DecoEvo: Score-Decoupled Co-Evolution of Solver and Rubric-Generator Skills in Text Space

Hugging Face Daily Papers Papers

Summary

Introduces DecoEvo, a score-decoupled co-evolution method for LLM optimization in text space that jointly improves solver and rubric-generator skills without gold rubrics, achieving 2.8–5.0% relative gains over baselines across five benchmarks.

Text-space optimization adapts large language models (LLMs) by editing external natural-language artifacts rather than model weights, so the optimized artifacts remain inspectable and the model can be treated as a black box. However, most existing text-space methods keep evaluation fixed. On open-ended tasks, this can become a bottleneck: once the solver improves on the criteria a rubric measures, omitted dimensions remain invisible to the optimization signal. Simply evolving the rubric is also unreliable when updates are selected by the current solver's score, because apparent progress can come from making the rubric easier to satisfy. We introduce DecoEvo (Decoupled Co-Evolution), which co-evolves a solver skill and a rubric-generator skill under decoupled objectives without using gold rubrics during optimization. The solver skill is updated using criterion-level feedback, while the rubric-generator skill is revised through complementary audits of requirement coverage and response discrimination that are independent of aggregate solver score. This separation focuses generator updates on newly exposed solver weaknesses, reducing repeated emphasis on criteria the solver already satisfies. Under each benchmark's official evaluation, DecoEvo outperforms all compared methods across five benchmarks and three LLM backbones, yielding 2.8--5.0\% relative gains over SkillOpt in the five-benchmark average.
Original Article
View Cached Full Text

Cached at: 07/30/26, 05:46 AM

Paper page - DecoEvo: Score-Decoupled Co-Evolution of Solver and Rubric-Generator Skills in Text Space

Source: https://huggingface.co/papers/2607.25675

Abstract

Text-spaceoptimizationadaptslargelanguagemodels(LLMs)byeditingexternalnatural-languageartifactsratherthanmodelweights,sotheoptimizedartifactsremaininspectableandthemodelcanbetreatedasablackbox.However,mostexistingtext-spacemethodskeepevaluationfixed.Onopen-endedtasks,thiscanbecomeabottleneck:oncethesolverimprovesonthecriteriaarubricmeasures,omitteddimensionsremaininvisibletotheoptimizationsignal.Simplyevolvingtherubricisalsounreliablewhenupdatesareselectedbythecurrentsolver’sscore,becauseapparentprogresscancomefrommakingtherubriceasiertosatisfy.WeintroduceDecoEvo(DecoupledCo-Evolution),whichco-evolvesasolverskillandarubric-generatorskillunderdecoupledobjectiveswithoutusinggoldrubricsduringoptimization.Thesolverskillisupdatedusingcriterion-levelfeedback,whiletherubric-generatorskillisrevisedthroughcomplementaryauditsofrequirementcoverageandresponsediscriminationthatareindependentofaggregatesolverscore.Thisseparationfocusesgeneratorupdatesonnewlyexposedsolverweaknesses,reducingrepeatedemphasisoncriteriathesolveralreadysatisfies.Undereachbenchmark’sofficialevaluation,DecoEvooutperformsallcomparedmethodsacrossfivebenchmarksandthreeLLMbackbones,yielding2.8--5.0\%relativegainsoverSkillOptinthefive-benchmarkaverage.

View arXiv pageView PDFAdd to collection

Get this paper in your agent:

hf papers read 2607\.25675

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2607.25675 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2607.25675 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2607.25675 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

Self-Evolving Deep Research via Joint Generation and Evaluation

arXiv cs.CL

Researchers from HKUST, ByteDance, and UCL propose SCORE, a co-evolutionary training framework that jointly trains an LLM as both a deep research report generator and an evaluator, using a meta-harness to dynamically adjust evaluation difficulty and prevent reward saturation. Experiments show consistent improvement in open-ended research report quality.