Metraipolis › archive › dc_68ff98681cd5

Decoding Task Progress from VLA Representations — arXiv:2608.13474

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.13474
License (as archived)
CC0-1.0
Archived
2026-10-19T04:56:34.121Z
SHA-256 of the archived text
96ff3dbabe5a663b0e23530221f2727e18f96c5fbd7ad3336b79932704e05baf
Stable URL of this page
https://metraipolis.com/documento/dc_68ff98681cd5

How to cite

Decoding Task Progress from VLA Representations — arXiv:2608.13474 — arXiv (en), https://arxiv.org/abs/2608.13474 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_68ff98681cd5 (Metraipolis archive dc_68ff98681cd5, sha256 96ff3dbabe5a663b0e23530221f2727e18f96c5fbd7ad3336b79932704e05baf)
BibTeX
@misc{dc68ff98681cd5,
  title = {Decoding Task Progress from VLA Representations — arXiv:2608.13474},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.13474},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_68ff98681cd5; sha256 96ff3dbabe5a663b0e23530221f2727e18f96c5fbd7ad3336b79932704e05baf},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.13474. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Vision-language-action models (VLAs) are moving rapidly towards deployment as general-purpose manipulation policies, but we currently lack basic tools for understanding what these models represent internally or for monitoring them at runtime. Leveraging ideas from mechanistic interpretability, we probe the residual stream of $\pi_{0.5}$ and find that task progress, the normalized time remaining in a trajectory, is linearly readable from the activations. We find that this signal is present in the pretrained PaliGemma backbone prior to training on any robot-specific data. A single linear probe generalizes to unseen tasks and varies under language counterfactuals when trained on multi-prompt data, but does not enable meaningful steering of the policy. These properties make the signal directly useful for instrumenting deployed VLAs. We use the probe as a simple label-free OOD detector, which detects stalled task progress, and find it competitive with state-of-the-art methods. Our results suggest that VLAs have rich, linearly readable internal representations of semantic quantities like task progress, and that learning to read these signals offers a lightweight, interpretable path toward monitoring deployed visuomotor policies.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-08-14 in cs.RO).
Categorie arXiv: Robotics (cs.RO).
Pagina del lavoro: https://arxiv.org/abs/2608.13474 — 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 2026-10-19T04:47:48.740Z dalla pagina https://arxiv.org/catchup/cs.RO/2026-08-14?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).