Metraipolis › archive › dc_56ef64a29f58

DENSE: Distilling Agent Trajectories into Evidence-Grounded Shortcut Trees for Self-Refinement — arXiv:2609.21423

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.21423
License (as archived)
CC0-1.0
Archived
2026-10-10T01:05:16.441Z
SHA-256 of the archived text
9aff88ebf5a1135b19839d52c742b38949b6f7d37b41060bc1719e4f7c6a270c
Stable URL of this page
https://metraipolis.com/documento/dc_56ef64a29f58

How to cite

DENSE: Distilling Agent Trajectories into Evidence-Grounded Shortcut Trees for Self-Refinement — arXiv:2609.21423 — arXiv (en), https://arxiv.org/abs/2609.21423 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_56ef64a29f58 (Metraipolis archive dc_56ef64a29f58, sha256 9aff88ebf5a1135b19839d52c742b38949b6f7d37b41060bc1719e4f7c6a270c)
BibTeX
@misc{dc56ef64a29f58,
  title = {DENSE: Distilling Agent Trajectories into Evidence-Grounded Shortcut Trees for Self-Refinement — arXiv:2609.21423},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.21423},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_56ef64a29f58; sha256 9aff88ebf5a1135b19839d52c742b38949b6f7d37b41060bc1719e4f7c6a270c},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.21423. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Online agent deployments accumulate execution trajectories at massive scale and behavioral diversity, for which predefined annotation criteria hardly exist. Extracting useful evidence therefore demands costly manual annotation or verifier signals that fail to scale, leaving valuable evidence buried among redundant, incomplete, and failed executions. This raises a question: without post-execution rewards or correctness labels, how can reusable experience be distilled from the trajectories themselves? To address this challenge, we introduce DENSE (Distilling Evidence from Nested Subtask Executions), which organizes trajectory-derived evidence into nested shortcut trees. By consolidating redundant attempts, identifying resolved subtasks, and retaining useful steps alongside outstanding requirements, DENSE transforms noisy execution traces into structured and reusable task-solving feedback. To evaluate whether such feedback helps agents retry the same task, we design REFIT, which measures success-rate changes between the initial attempt and feedback-guided retries. Among feedback methods without external outcome supervision, DENSE achieves the highest strict pass rate across four agent models on Terminal-Bench 2.1, improving over initial attempts by 7.12-21.81 percentage points with 19.0-43.6% fewer agent tokens on retries. In addition, on hard tasks DENSE consistently outperforms self-reflection in cumulative pass rate across multiple feedback iterations on all four models, demonstrating its strong potential for continual agent self-improvement.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-21 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.21423 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-10T00:23:42.450Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-21?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).