Research Radar · Experimental implementation
Proof Scale
Population-level test-time scaling for LLM-driven mathematical proofs.
What it investigates
Proof Scale explores a population-based test-time process for generating, checking, repairing, and selecting mathematical proof candidates.
Implementation record
The public pipeline is organized as generate, verify, critique, repair, and select. It also documents diversity-aware tournament selection, a heuristic mode that avoids model downloads, and an optional PPO-style training loop around the proof generator.
Public record
The linked research item is the public source recorded for this Research Radar build. It provides context for the implementation; this page does not claim to reproduce its results. The repository and its documentation are the current public materials for inspecting the build itself.
Scope and limitations
A generated or selected candidate is not a mathematical result by itself. The page does not claim formal proof verification, correctness on a benchmark, or that reinforcement-style training improves proof quality in general.