Metraipolis › archive › dc_aeae6ff5264a
EAVer: Long-Form Factuality Verification as an End-to-End Agentic Policy — arXiv:2609.22223
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.22223
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-09T06:49:20.048Z
- SHA-256 of the archived text
c25ccb8e917e5d700ad96bbb369ed72c2f44a1a6158d0c2ab3006f4babd73646- Stable URL of this page
https://metraipolis.com/documento/dc_aeae6ff5264a
How to cite
EAVer: Long-Form Factuality Verification as an End-to-End Agentic Policy — arXiv:2609.22223 — arXiv (en), https://arxiv.org/abs/2609.22223 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_aeae6ff5264a (Metraipolis archive dc_aeae6ff5264a, sha256 c25ccb8e917e5d700ad96bbb369ed72c2f44a1a6158d0c2ab3006f4babd73646)
BibTeX
@misc{dcaeae6ff5264a,
title = {EAVer: Long-Form Factuality Verification as an End-to-End Agentic Policy — arXiv:2609.22223},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.22223},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_aeae6ff5264a; sha256 c25ccb8e917e5d700ad96bbb369ed72c2f44a1a6158d0c2ab3006f4babd73646},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.22223. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Long-form factuality verification is commonly implemented as a static decompose-search-verify pipeline, with separately prompted modules processing claims and invoking external search. Treating claims independently makes LLM and search calls scale with claim count and causes repeated searches for overlapping evidence about related claims. We introduce EAVer, an End-to-end Agentic Verifier that learns to control the complete response-level verification workflow as a unified policy. EAVer groups semantically related claims, routes each group to direct verification or targeted search based on confidence, and keeps evidence returned by search in compact in-context memos for cross-claim reuse. To train this policy, we develop a privileged-teacher synthesis pipeline that converts gold claim annotations into executable multi-turn tool-interaction trajectories with live search rather than post-hoc rationales. Structural, label-alignment, tool-use, search-budget, and leakage checks yield 1,447 quality-controlled trajectories. We further construct 794 bidirectional same-trajectory preference pairs that keep claim grouping, search, and evidence fixed, enabling decision-focused Direct Preference Optimization (DPO) over factuality-decision tokens. The results with Qwen3-8B show that EAVer outperforms the strongest search-based baseline on each benchmark by 2.88 Macro-F1 points on VeriFastScore and 4.73 points on the out-of-distribution FaStFact-Bench, while using about 80% fewer searches than the most search-efficient baseline. Moreover, EAVer consistently improves performance across models ranging from 4B to 32B parameters, demonstrating its strong generalizability. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-22 in cs.CL). Categorie arXiv: Computation and Language (cs.CL) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.22223 — 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 2026-10-09T06:48:44.899Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-22?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).