'It just feels good when you solve the hard problems' - Harvard Gazette
Harvard students voluntarily take the extremely challenging Putnam math competition primarily for enjoyment, problem-solving, and camaraderie despite very low average scores.
DeepSeek's self-correcting AI model aces tough maths proofs
DeepSeekMath-V2 achieves near-perfect Putnam score (118/120) and matches IMO/China Olympiad gold levels by identifying and correcting its own mathematical reasoning errors.