Search papers, labs, and topics across Lattice.
5
0
9
1
You can now dial a knob to make your LLM either super-distillable or completely un-distillable, opening up new possibilities for both efficient knowledge transfer and robust model protection.
Ditch the learned router: a global scheduler for Mixture-of-Experts models unlocks state-of-the-art multi-domain learning by explicitly optimizing dataset-to-expert assignments.
Task arithmetic works because models internally allocate distinct features to different tasks, and enforcing this specialization via orthogonality regularization unlocks even better editing.
A stark capability cliff reveals that even leading AI models falter on complex workflows, achieving less than 15% success despite advancements in tool-use benchmarks.
LLMs can now navigate massive toolsets with a "Try-Check-Retry" loop, boosting tool-calling accuracy by up to 25% and letting smaller models punch above their weight.
