Repository navigation
golden: freeze the IMPROVE neoepitope example outputs - #4
Merged
Merged
Conversation
The IMPROVE example landed in #1 with scripts only; no run output was committed. This freezes a B=2000 run built from the public IMPROVE_paper release (commit c670942, results.zip sha256 2689700...) as a golden baseline, adds the opt-in regression test matching the GUSTO and PCam ones, and publishes the report under docs/reports/. A second independent run reproduced every artefact byte-for-byte (findings.json modulo provenance.git_commit). Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E5Q2q7a7gF1tUX8y6fRV7m
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Requested by Precious · project thread
Before: the IMPROVE neoepitope example (#1) had scripts only. Unlike GUSTO and PCam, it had no committed run output, no report and no regression baseline, so nobody could see what predval found on it without rebuilding everything.
After:
tests/golden/improve/freezes a full B=2000 run,docs/reports/improve.htmlpublishes the report, andtests/test_golden_improve.pychecks a fresh run against the frozen files. Like the other two golden tests, it is opt-in withPREDVAL_RUN_GOLDEN=1.How: the fixture was built from the public SRHgroup/IMPROVE_paper release, pinned to commit
c670942and to theresults.zipsha256 recorded inGOLDEN.md. A second independent run passed the new golden test, so every output reproduced byte-for-byte (findings.jsonexcludes its git-commit field). Nosrc/changes. Ruff and the default pytest suite pass locally.Headline numbers pinned (detail in
tests/golden/improve/GOLDEN.md):Interaction with the coverage-delta fix (#3 /
claude/project-thread-n263n3): IMPROVE has 100% coverage for every model, so every finitecov_deltais 0. Both fixes only change how missing or invented cells are handled, and neither touches the parquet artefacts. The golden files are therefore unaffected by either fix.🤖 Generated with Claude Code
https://claude.ai/code/session_01E5Q2q7a7gF1tUX8y6fRV7m
Generated by Claude Code