Skip to content

Corpus 100K: Coq — fix extractor (14 → 100K+ from existing CoqGym) #13

@hyperpolymath

Description

@hyperpolymath

Part of #12 campaign.

Observation: ECHIDNA has 235 MB of CoqGym corpus at
`external_corpora/CoqGym/` but `stats_coqgym.json` reports only
14 records. This is a ~4-order-of-magnitude shortfall vs the roughly
70 K proof states CoqGym is known to contain.

Action: audit `scripts/extract_coqgym.jl`; likely a regex /
gating bug. Rerun after fix and confirm ≥ 50 K records before
moving on.

Size estimate: CoqGym upstream ≈ 70 K, mathcomp adds ≈ 15 K,
Iris adds ≈ 10 K → realistic target 100 K+ without vendoring
anything new.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions