project-numina/Kimina-Prover-Distill-1.7B
AI-MO/Kimina-Prover-Distill-1.7B is a 1.7 billion parameter theorem proving model developed by Project Numina and Kimi teams, distilled from the larger Kimina-Prover-72B. This model specializes in competition-style problem solving within the Lean 4 theorem prover environment. It achieves 72.95% accuracy with Pass@32 on the MiniF2F-test benchmark, making it suitable for automated theorem proving tasks.
Loading preview...
Overview
AI-MO/Kimina-Prover-Distill-1.7B is a specialized language model designed for automated theorem proving in Lean 4. Developed by Project Numina and Kimi teams, this 1.7 billion parameter model is a distilled version of the larger Kimina-Prover-72B, which was trained using large-scale reinforcement learning.
Key Capabilities
- Theorem Proving: Excels at solving competition-style mathematical problems within the Lean 4 formal verification system.
- High Accuracy: Achieves a notable 72.95% accuracy with Pass@32 on the MiniF2F-test benchmark, indicating strong performance in formal mathematics.
- Distilled Efficiency: Provides a more compact and efficient solution compared to its larger 72B parameter predecessor, making it practical for deployment.
Use Cases
- Automated Theorem Proving: Ideal for generating formal proofs and solutions for mathematical problems in Lean 4.
- Educational Tools: Can be integrated into systems for teaching or assisting with formal mathematics and logic.
- Research in Formal Verification: Useful for researchers exploring the capabilities of AI in formal methods and proof generation.
Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.