Skip to content

feat: add SubPOP dataset adapter (sb-subpop)#10

Merged
openclaw-dv merged 1 commit into
mainfrom
merge/rust-subpop
Apr 11, 2026
Merged

feat: add SubPOP dataset adapter (sb-subpop)#10
openclaw-dv merged 1 commit into
mainfrom
merge/rust-subpop

Conversation

@openclaw-dv

Copy link
Copy Markdown
Collaborator

Summary

  • Adds SubPOP dataset adapter for subpopulation opinion benchmarking
  • Trivial conflict resolved in pyproject.toml and datasets/init.py (additive — both sb-gqa and sb-subpop add datasets)
  • Source: polecat/rust work on sb-subpop

Merge Queue

  • MR: sb-wisp-anz
  • Source issue: sb-subpop
  • Worker: rust
  • Rebase: conflict resolved (additive imports/deps from sb-gqa)

Test plan

  • All 138 tests pass locally
  • Conflicts resolved (trivial additive)
  • CI status checks

🤖 Generated with Claude Code

Add adapter for the SubPOP dataset (Suh et al., ACL 2025) from
HuggingFace (jjssuh/subpop). 3,362 questions across 22 US
subpopulations and 8 demographic attributes with pre-computed
per-group response distributions.

- SubPOPDataset with load(), info(), load_demographic_distributions()
- Lazy import of HuggingFace datasets library (optional dep)
- Population-level distributions via cross-group averaging
- Per-attribute demographic caching for P_sub evaluation
- Registered as "subpop" in DATASETS registry

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@openclaw-dv
openclaw-dv merged commit 6f918b2 into main Apr 11, 2026
6 checks passed
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.

1 participant