Yu Qiao

A new method combines language models with reinforcement learning for AI training.

Vlogger simplifies video blogging, making it quicker and easier for creators.

A-Eval assesses models for segmenting abdominal organs across diverse datasets.

A new method for improving AI's reasoning and explanation capabilities.

RobotScript enhances how robots execute tasks from natural language.

A new framework enhances robot actions through human commands.

The All-Seeing Project V2 improves AI's understanding of object relationships in images.

A high-quality dataset for training language models from English web content.

AI models improve understanding of driving scenes for safer navigation.

AVIBench tests LVLMs to ensure they withstand adversarial visual instructions.

A new model improves video understanding through innovative training techniques.

Researchers create a dataset to study how people learn by mimicking others.

DIBS enhances video event captioning by refining boundaries using unlabeled data.

Transform text into images, videos, and audio seamlessly with Lumina-T2X.

A new approach enhances self-driving cars by mimicking human thinking patterns.

This article details an innovative approach to improve language models using smaller models.

A new dataset and model enhance video captioning quality for machines.

A toolkit for assessing the safety of advanced language models.

New approach improves learning from interleaved image-text data.

A comprehensive dataset merging images and text to aid machine learning.

A structured approach to assess text-to-video models with improved efficiency.

A new framework helps language models learn symbolic language without human input.

A new dataset enhances AI's ability to process scientific documents effectively.

Researchers improve translation skills for over 100 languages, focusing on low-resource languages.

This method simplifies adding objects to images with text prompts, ensuring natural results.

A new model revolutionizes image generation from text descriptions, enhancing various industries.

GigaGS tackles challenges in large 3D scene modeling with innovative techniques.

A method enhancing language model alignment with human preferences.

A new method improves reasoning skills in language models using preference optimization.

SyncVIS enhances the tracking and segmentation of objects in videos for various applications.

New method boosts multimodal language models' visual task performance.

Vinci makes daily tasks easier with hands-free help and real-time guidance.