NinA: Normalizing Flows in Action. Training VLA Models with Normalizing Flows

2025-08-22

AI Summary

Diffusion models for vision-language-action (VLA) tasks suffer from high inference latency due to iterative denoising, hindering real-time, high-frequency robotic control. To address this, we propose NinA—the first VLA framework to integrate Normalizing Flows into the action decoder, enabling an invertible, single-step action generation architecture. NinA is end-to-end jointly fine-tuned with a pre-trained vision-language model, eliminating iterative sampling while preserving task success rates comparable to diffusion-based methods. Crucially, it achieves one-shot action prediction, drastically reducing inference latency. On the LIBERO benchmark, NinA attains state-of-the-art performance across multiple tasks and accelerates inference by several-fold—demonstrating practical viability for real-time robotic control applications.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Machine Learning: Imitation Learning & Inverse Reinforcement Learning

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Search and Retrieval-Augmented AI: Large language models for search
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Recent advances in Vision-Language-Action (VLA) models have established a two-component architecture, where a pre-trained Vision-Language Model (VLM) encodes visual observations and task descriptions, and an action decoder maps these representations to continuous actions. Diffusion models have been widely adopted as action decoders due to their ability to model complex, multimodal action distributions. However, they require multiple iterative denoising steps at inference time or downstream techniques to speed up sampling, limiting their practicality in real-world settings where high-frequency control is crucial. In this work, we present NinA (Normalizing Flows in Action), a fast and expressive alter- native to diffusion-based decoders for VLAs. NinA replaces the diffusion action decoder with a Normalizing Flow (NF) that enables one-shot sampling through an invertible transformation, significantly reducing inference time. We integrate NinA into the FLOWER VLA architecture and fine-tune on the LIBERO benchmark. Our experiments show that NinA matches the performance of its diffusion-based counterpart under the same training regime, while achieving substantially faster inference. These results suggest that NinA offers a promising path toward efficient, high-frequency VLA control without compromising performance.

Problem

• Replacing diffusion models for faster VLA action decoding
• Enabling one-shot sampling to reduce inference time
• Achieving efficient high-frequency control without performance loss

Innovation

• Replaces diffusion decoder with Normalizing Flow
• Enables one-shot sampling through invertible transformation
• Achieves faster inference without performance compromise