Jian Yang

A method to enhance language models by creating engaging multi-turn dialogues.

A new model uses text to guide LiDAR data creation, reducing costs and improving accuracy.

A new method improves computer-generated music quality by separating melody and rhythm.

Add-SD simplifies image editing by allowing realistic object additions through text prompts.

A new method predicts the significance of articles based on titles and abstracts.

AquilaMoE uses EfficientScale to optimize bilingual language model training with less data.

A new method streamlines action recognition in videos using existing image models.

New technique improves detection of camouflaged objects in various fields.

Exploring how preference learning improves language model alignment with human expectations.

wgatools streamlines whole genome alignment with versatile tools for researchers.

GLCONet improves the detection of camouflaged objects using local and global features.

SoccerNet 2024 challenges drive innovation in video understanding for soccer.

A new tool evaluates large language models' performance across multiple data types.

This article explores the significance of spin-space groups in magnetic materials.

A new method to balance general knowledge and task-specific adaptation in models.

UDCNet simplifies identifying key objects in satellite images.

TRIDENT enhances machine recognition of new object combinations, addressing key challenges in AI learning.

Learn how social event detection works and its importance in today's world.

CIDER improves recommendations by connecting preferences across different areas like books and movies.

Discover how video trimming transforms viewing experiences by highlighting the best moments.

STTrack advances object tracking by combining multiple data sources for better accuracy.

Introducing ERGNN, a new method improving graph neural networks with rational filters.