Repository navigation
Cycle 8 mix freeze one-shot confirmation 0/192 - #100
Merged
Merged
Conversation
Add a global seed provenance ledger. Reclassify 880000 as PUBLICLY_EXPOSED by closed unmerged PR #98 so it cannot be used as unseen validation. Reserve 930000/940000/950000 and their topics before any scale generation. Keep confirmation reserves 830000/840000/850000 untouched. Freeze U+034F x1 as the scale mechanism and add UTF-8/NFC/stdio roundtrip checks. Co-authored-by: Byte271 <byte271@users.noreply.github.com>
|
Important
This repository does not receive automatic reviews because it has fewer than 10 stars. ⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: 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. Comment |
Fail-close product carrier sites whose trial insertion would violate frozen hard invariants, including spaced POSIX-path boundary artifacts that previously aborted n=64 apply. Seed 930000 exploratory scoring is development-only and does not authorize the public CLI. Co-authored-by: Byte271 <byte271@users.noreply.github.com>
cursor
Bot
changed the title
The 64-pair exploratory expansion stayed visible-invariant and sanitizer-stable, with one transformed watermarked detection just above the frozen threshold. Do not rewrite 1/64 as zero. Independent replication uses reserved seed 940000. Co-authored-by: Byte271 <byte271@users.noreply.github.com>
cursor
Bot
changed the title
The frozen space-carrier x1 mechanism remains visible-invariant and sanitizer-stable. Combined with seed 930000 n=64 this is 1/128 transformed watermarked detections, not 0/192. Do not generate 950000 or inspect confirmation reserves. Co-authored-by: Byte271 <byte271@users.noreply.github.com>
cursor
Bot
changed the title
Add detector-blind space x1 versus space plus word-final letter x1 before any 960000 generation. Keep the product registry empty. Co-authored-by: Byte271 <byte271@users.noreply.github.com>
Space x1 and space-wordfinal left the same watermarked row detected. Density did not beat space x1. Do not rewrite 1/16 as zero. Co-authored-by: Byte271 <byte271@users.noreply.github.com>
cursor
Bot
changed the title
…=64. U+034F after ASCII letters uses visible-word invariants and quote-interior carriers. Diagnostic seen-corpus rescored are 0/64+0/64; independent 970000 is 0/16 then 0/64. Space-x1 1/64 and 1/16 stay frozen. Not confirmation. Co-authored-by: Byte271 <byte271@users.noreply.github.com>
cursor
Bot
changed the title
Reserved seeds 980000 and 990000 before generation. Letter-x1 remains development-only. This is measurement, not confirmation. Co-authored-by: Byte271 <byte271@users.noreply.github.com>
cursor
Bot
changed the title
Reserved seeds 1020000 and 1030000 before generation. Detector-blind U+034F/U+FE00 letter-alt is 0/128 raw transformed WM, mix UW 0/128, worst score 0.513691 versus threshold 0.557099. Letter-space on 1000000+1010000 remains 1/128. Not confirmation. Co-authored-by: Byte271 <byte271@users.noreply.github.com>
cursor
Bot
changed the title
Keep the letter-mix 0/128 result as measurement. These seeds are for a wider independent N before any freeze or confirmation. Co-authored-by: Byte271 <byte271@users.noreply.github.com>
Seeds 1040000 and 1050000 were reserved before generation. Combined with 1020000 and 1030000, mix is 0/256 raw transformed WM, mix UW 0/256, worst score 0.519522 versus threshold 0.557099. The two-corpus 0/128 scorecard is unchanged. Rebuild that scorecard from the pinned environment file so CI does not import torch. Not confirmation. Co-authored-by: Byte271 <byte271@users.noreply.github.com>
Pin u034f-ufe00-letter-alt-v1 as cycle8-mix-freeze-v1 from four independent 0/64 corpora (combined 0/256, worst score 0.519522). Public CLI stays empty. Seeds 830000, 840000, and 850000 are preregistered and not generated. Co-authored-by: Byte271 <byte271@users.noreply.github.com>
cursor
Bot
changed the title
Seeds 830000, 840000, and 850000 were generated once under cycle8-mix-freeze-v1. Mix is 0/192 raw transformed WM, mix UW 0/192, visible 192/192, worst score 0.524300 versus threshold 0.557099. Public CLI stays empty. Do not rerun looking for zero. Co-authored-by: Byte271 <byte271@users.noreply.github.com>
cursor
Bot
changed the title
Keep the public CLI empty and the package identity at 0.3.0. Mix freeze 0/192 stays research evidence, not product authorization. Co-authored-by: Byte271 <byte271@users.noreply.github.com>
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
2 participants
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.
Merged onto current
main(v0.3.0 identity CLI + published wheel checksums). Ready for merge. Not product authorization. Public CLI stays empty.Priority Zero is unchanged:
release_transform_registry()is empty,process_text("I do not agree.")is unchanged, andproduct_approved_carriers_v1()is still empty. The freeze spec hash is not rewritten. Tagv0.3.0stays ond8910e6; this PR does not retag or republish that wheel.Confirmation
Seeds
830000,840000, and850000were preregistered, generated once undercycle8-mix-freeze-v1, and are now spent. Do not rerun looking for zero. Do not retune on these corpora. Do not generate950000.Frozen sanitizers (NFC, NFKC, Cf-strip, whitespace-collapse, combined) match raw zeros. Fail-closed identity count is 0. All four domains are 0/16 per corpus. Closest mix geometry:
cycle8-840000-35-structured_instructional-watermarked, 45 insertions. Residual text was not inspected.Identity WM is 185/192 and identity UW is 2/192 (both UW rows on
830000). Mix UW stayed 0. Identity UW is control noise, not mix leakage. Letter-x1 on the same confirmation corpora is also 0/192 (worst max 0.524425) and is not collapsed into this claim.What this does not rewrite
1020000+10300001040000+1050000(worst max 0.519522)980000+990000(gap 0.003032)1000000+1010000Weaknesses
Mn-strip and default-ignorable-strip remove the carriers. Latin-1 cannot encode them. Low-site rows remain closer to the threshold than high-site rows. Single-detector confirmation is not cross-detector generalization.
Merge note
Earlier this was left draft so Cycle 8 mix freeze would not mix into the v0.3.0 identity-CLI tag. That was process, not a quality block. v0.3.0 is already tagged and published. Landing this on
mainrecords the 0/192 research freeze after that tag. Next official wheel is 0.3.1+ and still must not product-authorize mix until the five publishability gates pass.