Search papers, labs, and topics across Lattice.
University of Amsterdam
4
0
10
Forget relying on implicit reasoning: A-MAR's explicit reasoning plans unlock better artwork understanding by strategically retrieving relevant evidence.
Forget massive offline datasets: TCL slashes tensor program optimization time by 16x while *improving* inference latency, thanks to a Mamba-powered cost model and continual learning.
Ditch the GPU? Active Imitation Learning can tame the thermal chaos of running large foundation models on 3D-stacked CPUs, unlocking a cost-effective alternative for LFM inference.
Speaker drift, a subtle but pervasive flaw in modern TTS, can now be automatically detected by prompting LLMs with geometric representations of speaker embeddings.
