Search papers, labs, and topics across Lattice.
3
0
7
Achieving a 90% boost in prefill throughput for MoE models could redefine the efficiency of large-scale language model serving.
Neural ODEs aren't just for continuous-time dynamics anymore: this work shows they can unify and universally approximate both finite and infinite-dimensional operators, bridging the gap between NNs and neural operators.
OpenSearch-VL offers a fully transparent recipe for training state-of-the-art multimodal search agents, finally democratizing access to a capability previously locked behind closed doors.
