Skip to content

test: reproduce intermediate-layer heuristic release - #145

Closed
ShiboSoftwareDev wants to merge 1 commit into
agent/fix-cross-layer-heuristic-detourfrom
agent/repro-intermediate-layer-heuristic-release
Closed

test: reproduce intermediate-layer heuristic release#145
ShiboSoftwareDev wants to merge 1 commit into
agent/fix-cross-layer-heuristic-detourfrom
agent/repro-intermediate-layer-heuristic-release

Conversation

@ShiboSoftwareDev

Copy link
Copy Markdown
Contributor

Stacked on #144. The existing fixture now models a six-layer route from z0 to z5. Its legal transition region exposes z0-z5, but also exposes dead-end ports on z1-z4 before the z5 exit. The current heuristic stops guiding after the first layer change, so the bounded search follows those intermediate-layer exits and fails. This PR changes only the existing solver fixture; GitHub Actions generates and verifies the SVG snapshot. No project command is run locally.

@vercel

vercel Bot commented Aug 2, 2026

Copy link
Copy Markdown

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated (UTC)
tiny-hypergraph Ready Ready Preview Aug 2, 2026 8:17pm

Request Review

@tscircuitbot

Copy link
Copy Markdown

Benchmark This PR

Run benchmarks by commenting on this PR:

/benchmark [benchmark.sh args...]
/benchmark --dataset <hg07|srj18> --limit <n> --solver <core|poly> --families <default|default+deep|all> --concurrency <n|auto>

Everything after /benchmark is forwarded directly to ./benchmark.sh.
The result comment shows the matching dataset from the latest stored main-branch benchmark artifact beside the PR run; main is not rerun for each PR benchmark request.

Examples:

  • /benchmark -> hg07, all samples, core solver
  • /benchmark --dataset srj18 --limit 20 -> first 20 srj18 samples
  • /benchmark --dataset srj18 --concurrency auto -> srj18 samples with runner CPU-count concurrency
  • /benchmark --sample sample002 -> one hg07 sample
  • /benchmark --limit 20 --solver poly -> first 20 hg07 samples with the poly solver
  • /benchmark --limit 40 --families default+deep -> first 40 hg07 samples with deeper candidate families

Any PR whose title contains [BENCHMARK TEST] will automatically run the benchmark workflow on PR updates.

@ShiboSoftwareDev

Copy link
Copy Markdown
Contributor Author

Superseded by #146 and its unchanged #143 reproduction. The hosted sample 4 diagnostic disproved intermediate-layer release as the primary stall: the search never reached the transition. The retained reproduction now matches the actual graph-guidance failure.

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