AI研究论文每日精选

CLIMB: CLustering-based Iterative Data Mixture Bootstrapping for Language Model Pre-training

Antidistillation Sampling

Packing Input Frame Context in Next-Frame Prediction Models for Video Generation

Generate, but Verify: Reducing Hallucination in Vision-Language Models with Retrospective Resampling

WORLDMEM: Long-term Consistent World Simulation with Memory

Perception Encoder: The best visual embeddings are not at the output of the network

A Strategic Coordination Framework of Small LLMs Matches Large LLMs in Data Synthesis

VistaDPO: Video Hierarchical Spatial-Temporal Direct Preference Optimization for Large Video Models

ChartQAPro: A More Diverse and Challenging Benchmark for Chart Question Answering

DMM: Building a Versatile Image Generation Model via Distillation-Based Model Merging

NoisyRollout: Reinforcing Visual Reasoning with Data Augmentation

PerceptionLM: Open-Access Data and Models for Detailed Visual Understanding

InstantCharacter: Personalize Any Characters with a Scalable Diffusion Transformer Framework

70% Size, 100% Accuracy: Lossless LLM Compression for Efficient GPU Inference via Dynamic-Length Float

Sleep-time Compute: Beyond Inference Scaling at Test-time

Exploring Expert Failures Improves LLM Agent Tuning

CCMNet: Leveraging Calibrated Color Correction Matrices for Cross-Camera Color Constancy

FocusedAD: Character-centric Movie Audio Description

Complex-Edit: CoT-Like Instruction Generation for Complexity-Controllable Image Editing Benchmark

Retrieval-Augmented Generation with Conflicting Evidence

Set You Straight: Auto-Steering Denoising Trajectories to Sidestep Unwanted Concepts

MetaSynth: Meta-Prompting-Driven Agentic Scaffolds for Diverse Synthetic Data Generation

Learning Occlusion-Robust Vision Transformers for Real-Time UAV Tracking