AI研究論文デイリー

AlphaMaze: Enhancing Large Language Models' Spatial Intelligence via GRPO

How Much Knowledge Can You Pack into a LoRA Adapter without Harming LLM?

Geolocation with Real Human Gameplay Data: A Large-Scale Dataset and Human-Like Reasoning Framework

Logic-RL: Unleashing LLM Reasoning with Rule-Based Reinforcement Learning

Generating π-Functional Molecules Using STGG+ with Active Learning

Scaling Text-Rich Image Understanding via Code-Guided Synthetic Multimodal Data Generation

LongWriter-V: Enabling Ultra-Long and High-Fidelity Generation in Vision-Language Models

Enhancing Cognition and Explainability of Multimodal Foundation Models with Self-Synthesized Data

CLIPPER: Compression enables long-context synthetic data generation

How Much Do LLMs Hallucinate across Languages? On Multilingual Estimation of LLM Hallucination in the Wild

Does Time Have Its Place? Temporal Heads: Where Language Models Recall Time-specific Information

LLM-based User Profile Management for Recommender System

SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Features

NAVIG: Natural Language-guided Analysis with Vision Language Models for Image Geo-localization

Dynamic Concepts Personalization from Single Videos

Unstructured Evidence Attribution for Long Context Query Focused Summarization

How to Get Your LLM to Generate Challenging Problems for Evaluation

PC-Agent: A Hierarchical Multi-Agent Collaboration Framework for Complex Task Automation on PC

S^2R: Teaching LLMs to Self-verify and Self-correct via Reinforcement Learning

Generating Skyline Datasets for Data Science Models

SuperGPQA: Scaling LLM Evaluation across 285 Graduate Disciplines

Multimodal RewardBench: Holistic Evaluation of Reward Models for Vision Language Models

LServe: Efficient Long-sequence LLM Serving with Unified Sparse Attention

S*: Test Time Scaling for Code Generation

From RAG to Memory: Non-Parametric Continual Learning for Large Language Models

Symmetrical Visual Contrastive Optimization: Aligning Vision-Language Models with Minimal Contrastive Images

Discovering highly efficient low-weight quantum error-correcting codes with reinforcement learning

RelaCtrl: Relevance-Guided Efficient Control for Diffusion Transformers

MLGym: A New Framework and Benchmark for Advancing AI Research Agents