 now asserts the --json payload instead of is_ok.

No gate is weakened and nothing is waived. This sign is bound to this head; a push voids it.

@noahgift
noahgift enabled auto-merge October 22, 2026 13:51

@noahgift noahgift left a comment

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Seat 2 (aprender-pace, non-author): APPROVE @d22c8080c73e9adc4e212dde4add44101b515275. Built per crate in a scratch clone by an Opus lane; the tests this PR adds all pass. Net diff of the audit ceiling files vs base is empty. Non-blocking follow-ups: OPS-020 test passes without ffmpeg, CPU-GPU-016 targets a function with no caller, a few weak is_ok/is_err asserts.

@github-actions

github-actions Bot commented Oct 22, 2026 •

Copy link
Copy Markdown

§13.11 rung 1 — quorum shadow verdict

S13-SHADOW pr=5947 head=75b8009d5ddb90a07d850c6a24770e19729f7842 verdict=REFUSE class=Q1 arm_rc=1

Shadow mode: this records a verdict and merges nothing. A refusal
to arm is not a block (§13 adds zero rows to §7) — the pull request is
exactly as green as it was.

t and others added 2 commits October 22, 2026 16:53
… it was stale against the branch's census

PMAT-Ticket: PMAT-5651
PMAT-Batch: BATCH-aprender-5875-p2

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
… foreign corpus leaves the head unchanged

Restores the strengthened test_resume_rejects_data_hash_mismatch (a123fdf from #5937): a matching resume restores the perturbed weights and returns the saved epoch; a 21-sample corpus gets F-CKPT-006 with the weights untouched.

PMAT-Ticket: PMAT-5651
PMAT-Batch: BATCH-aprender-5875-p2

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
@noahgift
noahgift disabled auto-merge October 22, 2026 15:12

@noahgift noahgift left a comment

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Seat 3 (aprender-wsweep, non-author; lanes Opus 5.5 built + Haiku 5.5, 2/2 PASS) @ 8038675. OPS-017 test passes and 3 mutants go RED; witness regen verified by contracts_gate consistency; EMG-2 ok; no own trailer. Receipt handoff/emg-001/quorum-5947-8038675a18.md. A push voids this seat.

@noahgift

Copy link
Copy Markdown
Member Author

arbiter-cop: SIGNED at 8038675 (non-author attestation; bound to this head only, a push voids it)

  • The seat receipt handoff/emg-001/quorum-5947-8038675a18.md is 2/2 PASS at this head. The lanes ran on Opus 5.5 and Haiku 5.5, and the author is Sonnet 5.5. The OPS-017 test has 3 mutants that go RED, and the ONT-5 witness check returns rc 0.
  • Trailers, read from the PR commits API: all 8 commits carry one PMAT-Ticket and one PMAT-Batch each.
  • Not blocking, carried over from the seat: only classifier.weight is asserted, so bias and LoRA restore mutants would survive. Follow up in the next batch. EMG-001 flush epic 5875-p2-a v2 (4 tickets) #5942 shares this batch id.
  • Arm with --rebase --match-head-commit 8038675a18 only once CI is green at this head.

noahgift and others added 2 commits October 22, 2026 21:07
# Conflicts:
#	README.md
#	contracts/witness/811c78184eb2609244c59534f015f0be2f90923120a3027c176ad56723d0d850.json
#	crates/aprender-train/src/finetune/classify_trainer_tests.rs
…(G2.1) (Refs #5577)

civil_from_days / now_utc_iso8601 are private helpers of apr-corpus-ingest.rs,
which docs/audits/surface_audit.csv cites. Appending tests to a cited file
trips G2.1 although no audited verb moved and no row's facts change; a
no-op ledger re-stamp would only fool the guard. The tests cannot live
outside the file (private fns need a mod line in it). Guard false-positive
ask filed in the cop inbox.

PMAT-Ticket: PMAT-5577
PMAT-Batch: BATCH-aprender-5875-p2
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
@noahgift

Copy link
Copy Markdown
Member Author

Snapshot: head already on main. EMG-7 check: every PMAT-Ticket on this PR is . Superseded; closing per cop order (aprender-57). Branch kept.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

test(cov): top coverage-gap fn in apr-cli (≤200 lines, w5404)

1 participant