Reference implementation for the Phoenix Architecture. Work in progress. aicoding.leaflet.pub/
ai coding crazy

experiment: DOC_FREQ_CUTOFF 0.4→0.5 score=0.8816

D-rate improved from 66%→64% across specs. Analytics spec notably
improved (57%→44%). Raising the cutoff makes more tags trivial,
so remaining shared tags carry higher signal for typed edge inference.

+1 -1
+1 -1
src/experiment-config.ts
··· 11 11 MIN_SHARED_TAGS: 2, 12 12 JACCARD_DEDUP_THRESHOLD: 0.7, 13 13 FINGERPRINT_PREFIX_COUNT: 8, 14 - DOC_FREQ_CUTOFF: 0.4, 14 + DOC_FREQ_CUTOFF: 0.5, 15 15 16 16 // ─── canonicalizer.ts — scoring weights ─────────────────────────────────── 17 17 CONSTRAINT_NEGATION_WEIGHT: 4,