Metraipolis › archive › dc_64aaceffe137

Agent Self-Improvement with Frozen Weights: The Added Value and Limits of Failure Traces — arXiv:2611.11342

Source: arXiv (en)
Original: https://arxiv.org/abs/2611.11342
License (as archived): CC0-1.0
Archived: 2026-11-10T18:43:20.530Z
SHA-256 of the archived text: e50dd55714b7296fe425d4c3460bb0929d22b1046862aad710a0bd8f09c0d0a0
Stable URL of this page: https://metraipolis.com/documento/dc_64aaceffe137

How to cite

Agent Self-Improvement with Frozen Weights: The Added Value and Limits of Failure Traces — arXiv:2611.11342 — arXiv (en), https://arxiv.org/abs/2611.11342 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_64aaceffe137 (Metraipolis archive dc_64aaceffe137, sha256 e50dd55714b7296fe425d4c3460bb0929d22b1046862aad710a0bd8f09c0d0a0)

BibTeX

@misc{dc64aaceffe137,
  title = {Agent Self-Improvement with Frozen Weights: The Added Value and Limits of Failure Traces — arXiv:2611.11342},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.11342},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_64aaceffe137; sha256 e50dd55714b7296fe425d4c3460bb0929d22b1046862aad710a0bd8f09c0d0a0},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.11342. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

We evaluate whether an AI agent's harness, the notes around its model, can learn from the agent's failure traces beyond a matched search that never sees them. It can, but less than it looks. We tested an open 9-billion-parameter model with fixed weights; a larger teacher model wrote its notes. On generated customer-service decisions, where both note writers were told the target rules, notes based on the agent's mistakes improved accuracy by 4.0 percentage points over notes from the same process with the mistakes withheld. That process alone, trying and keeping better-scoring notes, reached about 60% of the total improvement. In full customer-service conversations the same notes showed no demonstrated improvement. Separately, a strict screen of single lessons admitted two customer-service lessons, still awaiting tests on new data, and one tool-use lesson, written without seeing the agent's mistakes, which showed no demonstrated gain on new tool-use tasks. A neutral note with no procedure scored higher than the learned lessons (exploratory). A blind audit judged 146 of 240 tool-use test items defective.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI); Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2611.11342 — 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 2026-11-10T18:37:37.637Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).