---
title: "自我分段的轨迹：以智能体声明的边界作为训练单元"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.02302v1"
permalink: "https://feed.astralor.com/items/itm_b804fa0df6cd006fe7434389"
published: "2026-08-03T14:27:58.000Z"
score: 78
channel: "ai"
tags: ["DPO","智能体","训练数据","语义自分段","长程轨迹"]
---

# 自我分段的轨迹：以智能体声明的边界作为训练单元

## AI 点评

> 这篇论文值得关注，因为它提出了一种无需外部标注的轨迹分段方法，对做Agent训练和强化学习的人很有价值。它把数据构造变成智能体自我声明的过程，思路值得借鉴。

## 速览

长期编码智能体轨迹与训练信用单元不匹配，单一动作价值不稳定，整集标签混入无效探索，固定窗口切错位置。本文提出采集时语义自分段，让智能体在轨迹生成时声明边界，通过可证伪假设产生可变长度语义段，并允许评审者按名否定，制造错误-纠正转换。实验显示，给定切割点，模型归因假设超随机两倍；标注者边界放置准确率高，而机械规则失效。DPO训练后，部分错误被纠正，表明边界可作为有效训练单元。

## 关键点

- 提出采集时语义自分段，让智能体在轨迹生成中声明边界，不再依赖固定窗口或事后分段。
- 实验证明模型对假设的归因显著优于随机，且边界放置无法用机械规则复现。
- DPO训练中边界对约束结构有效的纠错，但对抗项不变，指向语料多样性是下一步关键。

## 全文

Long-horizon coding-agent trajectories are poorly matched to the credit units available to train on: a single action has no stable value, an episode label merges productive exploration with abandoned directions, and a fixed window cuts where the logging mechanics fall. We introduce collection-time semantic self-segmentation, in which a declarative contract has the acting agent expose its own boundaries while the trajectory is generated. Instantiated with falsifiable causal hypotheses, successive adoptions expose variable-length semantic phases, and no milestone vocabulary, gold patch, environment replay, teacher logits, or retrospective segmenter places a boundary. Because the agent names its conjecture, a reviewer can negate it by name, which lets our protocol manufacture wrong-cause-then-correction transitions that recorded work rarely contains; one collection then yields four supervised targets, including audit supervision from exactly the failed regions an episode label discards. We then ask what survives deleting the declaration. Given the cut points but not the hypothesis, a model attributes action blocks to their governing hypothesis at over twice chance, beating equal-length blocks over the same trajectories (paired sign test $p = 0.0002$), surviving a lexical control and collapsing under label permutation. Asked instead to place boundaries, a code-blind annotator matches 24 of 40 where random placement matches 11.5, while a mechanical test-event rule beats chance at neither end of a strict-to-permissive sweep. The segments are therefore coherent and not cheaply reproducible. Downstream, DPO on 2,551 phase-boundary pairs changes no decision on 91 adversarial held-out items, while four of 60 change on matched-construction items, all wrong to right, where two controls change none: with 1,825 pairs from one generator, the variable to vary next is corpus diversity, not the boundary.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_b804fa0df6cd006fe7434389