Skip to content

FIX-0-8: fix group-balanced subsampling in the DFR baseline #401

Description

@mateuszwalo

Track: research · Level: core · Effort: ~6h · Phase: 0 · Depends on: none
Source: T20 findings III.6 (#389)

Why this matters

Our DFR-style baseline reaches 83.4 worst-group accuracy against 92.9 published. The findings
identify four reasons, one of which is an implementation defect in the group-balanced
subsampling, the other three being no l1 regularisation, no averaging over retrains, and a
weaker base extractor. A baseline that is wrong for an avoidable reason makes every
comparison against it softer than it should be.

Steps

  1. Fix the group-balanced subsampling in benchmarks/spurious_repair.py.
  2. State the three remaining known deviations from published DFR in the docstring, rather than
    leaving them implicit.
  3. Re-run the dfr condition and update findings_waterbirds.md.

Done when

The DFR arm is correct as far as group balancing goes, and every remaining deviation from the
published method is written down.


How to take this: comment "taking this" and wait to be assigned. Branch fix-0-8-dfr-group-balancing from upstream/main, and put Closes #<this issue number> in your PR. Full workflow: the Cohort Handbook (pinned in Discord).

Plan: plans/FIX-PLAN-T20-2026-08-23.md, section 4.

Activity

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

Metadata

Metadata

Assignees

No one assigned

    Labels

    P2Audit backlog: next cyclebugSomething isn't workingcohortBNNR 2026 cohort tasklevel:coreCore tasktrack:researchCohort research task

    Type

    No type

    Projects

    No projects

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions