aiqbench analysis
This paper proposes a streamlined approach for Vision-Language-Action (VLA) models by biasing diffusion training schedules toward high-noise states, eliminating the need for complex teacher models or distillation stages. Experimental results on standard and long-horizon robotics benchmarks demonstrate that this simple modification allows one-step action generation to match or exceed the performance of traditional multi-step decoding methods. For practitioners, this suggests that effective one-step policies can emerge from standard training pipelines without importing specialized image-synthesis machinery.
Let It Be Simple: One-Step Action Generation for Vision-Language-Action Models
arXiv:2606.05737v1 Announce Type: cross Abstract: Diffusion-based vision-language-action (VLA) models often inherit the image-generation view: actions are generated by iterative denoising. We argue that VLA action generation has a different condition-target structure: the policy is conditioned on rich observations, language, and state, but predicts only a compact, low-dimensional action chunk. Under this asymmetry, strong one-step action generation should not necessarily require the advanced one-step methods developed for image synthesis. We keep standard velocity prediction and add no teacher model, distillation stage, or auxiliary objective; in our main recipe, we simply bias the training time distribution toward high-noise states. We first isolate the effect in a controlled MNIST grid-to-sequence task, then test it with extensive robot-policy experiments. Across standard LIBERO, LIBERO-Plus, and LIBERO-Pro, one-step policies trained with high-noise biased schedules generally match ten-step decoding under the same recipe, and on standard LIBERO can exceed ten-step policies trained with a uniform time distribution. A real-robot bimanual YAM RSS evaluation gives a small-sample cross-architecture check of the same sampler trend. On a 1.4B VLM model with a 30M action head, one-step decoding reaches 95.6\% on LIBERO-Long. These results show that strong one-step VLA action generation can emerge from standard diffusion training, without importing the full few-step diffusion machinery developed for image generation.
Abstract:Diffusion-based vision-language-action (VLA) models often inherit the image-generation view: actions are generated by iterative denoising. We argue that VLA action generation has a different condition-target structure: the policy is conditioned on rich observations, language, and state, but predicts only a compact, low-dimensional action chunk. Under this asymmetry, strong one-step action generation should not necessarily require the advanced one-step methods developed for image synthesis. We keep standard velocity prediction and add no teacher model, distillation stage, or auxiliary objective; in our main recipe, we simply bias the training time distribution toward high-noise states. We first isolate the effect in a controlled MNIST grid-to-sequence task, then test it with extensive robot-policy experiments. Across standard LIBERO, LIBERO-Plus, and LIBERO-Pro, one-step policies trained with high-noise biased schedules generally match ten-step decoding under the same recipe, and on standard LIBERO can exceed ten-step policies trained with a uniform time distribution. A real-robot bimanual YAM RSS evaluation gives a small-sample cross-architecture check of the same sampler trend. On a 1.4B VLM model with a 30M action head, one-step decoding reaches 95.6\% on LIBERO-Long. These results show that strong one-step VLA action generation can emerge from standard diffusion training, without importing the full few-step diffusion machinery developed for image generation.
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Robotics (cs.RO)
arXiv:2606.05737 [cs.CV]
(or arXiv:2606.05737v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2606.05737 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 308 words
Related coverage
Dyna-2: General Purpose Physical Agents with Hour-Long Autonomy — Jason Ma, Dyna Robotics
Latent Space · Oct 13, 2026
Links #5: 2026/07
LessWrong · Jul 31, 2026
Runway’s WorldPrompt and the Engineering of Real-Time Worlds
Latent Space · Sep 25, 2026
[AINews] The Future of Latent Space
Latent Space · Sep 25, 2026