AdaMoE: Token-Adaptive Routing with Null Experts for Mixture-of-Experts Language Models Paper • 2406.13233 • Published Jun 19, 2024 • 1
MatryoshkaKV: Adaptive KV Compression via Trainable Orthogonal Projection Paper • 2410.14731 • Published Oct 16, 2024
Show-o Turbo: Towards Accelerated Unified Multimodal Understanding and Generation Paper • 2502.05415 • Published Feb 8 • 22