Yejin Choi

Learn how preference feedback shapes better language model outputs.

MINT-1T is the largest open-source dataset for training multimodal models.

A new framework improves language models' representation of diverse human values.

Learn about the importance of safety measures in language models.

Introducing a tool to enhance safety in language model interactions.

Using technology to improve fact-checking and combat misinformation effectively.

Analyzing how LLMs make moral choices across languages and cultures.

Study reveals surprising risks of sharing sensitive data with chatbots.

A new benchmark evaluates LLMs for factual accuracy.

xGen-MM enhances multimodal models for better image and text learning.

Enhancing LLMs with external memory aids multi-step reasoning tasks.

A tool to analyze chat logs quickly and effectively for researchers.

KALE combines images with rich captions for better understanding.

Learn how Negative Token Merging is changing AI image generation.

Researchers develop tools to refine text and design proteins efficiently.