tobjend b92b7653e2 feat: iDRegEx refinement for CRX flat bags (Round 16) ... Heuristic: only run iDRegEx when n_methods ≤ 10 AND CRX grammar has >50% top-level optional parts (flat chain signal). If iDRegEx grammar is >10x tighter by lang_size, use it. Otherwise keep CRX. RAGSAK result: agents/capability (5 methods) refined from slot?.(defaultCapabilityId+summarize)?... (lang_size=1432) to (defaultCapabilityId|summarize) (lang_size=3) — 477x tighter. Speed cost: ~0.7s per candidate, negligible on 74s pipeline. CLI: --idregex-refine flag (default off). Also adds _count_optionals() and _should_try_idregex() helpers with 8 pytest tests. 234 tests pass.		2026-07-12 16:54:40 +02:00
..
results	feat: grammar_structure_score + min_structure filter	2026-07-12 03:02:50 +02:00
coarsen_eval.py	feat: 4-codebase evaluation — conventions vs completions tradeoff	2026-07-12 01:14:09 +02:00
context_eval.py	feat: implement Reduce algorithm (Algorithm 4, TODS 2010)	2026-07-12 00:11:38 +02:00
EXPERIMENT_LOG.md	feat: iDRegEx refinement for CRX flat bags (Round 16)	2026-07-12 16:54:40 +02:00
freq_eval.py	feat: frequency threshold sweep — 0.01-0.20 across 4 codebases	2026-07-12 01:26:45 +02:00
gbnf_eval.py	perf: iDRegEx opt-in, GBNF newline fix, OverflowError fix	2026-07-12 02:52:05 +02:00
RESULTS.md	docs: experiment log + cross-package analysis	2026-07-12 00:45:21 +02:00