Metraipolis › archive › dc_c23b7c19fc7a

STAR: Steering Action Representations during Inference — arXiv:2610.41009

Source: arXiv (en)
Original: https://arxiv.org/abs/2610.41009
License (as archived): CC0-1.0
Archived: 2026-11-03T07:32:36.464Z
SHA-256 of the archived text: 970d27917d688b72d919d09e07887bc39a43d233a6e2c85301edabd141b535dc
Stable URL of this page: https://metraipolis.com/documento/dc_c23b7c19fc7a

How to cite

STAR: Steering Action Representations during Inference — arXiv:2610.41009 — arXiv (en), https://arxiv.org/abs/2610.41009 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_c23b7c19fc7a (Metraipolis archive dc_c23b7c19fc7a, sha256 970d27917d688b72d919d09e07887bc39a43d233a6e2c85301edabd141b535dc)

BibTeX

@misc{dcc23b7c19fc7a,
  title = {STAR: Steering Action Representations during Inference — arXiv:2610.41009},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.41009},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_c23b7c19fc7a; sha256 970d27917d688b72d919d09e07887bc39a43d233a6e2c85301edabd141b535dc},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.41009. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Vision-language-action models (VLAs) map language instructions and visual observations to robot actions, showing strong capabilities across robotic manipulation tasks. However, in our probing experiment, we find that action representations near the action output do not fully preserve task-relevant visual grounding. Motivated by this late-layer grounding gap, we introduce STAR, for STeering Action Representations, a training-free inference method for pre-trained VLAs. STAR applies an exponential moving average over late-layer residual updates, allowing later action representations to inherit object-aware update directions from earlier layers without model updates or architectural changes. It further uses recent proprioceptive states to adapt the mixing strength, using conservative mixing when execution dynamics change rapidly. Experiments on robot manipulation benchmarks show that STAR improves late-layer object awareness and increases average task success rates by 4.0 percentage points. Our code is available at this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Robotics (cs.RO) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.41009 — 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 2026-11-03T07:29:33.698Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).