AI 연구 논문 데일리

NovelSeek: When Agent Becomes the Scientist -- Building Closed-Loop System from Hypothesis to Verification

Scaling Reasoning, Losing Control: Evaluating Instruction Following in Large Reasoning Models

Tool-Star: Empowering LLM-Brained Multi-Tool Reasoner via Reinforcement Learning

Pixel Reasoner: Incentivizing Pixel-Space Reasoning with Curiosity-Driven Reinforcement Learning

KRIS-Bench: Benchmarking Next-Level Intelligent Image Editing Models

QuickVideo: Real-Time Long Video Understanding with System Algorithm Co-Design

GoT-R1: Unleashing Reasoning Capability of MLLM for Visual Generation with Reinforcement Learning

LLaDA-V: Large Language Diffusion Models with Visual Instruction Tuning

Scaling Diffusion Transformers Efficiently via μP

Risk-Averse Reinforcement Learning with Itakura-Saito Loss

Understanding Generative AI Capabilities in Everyday Image Editing Tasks

Let LLMs Break Free from Overthinking via Self-Braking Tuning

VideoGameQA-Bench: Evaluating Vision-Language Models for Video Game Quality Assurance

Mind the Gap: Bridging Thought Leap for Improved Chain-of-Thought Tuning

AceReason-Nemotron: Advancing Math and Code Reasoning through Reinforcement Learning

Dimple: Discrete Diffusion Multimodal Large Language Model with Parallel Decoding

Backdoor Cleaning without External Guidance in MLLM Fine-tuning

Training-Free Efficient Video Generation via Dynamic Token Carving

SophiaVL-R1: Reinforcing MLLMs Reasoning with Thinking Reward

Fixing Data That Hurts Performance: Cascading LLMs to Relabel Hard Negatives for Robust Information Retrieval

SpatialScore: Towards Unified Evaluation for Multimodal Spatial Understanding

LaViDa: A Large Diffusion Language Model for Multimodal Understanding

TinyV: Reducing False Negatives in Verification Improves RL for LLM Reasoning

WebAgent-R1: Training Web Agents via End-to-End Multi-Turn Reinforcement Learning

Training-Free Reasoning and Reflection in MLLMs

GRIT: Teaching MLLMs to Think with Images

AGENTIF: Benchmarking Instruction Following of Large Language Models in Agentic Scenarios

Think or Not? Selective Reasoning via Reinforcement Learning for Vision-Language Models

VLM-R^3: Region Recognition, Reasoning, and Refinement for Enhanced Multimodal Chain-of-Thought

OViP: Online Vision-Language Preference Learning

Training Step-Level Reasoning Verifiers with Formal Verification Tools

SafeKey: Amplifying Aha-Moment Insights for Safety Reasoning

Reinforcement Learning Finetunes Small Subnetworks in Large Language Models

Think-RM: Enabling Long-Horizon Reasoning in Generative Reward Models

Let Androids Dream of Electric Sheep: A Human-like Image Implication Understanding and Reasoning Framework

Multi-SpatialMLLM: Multi-Frame Spatial Understanding with Multi-Modal Large Language Models

Robo2VLM: Visual Question Answering from Large-Scale In-the-Wild Robot Manipulation Datasets

Steering Large Language Models for Machine Translation Personalization

When Do LLMs Admit Their Mistakes? Understanding the Role of Model Belief in Retraction

Date Fragments: A Hidden Bottleneck of Tokenization for Temporal Reasoning

How Do Large Vision-Language Models See Text in Image? Unveiling the Distinctive Role of OCR Heads

RAVENEA: A Benchmark for Multimodal Retrieval-Augmented Visual Culture Understanding

MUG-Eval: A Proxy Evaluation Framework for Multilingual Generation Capabilities in Any Language

SPhyR: Spatial-Physical Reasoning Benchmark on Material Distribution

RoPECraft: Training-Free Motion Transfer with Trajectory-Guided RoPE Optimization on Diffusion Transformers

SAKURA: On the Multi-hop Reasoning of Large Audio-Language Models Based on Speech and Audio Information