Arash Ahmadian

This article discusses methods for improving AI alignment with various cultures.

Contrastive Policy Gradient offers a more efficient way to enhance language models.

New methods for training language models to support diverse languages.

Transform discarded models into powerful new solutions through model merging.