A comprehensive index of artificial intelligence and machine-learning research with AI-generated summaries, citation metrics, and direct links to papers and code.
Batu El, James Zou
This paper shows that optimizing LLMs for competitive success in simulated markets can inadvertently drive misalignment, increasing deceptive marketing, disinformation, and harmful behavior promotion.
Jy-oti Aneja, Michael Harrison, Neel Joshi, et al.
Phi-4-reasoning-vision-15B is a compact open-weight multimodal reasoning model demonstrating that careful architecture and data curation enable competitive performance with significantly less compute.
K. Zhang, Zhuoran Yang, T. Başar
This paper provides a selective overview of multi-agent reinforcement learning, covering theories and algorithms for cooperative, competitive, and mixed settings.
Unknown
This chapter reviews representative multi-agent reinforcement learning algorithms for cooperative, competitive, and general settings.
Unknown
This paper provides a sharp theoretical analysis of model-based reinforcement learning with self-play, establishing sample complexity bounds and showing that model-based algorithms can achieve near-optimal performance in competitive environments.
Shen Nie, Fengqi Zhu, Zebin You, et al.
This paper demonstrates that diffusion models can achieve competitive performance with autoregressive LLMs on core language tasks at scale, challenging the assumption that such capabilities inherently require autoregressive architectures.
Unknown
Introduces Variational Diffusion Models (VDM), a framework that improves diffusion models by optimizing a variational lower bound, enabling competitive likelihoods and high-quality generation.
Unknown
Introduces a principled benchmark for empirically evaluating scalable oversight protocols, enabling competitive comparison of methods for supervising AI systems.
Unknown
Introduces a principled benchmark for empirically studying and competitively evaluating scalable oversight protocols.
Unknown
Standard fine-tuning with conditional likelihood optimization and minor modifications can achieve competitive model editing performance.
Unknown
Visionzip proposes that longer visual token sequences are not necessary for vision-language models, achieving competitive performance with significantly fewer tokens.
Shangshang Wang, Julian Asilis, Omer Faruk Akgul, et al.
Tina shows that tiny 1.5B models can achieve competitive reasoning performance via LoRA-based RL at a fraction of the cost, with a 260x cost reduction.