Metraipolis › archive › dc_0bded0f397d3

When Faster VLA Deployment Changes Closed-Loop Behavior: Task Success-Latency Analysis of SmolVLA Across PyTorch and ONNX Variants — arXiv:2609.14146

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.14146
License (as archived)
CC0-1.0
Archived
2026-10-16T00:15:29.693Z
SHA-256 of the archived text
68dda178eb768f2b560d0ecf041d9d053781c380f3f748c62604ad41970ca32b
Stable URL of this page
https://metraipolis.com/documento/dc_0bded0f397d3

How to cite

When Faster VLA Deployment Changes Closed-Loop Behavior: Task Success-Latency Analysis of SmolVLA Across PyTorch and ONNX Variants — arXiv:2609.14146 — arXiv (en), https://arxiv.org/abs/2609.14146 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0bded0f397d3 (Metraipolis archive dc_0bded0f397d3, sha256 68dda178eb768f2b560d0ecf041d9d053781c380f3f748c62604ad41970ca32b)
BibTeX
@misc{dc0bded0f397d3,
  title = {When Faster VLA Deployment Changes Closed-Loop Behavior: Task Success-Latency Analysis of SmolVLA Across PyTorch and ONNX Variants — arXiv:2609.14146},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.14146},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0bded0f397d3; sha256 68dda178eb768f2b560d0ecf041d9d053781c380f3f748c62604ad41970ca32b},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.14146. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Vision-language-action (VLA) deployment can reduce inference latency while changing closed-loop task behavior. We evaluate HuggingFaceVLA/smolvla_libero on an RTX 2060 (6 GB) in LIBERO Spatial and Object (MuJoCo 3.3.2, LeRobot 0.6.1, seed 42), comparing PyTorch+AMP with ONNX Runtime CUDA Execution Provider (CUDA EP). The main evaluation uses 100 episodes/suite; a paired rollout uses 300 episodes/suite. PyTorch+AMP reaches 70.0%/88.0% Spatial/Object success at 1181 ms p99. Requested-FP16 and requested-INT8 ONNX reduce tether-inspect p99 to 601 ms and 532 ms, while Spatial success falls to 41.0% and 40.0% and Object remains at 89.0%. A graph audit shows those artifacts are byte-identical FP32 graphs, so the requested-INT8 row is not operator-level INT8 quantization. A static language-width ablation (16/24/32 tokens) yields Spatial success of 41.0%, 75.0%, and 71.0%; widths 24 and 32 recover much of the Spatial drop while Object success and uniform-bench latency stay approximately stable. Width-24 ONNX Spatial success is comparable to the PyTorch+AMP baseline at roughly half the latency (Wilson intervals overlap; two-proportion chi-squared p=0.53). Context width is an important contributor in this stack; it does not account for every PyTorch-vs-ONNX difference. Deployment evaluation should jointly report latency, artifact inspection, interface constraints, and closed-loop success. Code: this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-15 in cs.LG).
Categorie arXiv: Robotics (cs.RO) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.14146 — 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 2026-10-15T23:54:14.892Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-15?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).