Rewrite _EvalKing as a linear ring-scan and promote its terms to DNA-tunable constants; retune via mate-mining regression, ElasticNet/Ridge, and logistic-vs-material refits.

2026-09-15T20:31:36Z

Splits _EvalKing's own king-location table from CountKingSafetyDefects' search-gate copy so future eval retunes can't silently perturb extension/reduction gate frequency. Bumps KING_BACK_FLIGHT_BONUS (+1 -> +3) and reorders/nudges the ring-scan sub-terms (minor/pawn/xray/undefended) to match relative rankings that replicated across three independent calibration methods, while leaving KING_QUEEN_PROXIMITY_DANGER and KING_MISSING_ONE_CASTLE_OPTION untouched pending further discussion. Self-play (640 games) is roughly even (score 0.509, llr=+0.17, far from either SPRT bound) but consistently faster (nps candidate=1362140 vs head=1333682) at equal average depth -- landing as a wash/speed-neutral checkpoint, not a confirmed tactical improvement, for further tuning to build on. Co-Authored-By: Claude Sonnet 5

Add king-safety mate-mining regression pipeline; retune file-storm/

2026-09-14T06:18:07Z

queen-proximity king-safety weights and add a back-flight bonus from it. Mines real human-game checkmates and matched sound-draw controls (eval_tune/mine_king_mates.py, shard_mine.py, extract_king_features.py, match_controls.py, train_king_safety_model.py), then fits a logistic regression against _EvalKing's raw per-defect breakdown (engine-side support: chess.h's KING_DEFECT_DUMP_INFO, eval.c's KING_DEFECT_DUMP instrumentation, command.c's evalkingdump command, all gated behind the KING_DEFECT_DUMP=1 build flag) to see which defects actually predict getting mated. See eval_tune/king_safety_data/HANDOFF.md for the full methodology and findings. Two rounds of retuning landed from this data so far, both validated via sd10 curated-suite comparison plus per-position Crafty corroboration (tests/ECM.md) and gated self-play: - KingFileDefects/KingStormingPawnDefects (_ComputeFileStormDefects) bumped ~1.75-2x relative to the rest of the ring-scan defects -- the regression's most dominant, most stable tier-1 predictors, previously weighted the same as every other hand-picked defect. - KING_QUEEN_PROXIMITY_DANGER rescaled from -8/count to -32/count, and a new KING_BACK_FLIGHT_BONUS (+1/square) term added for empty, unattacked ring squares on the side of the king away from the attacker -- a real, non-collinear protective signal the current EmptyUnattackedSquare table didn't credit at all. Both calibrated by unstandardizing the regression's coefficients and anchoring against KING_MISSING_ONE_CASTLE_OPTION's already-live -23 value (same direct-EVAL_TERM unit family). Co-Authored-By: Claude Sonnet 5 Claude-Session: https://claude.ai/code/session_01P9EDZVcqu6E2bNRm4zDrTf

typhoon.git/src/eval_tune/extract_king_features.py, branch master

Rewrite _EvalKing as a linear ring-scan and promote its terms to DNA-tunable constants; retune via mate-mining regression, ElasticNet/Ridge, and logistic-vs-material refits.

Add king-safety mate-mining regression pipeline; retune file-storm/