Metraipolis › archive › dc_ab2201c27ae1

Anatomy of a Closed-Loop Collapse: A Causal Case Study of a Compressed VLA Policy — arXiv:2609.23048

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.23048
License (as archived)
CC0-1.0
Archived
2026-10-09T07:08:22.405Z
SHA-256 of the archived text
6d6376318318a5a6d5992e8c6725a1e74c62bfd6109e159332fa7ab9f127e94a
Stable URL of this page
https://metraipolis.com/documento/dc_ab2201c27ae1

How to cite

Anatomy of a Closed-Loop Collapse: A Causal Case Study of a Compressed VLA Policy — arXiv:2609.23048 — arXiv (en), https://arxiv.org/abs/2609.23048 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ab2201c27ae1 (Metraipolis archive dc_ab2201c27ae1, sha256 6d6376318318a5a6d5992e8c6725a1e74c62bfd6109e159332fa7ab9f127e94a)
BibTeX
@misc{dcab2201c27ae1,
  title = {Anatomy of a Closed-Loop Collapse: A Causal Case Study of a Compressed VLA Policy — arXiv:2609.23048},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.23048},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ab2201c27ae1; sha256 6d6376318318a5a6d5992e8c6725a1e74c62bfd6109e159332fa7ab9f127e94a},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.23048. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Compressed manipulation policies can pass offline evaluation while failing in closed-loop execution; this dissociation is established in prior work and is not our claim. We contribute a causal anatomy of one naturally occurring case. An 8-layer distillation of Octo-Base retains 86% of parameters, passes every offline check we applied (0.996 and 1.000 teacher-ratios on the family's own validation metrics), and collapses in closed loop: 0/72 vs. the teacher's 40/72 on a simulated WidowX pick-and-place task. The collapse is structured, not diffuse: early task stages degrade gradually (the student moves the object at 90% of the teacher's rate and grasps at 55%), while transport-to-target fails categorically, at 0% in every training variant. Paired action-trace forensics isolate the signature: a negative, late-heavy $z$ residual, roughly 10x its post-repair magnitude, and persistent across the base distillation and both continuation branches. Four standard therapies fail under matched controls: continued training and in-domain offline data leave success at zero, even though the latter measurably improves marginal action statistics; command-level compensation recovers nothing at any offset, although the same perturbations degrade healthy policies; clamping the symptom in the command channel preserves grasping, yet success stays at floor. A minimal-pair intervention that substitutes half of the training stream with deployment-distribution teacher rollouts, with every other setting held fixed, restores parity with the teacher (18/36 vs. 17/36 held-out), eliminates that signature, and recovers a teacher-like perturbation-response profile. We claim existence, not universality. Operationally, offline gates, including a family's own validation metrics, are insufficient acceptance tests for compressed policies; a few dozen closed-loop trials sufficed to find what they missed.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-22 in cs.LG).
Categorie arXiv: Robotics (cs.RO) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.23048 — 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 2026-10-09T06:58:09.549Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-22?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).