Evoduet improves scientific discovery by evolving search and solutions together
EvoDuet: Bilevel Co-Evolution of Web Searching and Task Solving for Scientific Discovery
Computation and Language
Summary
Sometimes, AI models get stuck when they try to solve problems because they don’t have all the right information. The authors created EvoDuet, a method that helps the AI figure out when it needs to look up new information on the web and when it can use what it already knows. EvoDuet tries different ways of searching and solving problems at the same time, improving how it finds answers. Tests show that this approach helps find better solutions faster on a variety of scientific tasks.
What this means in practice
- •For software developers: Integrate co-evolving web search with AI models to improve problem-solving in applications needing up-to-date external knowledge.
- •For data engineers: Enhance document retrieval systems with adaptive query refinement to match evolving solution criteria dynamically.
Authors
Young-Jun Lee, Jinheon Baek, Soyeong Jeong, Minki Kang, Seungyeon Jwa, Jonghyun Choi, Seungho Han, Dongyeop Kang
Abstract
Evolutionary search with large language models (LLMs) can stall when progress requires external knowledge the model lacks. Supplying relevant documents helps, but simply adding web search tool can keep returning the same pages as solutions change. We introduce EvoDuet, a bi-level optimization method that co-evolves solutions and search queries with fixed model parameters. At each iteration, a retrieval gate lets the LLM assess its knowledge gap and choose to retrieve new documents, reuse stored ones, or proceed without them. An inner loop refines queries and ranks documents by the solution scores they are predicted to yield; an outer loop generates candidates in parallel from these documents and records the evaluated outcomes for later searches. Across 21 optimization tasks with one candidate per iteration, EvoDuet raises OpenEvolve's normalized discovery gain from 74.1% to 78.0% with GPT-5.6-Luna and from 61.3% to 82.3% with Gemini-3.8-Flash, whereas Qwen3.5-9B does not benefit. Our best runs surpass the previously reported best scores on eight tasks, including Swap Reduction on Q20 and Rosetta, and match them on three more. EvoDuet also improves with other scaffolds (e.g., Top-K, EvoX) on Sums/Diffs and Denoising, demonstrating its applicability across evolutionary search scaffolds.