AI 연구 논문 데일리

Baichuan-Omni Technical Report

Meissonic: Revitalizing Masked Generative Transformers for Efficient High-Resolution Text-to-Image Synthesis

StructRAG: Boosting Knowledge Intensive Reasoning of LLMs via Inference-time Hybrid Information Structurization

From Generalist to Specialist: Adapting Vision Language Models via Task-Specific Visual Instruction Tuning

Multi-Agent Collaborative Data Selection for Efficient LLM Pretraining

Mechanistic Permutability: Match Features Across Layers

EvolveDirector: Approaching Advanced Text-to-Image Generation with Large Vision-Language Models

SuperCorrect: Supervising and Correcting Language Models with Error-Driven Insights

PositionID: LLMs can Control Lengths, Copy and Paste with Explicit Positional Awareness

Semantic Score Distillation Sampling for Compositional Text-to-3D Generation

KV Prediction for Improved Time to First Token

Think While You Generate: Discrete Diffusion with Planned Denoising

ZeroComp: Zero-shot Object Compositing from Image Intrinsics via Diffusion

I-Max: Maximize the Resolution Potential of Pre-trained Rectified Flow Transformers with Projected Flow

DA-Code: Agent Data Science Code Generation Benchmark for Large Language Models

MiRAGeNews: Multimodal Realistic AI-Generated News Detection

SimpleStrat: Diversifying Language Model Generation with Stratification

Mentor-KD: Making Small Language Models Better Multi-step Reasoners

GenARM: Reward Guided Generation with Autoregressive Reward Model for Test-time Alignment

Synth-SONAR: Sonar Image Synthesis with Enhanced Diversity and Realism via Dual Diffusion Models and GPT Prompting