HyperTree Proof Search introduces online self-improvement for neural theorem proving

The system learns from its own proof searches and uses a hyper-tree search algorithm inspired by Monte Carlo tree search and AlphaZero.

What happened?

HTPS plus online training reached 82.6% on held-out Metamath theorems and improved miniF2F-curriculum performance to 42%.

Why it matters

It foreshadowed the verifier-guided self-improvement loops that became common in later formal reasoning systems.

Significance rationale: HTPS coupled AlphaZero-inspired search with online learning and substantially advanced both Metamath and Lean theorem-proving performance.

Historical context

Primary evidence

Evidence & status at a glance

Significance
H2 · Field milestone
Verification
Paper released
Evidence level
E3 · Inspectable artifact
AI role
AI primary, human verified
Formal assurance
Machine-checked

How should these labels be read? →

Contribution & verification

Mathematical novelty
Benchmark result
Interface
Formal, Symbolic
AI contribution
The prover searches, learns from successful and failed search trajectories, and iteratively improves its policy.
Human contribution
Researchers designed HTPS, training environments and benchmark protocols.
Verification
Paper released · E3 · Inspectable artifact · Machine-checked
Last reviewed
2026-09-11

Technical details

Systems
htps
Organizations
meta-ai
Problems
metamath, minif2f
Methods
online-training, tree-search, expert-iteration

Sources & artifacts

Verification history

  • 2022-05-23 · Paper released · E2 · Research record

    The Chronicle records this event from the cited primary or official source; the status does not imply independent Chronicle verification.

  • 2026-09-11 · Paper released · E3 · Inspectable artifact

    This corpus refresh added a directly inspectable public artifact maintained by the authors or releasing organization. E3 records artifact-level inspectability and does not imply independent verification of the underlying scientific or mathematical claim.

Corrections

No corrections recorded.