Metraipolis › archive › dc_0d699eae0aa9

Same Trajectory, Contradictory Rewards (ROBORMBENCH): Paraphrase Fragility in Vision Language Reward Models — arXiv:2609.05401

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.05401
License (as archived)
CC0-1.0
Archived
2026-10-16T22:36:07.501Z
SHA-256 of the archived text
be3ecc94b27b9abf87ebc96aa63fa6ab1b4b3cd37068e5dc94466ce92bbdf611
Stable URL of this page
https://metraipolis.com/documento/dc_0d699eae0aa9

How to cite

Same Trajectory, Contradictory Rewards (ROBORMBENCH): Paraphrase Fragility in Vision Language Reward Models — arXiv:2609.05401 — arXiv (en), https://arxiv.org/abs/2609.05401 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0d699eae0aa9 (Metraipolis archive dc_0d699eae0aa9, sha256 be3ecc94b27b9abf87ebc96aa63fa6ab1b4b3cd37068e5dc94466ce92bbdf611)
BibTeX
@misc{dc0d699eae0aa9,
  title = {Same Trajectory, Contradictory Rewards (ROBORMBENCH): Paraphrase Fragility in Vision Language Reward Models — arXiv:2609.05401},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.05401},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0d699eae0aa9; sha256 be3ecc94b27b9abf87ebc96aa63fa6ab1b4b3cd37068e5dc94466ce92bbdf611},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.05401. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Vision-language models are increasingly used as reward functions for robotic learning, but this role requires paraphrase invariance: the same trajectory should receive the same reward under semantically equivalent goal descriptions. We show that current VLM reward models often violate this property. Paraphrasing the instruction alone can substantially change predicted progress scores, and can even flip identical robot behavior between failure and success. To measure this failure mode, we introduce ROBORMBENCH, a benchmark with 2,390 real-robot trajectories, ground-truth progress labels, and 21,673 verified paraphrases spanning lexical, syntactic, and action-goal rewrites. Across proprietary and open-source VLMs, paraphrase-induced instability is widespread and severe, grows under more divergent rewrites, and is not reliably reduced by scale or explicit reasoning. Dedicated reward models trained with trajectory-grounded supervision are substantially more stable. These results show that paraphrase robustness is a core requirement for reliable VLM-based reward modeling in robotics.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-07 in cs.CL).
Categorie arXiv: Robotics (cs.RO) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.05401 — 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 2026-10-16T22:21:16.329Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-07?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).