release_tools/order5_deep.json

{
 "sampled_pairs": 600,
 "phase1_fin<=6": {
  "ce_at_fin<=6": 402,
  "min_ce_distribution": {
   "2": 348,
   "3": 40,
   "4": 8,
   "5": 6
  },
  "no_ce_at_fin<=6": 198
 },
 "phase2_sat_spotcheck_of_no_ce6": {
  "checked": 40,
  "fin7": 0,
  "fin8": 0,
  "none_le8": 40,
  "examples": []
 },
 "interpretation": "Of 600 random order-5 pairs, 402 are FALSE with a CE at Fin<=6 (in-solver reachable). 198 had no CE<=6; a SAT spot-check of 40 found 0 at Fin7 + 0 at Fin8 (the in-solver-MISS tail) and 40 with no CE<=8 (TRUE/huge). Low fin7/8 counts => the flagship FALSE engine (Fin<=6) covers order-5 OOD FALSE."
}