Demonstrates advanced fine-tuning techniques for pushing foundation models to world-class performance in specialized domains.
NVIDIA's Nemotron model family achieved gold-medal level performance on both International Olympiad in Informatics and International Mathematical Olympiad through specialized recipes.
New simulation research evaluates budget phasing algorithms that address three core problems in Marketing Mix Modeling: multi-collinearity, selection bias, and non-identifiable adstock/saturation effects.
OpenAI published 722 math papers solving 90 top open problems using an internal AI model, including a breakthrough 'Quasi-Riemann Hypothesis' result.
Researchers present a deterministic polynomial-time algorithm achieving optimal O(√d) regret for online inverse linear optimization, solving an open problem.
Researchers developed an information-theoretic framework to quantitatively measure when specific features are generated during diffusion model denoising, revealing representation differences across pixel, SDVAE, VAVAE, and RAE models.
PINNs outperform classical finite difference methods in 5D quantum harmonic oscillator problems, reducing memory requirements from 1TB+ to practical levels while maintaining 0.1% accuracy.
