Skip standalone inner-fold evaluation for default-only models - #58
Merged
Merged
Conversation
adrian-prior
marked this pull request as ready for review
September 23, 2026 09:03
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Generated by Codex
Standalone fitting skips inner-fold evaluation when only the default config is requested, avoiding unnecessary DFS computation. Benchmark runs retain all raw scores and timings; leaderboard reporting identifies default-only runs from the trial rows and excludes their inner fit time, including for existing CSVs. The runtime table is recalculated to exclude diagnostic inner evaluation and extra outer evaluation of an unselected default config.
The standalone path checks the search-space structure without sampling a config plan twice. CSV columns and raw baseline results are unchanged. The documentation distinguishes full-sweep runtime from the leaderboard's selected-trial training time.
Validation: 105 focused tests passed, 10 optional tests skipped; all pinned pre-commit hooks passed. Each model run's revised total was independently checked against the accounting rule. A separate check with the real TabPFN-Rel wrapper and DFS engine on a tiny database confirmed one training DFS computation, one prediction DFS computation, and no inner-fold work; only the final estimator was replaced with a dummy classifier.
Prepares patch releases
relarena-core 0.0.4andrelarena 0.1.2, including matching core pins and the workspace lockfile. Both wheels and source distributions build successfully and pass package-data checks.