Skip to content

Instantly share code, notes, and snippets.

@ruvnet

ruvnet/gist479-2.md

Created June 17, 2026 22:10
Show Gist options

  • Select an option

    Learn more about clone URLs
  • Save ruvnet/813e0aeca7dc7f3a0b2e84fa73180ee6 to your computer and use it in GitHub Desktop.

Select an option

Learn more about clone URLs
Save ruvnet/813e0aeca7dc7f3a0b2e84fa73180ee6 to your computer and use it in GitHub Desktop.
Download ZIP
Productionizing ruvector gnn-rerank (#479) — Step 2: CI-guarded recall test
Raw
gist479-2.md

Productionizing ruvector gnn-rerank (#479) — Step 2: CI-guarded recall test

Turned the research win into a regression guard. A deterministic integration test reproduces the exact regime (N=5000, D=128, noise σ=0.40, seed=42) through the public reranker API and asserts GnnDiffusion beats the no-rerank baseline.

recall@10:  noisy=0.280   gnn=0.384   delta=+0.104   (matches #479 exactly)
test result: ok. 1 passed

Why it matters: the headline "+10.4pp" now lives in cargo test (CI), with a +0.03 floor so it can't silently regress — and it runs without the standalone benchmark binary (which endpoint security blocks on the dev box). Branch: feat/productionize-gnn-rerank.

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