16 TriForce: Lossless Acceleration of Long Sequence Generation with Hierarchical Speculative Decoding · 5 authors 1
12 MoA: Mixture-of-Attention for Subject-Context Disentanglement in Personalized Image Generation · 6 authors 1
11 OpenBezoar: Small, Cost-Effective and Open Models Trained on Mixes of Instruction Data · 4 authors 1