Shuming Shi

Disco-Bench measures how models relate sentences and understand language connections.

This article examines hallucination in AI language models and ongoing research.

A new way to collect instruction-tuning data for large language models.

A new framework enhances language models by blending text and images for richer interactions.

A look at the obstacles and progress in translation technology.

This study highlights the importance of measuring uncertainty in language model evaluations.

A new framework reduces data biases improving instruction-following performance in language models.

Exploring methods to enhance LLMs for practical applications.

A new method enhances word suggestions for translators using contextual understanding.