self-evolving-agents-accumulate-reusable-skills-bank-admit-new-skill CLM-0483mechanismsingle paperReviewed by AI
For self-evolving agents that accumulate reusable skills in a bank and admit a new skill whenever aggregate task reward rises, the accepted skills can be collectively net-negative, because a bank-level reward change cannot identify which skill contributed and a skill's contribution depends on the other skills present and on the deployment domain.
Capability: Improving its own scaffolding from failures
Observed on
Skill-bank self-evolving agents (MemSkill-style) with an outcome-only acceptance gate, audited on LoCoMo, LongMemEval, HotpotQA and ALFWorld with a Llama-3.3-70B skill-extraction p.
Sources
- Validated by: Matched-protocol comparison against MemSkill and six memory-agent baselines: the coalition gate accepted 1 of 12 candidates versus MemSkill's 6, and a K-controlled ablation showed MemSkill's extra additions were -1.38 pp on average; label-free masking improved all six trained banks on HotpotQA-50 LLM-judge (+3.59 pp average on coalition-gated banks, +7.17 pp on baseline banks), and the retrieval-quality signal correlated with label-based skill ranking at rho=+0.76 versus -0.18 for a policy-entropy signal.. [truncated]
Either way, bring evidence: a link, or what you saw.
Disagreeing is the most useful thing you can do here. Nothing on this page waits for a reviewer: the counts above move when evidence lands, and every source stays attached with who brought it.
Related claims
- For self-evolving LLM agents that admit a new skill into a persistent skill bank whenever aggregate outcome reward rises, the accepted skills are collectively net-negative and the gate cannot do better in principle, because the bank-level gain is one scalar that leaves every incumbent skill's contribution unconstrained; scoring sampled coalition knockouts at the gate, and masking transferred skills whose removal improves retrieval quality on unlabeled target queries, recovers the missing composition and domain dependence.Improving its own scaffolding from failures · reviewed by AI
- When an agent harness mines reusable rules and tools from its own labelled diagnosis trajectories, promoting every proposal that helps the cases it was mined from makes the harness worse over time than not evolving at all, and only admitting a proposal when it also fails to regress on a freshly resampled held-out case set keeps accuracy climbing across successive evolution rounds.Improving its own scaffolding from failures · reviewed by AI
- For a frozen-weight video agent on open-ended surgical video QA, mining candidate retrieval skills from the agent's own low-scoring traces and keeping each only when it raises a four-dimensional validation reward on ~40 held-out questions beats installing a hand-written general 'looking discipline' skill, and the single distilled temporal skill (re-read the window adjacent to a localized event for before/after questions) carries most of the gain.Improving its own scaffolding from failures · reviewed by AI
- In label-free RL post-training of reasoning models, deriving each model's reward from a separately-trained peer model's majority vote — rather than from its own majority vote — avoids the reward collapse and length degeneration that self-rewarding runs show, and the benefit grows as the paired models differ in family or size because their errors overlap less.Getting more from several models than from one · reviewed by AI
- For LLM agents that score their own actions during exploration in text-based games and then reuse that experience, the accuracy of those self-scores against the environment verifier does not predict later gains on held-out stricter configurations: block-level correlations between judgment agreement or calibration and the next checkpoint's score change sit near zero, so recognizing a good action is not the bottleneck — converting it into a transferable policy is.Improving its own scaffolding from failures · reviewed by AI
Notes
Drafted from the paper by a model and filed unreviewed. Visible here so it can be read, not because anyone has vouched for it: it does not move any technique's standing and does not count toward the internal scorecard. Drafted confidence: medium. Falsifier as drafted: A knockout audit on skill banks grown by an outcome-only gate showing that accepted skills have positive individual marginal contributions, stable across bank compositions and across source and target domains.