跳到正文
HuggingFace Daily Papers(社區熱門論文)·· 2026-10-06AI 評分46

Self-Retrospection Distillation: Turning Post-hoc Experiences into Prior Foresight

Self-Retrospection Distillation: Turning Post-hoc Experiences into Prior Foresight

AI 導讀

Self-Retrospection Distillation(SRD)利用互動後的軌跡經驗,監督智能體在互動前作出預見,並作為訓練目標,無需在推理時明確生成。在 10 項工具整合推理及長期智能體任務中,SRD 相較 RLVR 和自我蒸餾基線最高提升 24.2 pp。2B 設定下,當 98% 的 rollout 羣組全部失敗時,RLVR 成功率為 0.0%,加入 SRD 後達 60.6%。

正文

Published on Oct 6

Authors:

,

,

,

,

,

,

,

,

,

,

Abstract

Reinforcement learning with verifiable rewards (RLVR) turns agent experience into learning signals primarily through scalar outcome rewards after interaction. For group-relative objectives, however, this signal vanishes when all rollouts receive the same reward, even though their trajectories may reveal useful information about what the task requires and how the agent fails. We ask a complementary question: can hindsight teach an agent what it could have anticipated before acting? We introduce prospective learning, which uses post-hoc experience to supervise foresight predictions from the pre-interaction view, and instantiate it with Self-Retrospection Distillation (SRD). Intuitively, a completed trajectory reveals knowledge that would have been useful and pitfalls that should be avoided; SRD distills this privileged hindsight into trajectory-blind foresight of the same policy. Foresight serves only as a training target and need not be explicitly generated at inference time. Across 10 tool-integrated reasoning and long-horizon agentic tasks, SRD complements RLVR and self-distillation baselines with gains of up to 24.2 pp. Its advantage is especially pronounced when reward contrast is scarce: when 37--98% of rollout groups are reward-uniform across model scales, yet SRD can still exploit learning signal from sampled trajectories. In the 2B setting, where 98% of groups are all-failure, the RLVR training ends up at 0.0% success, while adding SRD reaches 60.6% under the same rollout budget. Our results suggest that post-hoc agent experience is useful not only for evaluating or improving behavior, but also for shaping predictive representations before available interaction.

View arXiv page View PDF Add to collection

Models citing this paper 0

No model linking this paper

Cite arxiv.org/abs/2610.08077 in a model README.md to link it from this page.

Datasets citing this paper 0

No dataset linking this paper

Cite arxiv.org/abs/2610.08077 in a dataset README.md to link it from this page.

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2610.08077 in a Space README.md to link it from this page.

Collections including this paper 0

No Collection including this paper

Add this paper to a collection to link it from this page.

來源:HuggingFace Daily Papers(社區熱門論文) · huggingface.co