AI研究论文每日精选

MiMo-VL Technical Report

AmbiK: Dataset of Ambiguous Tasks in Kitchen Environment

Advancing Multimodal Reasoning: From Optimized Cold Start to Staged Reinforcement Learning

A Controllable Examination for Long-Context Language Models

MMR-V: What's Left Unsaid? A Benchmark for Multimodal Deep Reasoning in Videos

SuperWriter: Reflection-Driven Long-Form Generation with Large Language Models

OpenThoughts: Data Recipes for Reasoning Models

Establishing Trustworthy LLM Evaluation via Shortcut Neuron Analysis

Voyager: Long-Range and World-Consistent Video Diffusion for Explorable 3D Scene Generation

VisCoder: Fine-Tuning LLMs for Executable Python Visualization Code Generation

IllumiCraft: Unified Geometry and Illumination Diffusion for Controllable Video Generation

Image Editing As Programs with Diffusion Models

Unleashing the Reasoning Potential of Pre-trained LLMs by Critique Fine-Tuning on One Problem

Ψ-Sampler: Initial Particle Sampling for SMC-Based Inference-Time Reward Alignment in Score Models

LayerFlow: A Unified Model for Layer-aware Video Generation

DenseDPO: Fine-Grained Temporal Preference Optimization for Video Diffusion Models

SVGenius: Benchmarking LLMs in SVG Understanding, Editing and Generation

TimeHC-RL: Temporal-aware Hierarchical Cognitive Reinforcement Learning for Enhancing LLMs' Social Intelligence

Rectified Sparse Attention

Orak: A Foundational Benchmark for Training and Evaluating LLM Agents on Diverse Video Games

Beyond the Surface: Measuring Self-Preference in LLM Judgments

BenchHub: A Unified Benchmark Suite for Holistic and Customizable LLM Evaluation

TalkingMachines: Real-Time Audio-Driven FaceTime-Style Video via Autoregressive Diffusion Models

POSS: Position Specialist Generates Better Draft for Speculative Decoding

Robustness in Both Domains: CLIP Needs a Robust Text Encoder

Critique-GRPO: Advancing LLM Reasoning with Natural Language and Numerical Feedback

DiffDecompose: Layer-Wise Decomposition of Alpha-Composited Images via Diffusion Transformers

Adapt before Continual Learning

Video-Skill-CoT: Skill-based Chain-of-Thoughts for Domain-Adaptive Video Reasoning

CapSpeech: Enabling Downstream Applications in Style-Captioned Text-to-Speech

RefEdit: A Benchmark and Method for Improving Instruction-based Image Editing Model on Referring Expressions

Quantitative LLM Judges

Improving Knowledge Distillation Under Unknown Covariate Shift Through Confidence-Guided Data Augmentation

Follow the Flow: Fine-grained Flowchart Attribution with Neurosymbolic Agents

DLP: Dynamic Layerwise Pruning in Large Language Models

Unleashing Hour-Scale Video Training for Long Video-Language Understanding

TRiSM for Agentic AI: A Review of Trust, Risk, and Security Management in LLM-based Agentic Multi-Agent Systems

HTSC-2025: A Benchmark Dataset of Ambient-Pressure High-Temperature Superconductors for AI-Driven Critical Temperature Prediction

Segment Policy Optimization: Effective Segment-Level Credit Assignment in RL for Large Language Models

Rex-Thinker: Grounded Object Referring via Chain-of-Thought Reasoning

Rethinking the Stability-Plasticity Trade-off in Continual Learning from an Architectural Perspective

CRAWLDoc: A Dataset for Robust Ranking of Bibliographic Documents

VLMs Can Aggregate Scattered Training Patches

Robust Neural Rendering in the Wild with Asymmetric Dual 3D Gaussian Splatting

Solving Inverse Problems with FLAIR

FinChain: A Symbolic Benchmark for Verifiable Chain-of-Thought Financial Reasoning

Small Language Models are the Future of Agentic AI

Sounding that Object: Interactive Object-Aware Image to Audio Generation

Survey of Active Learning Hyperparameters: Insights from a Large-Scale Experimental Grid

RiOSWorld: Benchmarking the Risk of Multimodal Compter-Use Agents