Luke Zettlemoyer

7 papers University of Washington; Meta

• CAT: Content-Adaptive Image Tokenization
• FlexOLMo: Open Language Models for Flexible Data Use
• Heterogeneous Swarms: Jointly Optimizing Model Roles and Weights for Multi-LLM Systems
• LMFusion: Adapting Pretrained Language Models for Multimodal Generation
• Meta CLIP 2: A Worldwide Scaling Recipe
• Precise Information Control in Long-Form Text Generation
• When Worse is Better: Navigating the Compression Generation Trade-off In Visual Tokenization