2
0

Paper page - MoTE: Mixture of Ternary Experts for Memory-efficient Large Multimodal Models

11mon 2d ago by lemmy.world/u/cm0002 in gai@sopuli.xyz from huggingface.co