TL;DR
Ace-Skill introduces a co-evolutionary framework that enhances self-evolving multimodal agents by optimizing rollout sampling and knowledge organization, leading to significant performance improvements and knowledge transfer capabilities.
Contribution
It presents a novel joint optimization approach combining prioritized sampling and semantic clustering to improve self-evolution in multimodal agents.
Findings
Achieved +35.46% in Avg@4 accuracy across benchmarks.
Enabled a 35B MoE model to outperform proprietary counterparts.
Transferred knowledge effectively to smaller models in zero-shot settings.
Abstract
Self-evolving agents present a promising path toward continual adaptation by distilling task interactions into reusable knowledge artifacts. In practice, this paradigm remains hindered by two coupled bottlenecks: data inefficiency, where costly rollout effort is disproportionately spent on low-value samples rather than informative ones, and knowledge interference, where heterogeneous knowledge stored in shared repositories leads to noisy retrieval and task-misaligned guidance. Together, these issues form a self-reinforcing failure loop in which uninformative rollouts yield noisy knowledge, which in turn degrades subsequent rollouts. In this work, we introduce Ace-Skill, a co-evolutionary framework that jointly optimizes rollout allocation and knowledge organization for self-evolving multimodal agents. Specifically, Ace-Skill combines aprioritized sampler with lazy-decay proficiency…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Code & Models
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
