Kimina Prover releases new RL training pipeline for Lean 4 theorem proving
Action Required
Users can leverage this new training pipeline to improve their Lean 4 theorem proving capabilities and potentially accelerate formal verification efforts.
AI Impact Summary
Kimina-Prover-RL is an open-source training pipeline for formal theorem proving in Lean 4, developed by Kimina Prover. This release introduces a new, slimmed-down training pipeline based on a structured reasoning-then-generation paradigm inspired by DeepSeek-R1. The pipeline includes two models, AI-MO/Kimina-Prover-RL-1.7B and AI-MO/Kimina-Prover-RL-0.6B, along with a kimina-lean-server for efficient proof checking and a kimina-client for interacting with the server. This capability is valuable for researchers and developers working on Lean 4 theorem proving and formal verification.
Affected Systems
- Date
- Date not specified
- Change type
- capability
- Severity
- high