CSPComing soon
Constraint-satisfaction puzzles at controlled size with guaranteed-unique solutions.
- Source
- Generated
- Score
- ModelAxes score, λ = 1
- Ladder
- structural · positions x attributes · 3×3, 4×3, 4×4, 5×4, 5×5, 6×6, 7×7, 8×8
- Items
- Not built yet
- Dimensions
- —
No results yet: this benchmark is planned and is not part of any score. Everything on this page describes the plan and may change before it ships.
The plan
Draft specWritten up in a draft specification; details may change once items are built and piloted.
What it will measure
Holding many interacting constraints at once and reaching the one assignment consistent with all of them.
How items will be built
- N positions (houses, seats, time slots) × M attributes. Clues are typed: in position k, same as, not the same as, next to, immediately left of, somewhere left of, at one of the ends.
- Clues are added until an exact solver finds exactly one solution, then removed wherever the solution stays unique, so every remaining clue is necessary. Clue order is shuffled.
- 8 rungs from 3 × 3 to 8 × 8, with more clue types as rungs rise. From rung 5, half the items use invented names so real-world associations can't help.
- The classic "Einstein" puzzle and 10 re-encodings of it are included as a memorization probe.
How it will be scored
- Exact full-solution match graded by code; +1 / 0 / −1. Cell accuracy is reported alongside.
- x50 over the size of the search space, reasoning lift, and accuracy split by whether pure propagation solves the puzzle or search is needed.
Example (burned: never used in a scored release)
Three houses in a row, numbered 1–3 from left to right. Colors: red, green, blue. Pets: cat, dog, fish. Drinks: tea, coffee, milk. 1. The cat lives in the blue house. 2. The green house is immediately to the right of the red house. 3. Tea is drunk in house 2. 4. The dog's owner drinks milk. 5. The fish lives in the green house.
Planned size1,200 items (8 rungs × 150).