Search papers, labs, and topics across Lattice.
3
0
4
9
SNIPER achieves a remarkable CRAFT score of 0.98, ensuring near-perfect adherence to compression budgets while maintaining model performance.
MAESTRO prunes MoE models more effectively by leveraging the interdependencies of expert routing, achieving superior performance retention and consistency across tasks.
Vision-language models can significantly enhance language models through knowledge distillation, even without direct textual understanding, challenging conventional KD paradigms.
