My Extraction Score Was 0.08 and the Model Was Innocent: Rebuilding the Ruler

Automated excerpt

Rejection corpora (nearmiss, adversarial/*) report false_accept_rate; silence and extraction corpora report acceptance_rate. Same model, same corpus, three comparators, three very different stories. The model extracts the right rule in different words. Is a right-when/wrong-do rule "extracted correctly"? References CauterRule v0.3.1 release notes v0.3.1 field test report (§8 extraction quality, Appendix A J6/J10/J13) Auto-generated per-corpus results Extraction accuracy tests Corpus format spec (expected_rule) User guide · Changelog CauterRule v0.3.1 is released.

Selected automatically from source text; not independently written or fact-checked. Read the original for full context.