AI研究論文デイリー

MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmark

Show, Don't Tell: Aligning Language Models with Demonstrated Feedback

Learning Temporally Consistent Video Depth from Video Diffusion Priors

Artificial Generational Intelligence: Cultural Accumulation in Reinforcement Learning

μLO: Compute-Efficient Meta-Generalization of Learned Optimizers

ZeroSmooth: Training-free Diffuser Adaptation for High Frame Rate Video Generation