Skip to content

Expand the bilingual synthetic benchmark (eval/dataset) #5

Description

@fishonbike

Grow eval/dataset/docs.jsonl with harder, realistic-but-synthetic cases: more quasi-identifiers, nicknames, transliterations, regional terms, code-switching. Keep all data fabricated (.example domains, invalid IDs). PRs welcome.

Metadata

Metadata

Assignees

No one assigned

    Labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions