Lin Gui

FUTE offers a new approach to task embeddings across multiple models.

A new method enhances language models' reasoning capabilities through structured feedback.

This work highlights dependencies between content and style in counterfactual generation.

How the arrangement of examples shapes language model effectiveness.

PLAYER* enhances agent communication and problem-solving in murder mystery settings.

A new approach improves causal event extraction using human-centered evaluation.

Discover how language models can learn and adapt while avoiding harmful content.