Yongbin Li

Solar transforms data types into language for better question answering.

A novel approach uses wider networks to improve evaluation quality of language models.

Exploring the impact of instruction complexity on language model performance.

UniSA framework unifies tasks in sentiment analysis for better emotion recognition.

UniPCM leverages automatic prompt generation to improve dialogue systems efficiently.

A new method enhances language models' understanding of conversations.

A method called Masked Thought Fine-Tuning boosts math problem-solving in language models.

Examining how LLMs ensure safety and the impact of jailbreaks.

New framework improves training data for language models using images and text.