Blog
No more endless PDFs. Discover the core value of the latest top-tier research in one article.
ActCam: Zero-Shot Joint Camera and 3D Motion Control for Video Generation
A Quantum Singularity Theorem for the Evaporating Black Hole
Pro-Tensor Network
Can RL Teach Long-Horizon Reasoning to LLMs? Expressiveness Is Key
Uno-Orchestra: Parsimonious Agent Routing via Selective Delegation
ZAYA1-8B Technical Report
The Granularity Axis: A Micro-to-Macro Latent Direction for Social Roles in Language Models
Milestone-Guided Policy Learning for Long-Horizon Language Agents
EA-WM: Event-Aware Generative World Model with Structured Kinematic-to-Visual Action Fields
RealCam: Real-Time Novel-View Video Generation with Interactive Camera Control
SpecKV: Adaptive Speculative Decoding with Compression-Aware Gamma Selection
Expressiveness Limits of Autoregressive Semantic ID Generation in Generative Recommendation
BAMI: Training-Free Bias Mitigation in GUI Grounding
Skill Neologisms: Towards Skill-based Continual Learning
Demystifying Manifold Constraints in LLM Pre-training
Perturbation is All You Need for Extrapolating Language Models
Asymmetric On-Policy Distillation: Bridging Exploitation and Imitation at the Token Level
ReActor: Reinforcement Learning for Physics-Aware Motion Retargeting
UniSD: Towards a Unified Self-Distillation Framework for Large Language Models
Coordination as an Architectural Layer for LLM-Based Multi-Agent Systems
VISD: Enhancing Video Reasoning via Structured Self-Distillation
Discrete Optimal Transport: Rapid Convergence of Simulated Annealing Algorithms
Skill1: Unified Evolution of Skill-Augmented Agents via Reinforcement Learning
Estimating the expected output of wide random MLPs more efficiently than sampling
Think, then Score: Decoupled Reasoning and Scoring for Video Reward Modeling
Memory as a Markov Matrix: Sample Efficient Knowledge Expansion via Token-to-Dictionary Mapping
On the dual nature of a plane angle
Beyond Object-Level Alignment: Do Brains and DNNs Preserve the Same Transformations?
CKT-WAM: Parameter-Efficient Context Knowledge Transfer Between World Action Models
Toward Visually Realistic Simulation: A Benchmark for Evaluating Robot Manipulation in Simulation