Search papers, labs, and topics across Lattice.
4
0
5
4
This work evaluates three evolutionary search strategies on five optimization tasks, commonly used by papers in the genre to report results, and suggests that the best way to split a fixed budget between more seeds and more iterations changes with the strategy, the task, and the total budget.
Spotlights is presented, a system that performs optimization-opportunity discovery as a distinct and empirically evaluable step between a broad engineering objective and subsequent implementation and validation.
Evolving LLM prompts on a budget can cut search costs by up to 54x without sacrificing performance, reshaping the economics of prompt optimization.
Late-interaction retrieval just got a whole lot faster and cheaper: Flash-MaxSim slashes memory usage by 16x and speeds up inference by 4.7x on an H100 by ditching the massive similarity tensor.
