Blog
No more endless PDFs. Discover the core value of the latest top-tier research in one article.
Revealing Modular Gradient Noise Imbalance in LLMs: Calibrating Adam via Signal-to-Noise Ratio
Normalized Architectures are Natively 4-Bit
3DSS: 3D Surface Splatting for Inverse Rendering
OPSD Compresses What RLVR Teaches: A Post-RL Compaction Stage for Reasoning Models
Prediction and Empowerment: A Theory of Agency through Bridge Interfaces
Spark3R: Asymmetric Token Reduction Makes Fast Feed-Forward 3D Reconstruction
UniPool: A Globally Shared Expert Pool for Mixture-of-Experts
Attractor Geometry of Transformer Memory: From Conflict Arbitration to Confident Hallucination
The Interplay of Data Structure and Imbalance in the Learning Dynamics of Diffusion Models
Mind the Gap? A Distributional Comparison of Real and Synthetic Priors for Tabular Foundation Models
Long Context Pre-Training with Lighthouse Attention
Are We Making Progress in Multimodal Domain Generalization? A Comprehensive Benchmark Study
Quasi sdf-absorbing ideals in commutative rings
EnterpriseRAG-Bench: A RAG Benchmark for Company Internal Knowledge
Steering Visual Generation in Unified Multimodal Models with Understanding Supervision
Render, Don't Decode: Weight-Space World Models with Latent Structural Disentanglement
NavOne: One-Step Global Planning for Vision-Language Navigation on Top-Down Maps
Cited but Not Verified: Parsing and Evaluating Source Attribution in LLM Deep Research Agents
SkillRet: A Large-Scale Benchmark for Skill Retrieval in LLM Agents
Rigorous Interpretation Is a Form of Evaluation
A note on the modal logic of symmetric extensions
Horizontal transport as a source of disequilibrium chemistry on the nightside of a hot exoplanet
MASPO: Joint Prompt Optimization for LLM-based Multi-Agent Systems
Adaptive Q-Chunking for Offline-to-Online Reinforcement Learning
Understanding diffusion models requires rethinking (again) generalization
Text-Conditional JEPA for Learning Semantically Rich Visual Representations
Beyond Uniform Credit Assignment: Selective Eligibility Traces for RLVR
Rethinking RL for LLM Reasoning: It's Sparse Policy Selection, Not Capability Learning
Beyond Steering Vector: Flow-based Activation Steering for Inference-Time Intervention
Who Prices Cognitive Labor in the Age of Agents? A Position on Compute-Anchored Wages