lm-provers/QED-Nano-SFT
TEXT GENERATIONConcurrency Cost:1Model Size:4BQuant:BF16Ctx Length:32kPublished:Feb 12, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Warm

QED-Nano SFT by lm-provers is a 4 billion parameter mathematical reasoning model, fine-tuned on olympiad-level proof problems. This supervised fine-tuning (SFT) checkpoint is distilled from the 685B parameter DeepSeek-Math-V2, generating mathematical proofs with explicit chain-of-thought reasoning. It serves as the foundation for the full QED-Nano pipeline, excelling in theorem proving tasks.

Loading preview...