aiqbench analysis
This research introduces DecoEvo, a novel framework for text-space optimization that addresses the bottleneck of fixed evaluation criteria in large language model adaptation. By decoupling the evolution of solver skills from rubric-generator objectives using independent audits rather than aggregate scores, the method effectively uncovers previously invisible dimensions of performance gaps across multiple benchmarks and LLM backbones. The approach demonstrates significant relative gains over existing techniques while maintaining inspectable artifacts without modifying model weights directly.
DecoEvo: Score-Decoupled Co-Evolution of Solver and Rubric-Generator Skills in Text Space
arXiv:2607.25675v1 Announce Type: new Abstract: 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.
Abstract: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.
Artificial Intelligence (cs.AI)
arXiv:2607.25675 [cs.AI]
(or arXiv:2607.25675v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2607.25675 arXiv-issued DOI via DataCite (pending registration)
Full text under the publisher's open licence · 211 words
Related coverage
Who Grades the Grader? Co-Evolving Evaluation Metrics and Skills for Self-Improving LLM Agents
arXiv · Jul 31, 2026
SkillLift: Learning Dense Rubrics from Sparse Oracles for Efficient Skill Evolution
arXiv · Sep 16, 2026
Recursive Reward Improvement via Adversarial Meta-Prompt Evolution
arXiv · Oct 17, 2026
Socratic-SWE: Self-Evolving Coding Agents via Trace-Derived Agent Skills
arXiv · Jun 8, 2026