Skip to content

Cycle 7 Stage B: high-density durable catalog v3 - #96

Merged
byte271 merged 7 commits into
mainfrom
cursor/cycle7-stage-b-density-3fb1
Aug 25, 2026
Merged

byte271 merged 7 commits into
mainfrom
cursor/cycle7-stage-b-density-3fb1

Conversation

@byte271

@byte271 byte271 commented Aug 25, 2026 •

Copy link
Copy Markdown
Owner

Cycle 7 Stage B (catalog v3) and Stage C1 (catalog v4). Development evidence only. Not Cycle 7 formal confirmation.

Live state

  • PR Cycle 7 Stage A: collapse-resistant durable transforms #95 squash-merged. main HEAD is af341884fff211caeb24649a6ce79305c3491c6c.
  • This branch is based on that main. Cycle 6 remains frozen NONZERO_RESIDUAL 7/192.
  • Seeds 830000 / 840000 / 850000 were not inspected.
  • Seed 880000 is reserved unseen validation. It was not generated.

CI

The rebase SHA 05eab86 only received CodeRabbit because changing the PR base does not emit pull_request synchronize. Head is now 5d8140d so full Actions against main can run, including CI / Release Readiness.

Default pytest does not install transformers. GPT-2 tokenizer tests use pytest.importorskip("transformers").

Do not merge while required PR-head CI is red or still in progress.

Stage B (frozen catalog v3)

Seed 860000 then disjoint validation 820000. Catalog/scheduler were not changed after seeing 820000.

  • Density/geometry replicated at ~4 candidates/sample; collapsed intact/root ~0.72–0.74.
  • Detector reduction did not replicate.
  • Overall Stage B: INSUFFICIENT_EVIDENCE.

Stage C1 (catalog v4, seed 870000)

Metric Result
Mean durable candidates 4.75 (Stage C bar was ≥8)
Clause-newline mean 1.50
Quantifier-of mean 0.00
Collapsed intact / root 0.715
Spacing after collapse WM 4/4
Durable v4 after collapse WM 3/4
Combined after collapse WM 4/4
Transformed UW 0/4 all arms

Overall Stage C1: INSUFFICIENT_EVIDENCE as a Cycle 6 whitespace-collapse replacement.

Evidence: evidence/cycle7-stage-c-2026-08-25/. Writeup: docs/cycle7/stage-c-decision.md.

Open in Web Open in Cursor 

@chatgpt-codex-connector

Copy link
Copy Markdown

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.

@coderabbitai

coderabbitai Bot commented Aug 25, 2026 •

Copy link
Copy Markdown

Important

  • 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: d104422d-a734-47d4-bb28-1215839b4960


Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

Comment @coderabbitai help to get the list of available commands.

cursoragent and others added 6 commits August 25, 2026 21:23
Extend the seed ledger with exploratory seed 860000 before any
Stage B generation. Add collapse-resistant families for sentence
newlines, bounded optional that, discourse commas, prenominal
hyphens, parenthetical adverbs, and coordinating commas.

Co-authored-by: Byte271 <byte271@users.noreply.github.com>
Catalog v3 durable edits survive whitespace collapse and cut
collapse-surviving watermarked detections to 2/4 on the new
exploratory corpus. Freeze validation seed 820000 topic before
any disjoint look.

Co-authored-by: Byte271 <byte271@users.noreply.github.com>
Density and collapse-surviving geometry replicate catalog v3, but
the seed-860000 durable 2/4 detector reduction does not. Overall
Stage B remains INSUFFICIENT_EVIDENCE. Do not retune on 820000.

Co-authored-by: Byte271 <byte271@users.noreply.github.com>
Default CI pytest does not install transformers; the Stage B channel
tokenization check now importorsksips instead of failing collection.

Co-authored-by: Byte271 <byte271@users.noreply.github.com>
Clause-punctuation newlines and optional quantifier-of-determiner
pairs target higher natural site density. Ledger v3 reserves 880000
as unseen validation and leaves 830000/840000/850000 untouched.

Co-authored-by: Byte271 <byte271@users.noreply.github.com>
Catalog v4 raised mean durable density only to 4.75 sites/sample.
Clause newlines fired; quantifier-of did not. Collapse-surviving
intact windows stayed ~71.5%. Seed 880000 remains unseen.

Co-authored-by: Byte271 <byte271@users.noreply.github.com>
@cursor
cursor Bot force-pushed the cursor/cycle7-stage-b-density-3fb1 branch from 8be68eb to 05eab86 Compare August 25, 2026 21:23
@cursor
cursor Bot changed the base branch from research/cycle7-durable-transforms to main August 25, 2026 21:23
Retargeting #96 to main does not emit pull_request synchronize, so
the rebase SHA had only CodeRabbit. This push starts full CI.

Co-authored-by: Byte271 <byte271@users.noreply.github.com>
@byte271
byte271 merged commit d7eeb0f into main Aug 25, 2026
29 checks passed
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.

2 participants