Blog
告别冗长 PDF，一文读懂最新顶会与核心期刊的创新价值。
DeepScan: A Training-Free Framework for Visually Grounded Reasoning in Large Vision-Language Models
Introduction to holography
UltraDexGrasp: Learning Universal Dexterous Grasping for Bimanual Robots with Synthetic Data
Conditioned Activation Transport for T2I Safety Steering
Pointer-CAD: Unifying B-Rep and Command Sequences via Pointer-based Edges & Faces Selection
cuRoboV2: Dynamics-Aware Motion Generation with Depth-Fused Distance Fields for High-DoF Robots
Latent Particle World Models: Self-supervised Object-centric Stochastic Dynamics Modeling
MAGE: Meta-Reinforcement Learning for Language Agents toward Strategic Exploration and Exploitation
Observing and Controlling Features in Vision-Language-Action Models
On-Policy Self-Distillation for Reasoning Compression
DAGE: Dual-Stream Architecture for Efficient and Fine-Grained Geometry Estimation
Scaling Laws for Reranking in Information Retrieval
RoboMME: Benchmarking and Understanding Memory for Robotic Generalist Policies
BandPO: Bridging Trust Regions and Ratio Clipping via Probability-Aware Bounds for LLM Reinforcement Learning
MoRe: Motion-aware Feed-forward 4D Reconstruction Transformer
The Spike, the Sparse and the Sink: Anatomy of Massive Activations and Attention Sinks
Replaying pre-training data improves fine-tuning
Quantum field theories with many fields
Generalizing Fair Top-$k$ Selection: An Integrative Approach
Reasoning Theater: Disentangling Model Beliefs from Chain-of-Thought
PhysiFlow: Physics-Aware Humanoid Whole-Body VLA via Multi-Brain Latent Flow Matching and Robust Tracking
Critic in the Loop: A Tri-System VLA Framework for Robust Long-Horizon Manipulation
A Simple Baseline for Unifying Understanding, Generation, and Editing via Vanilla Next-token Prediction
Same Error, Different Function: The Optimizer as an Implicit Prior in Financial Time Series
Dissociating Direct Access from Inference in AI Introspection
SeedPolicy: Horizon Scaling via Self-Evolving Diffusion Policy for Robot Manipulation
How Does the ReLU Activation Affect the Implicit Bias of Gradient Descent on High-dimensional Neural Network Regression?
Accelerating Text-to-Video Generation with Calibrated Sparse Attention
Privacy-Aware Camera 2.0 Technical Report
ViterbiPlanNet: Injecting Procedural Knowledge via Differentiable Viterbi for Planning in Instructional Videos