Search papers, labs, and topics across Lattice.
3
0
7
3
Imagine getting the accuracy boost of "glimpsing into the future" for KV cache eviction, but without the hefty cost of draft generation – LookaheadKV makes it real.
Fine-tuning LLMs can kill their in-context learning abilities, but this work identifies a simple fix: only update the value matrix.
Language agents can now navigate complex, constrained environments with significantly improved success rates thanks to a new framework that combines multi-plan aggregation with constrained decoding and adaptive re-planning.
