Fast-dVLA: Accelerating Discrete Diffusion VLA to Real-Time Performance

Wenxuan Song; Jiayi Chen; Shuai Chen; Jingbo Wang; Pengxiang Ding; Han Zhao; Yikai Qin; Xinhu Zheng

cs.ROMar 26, 2026

Fast-dVLA: Accelerating Discrete Diffusion VLA to Real-Time Performance

Authors: Wenxuan Song, Jiayi Chen, Shuai Chen, Jingbo Wang, Pengxiang Ding, Han Zhao, Yikai Qin, Xinhu Zheng, +3 more

Organizations: The Hong Kong University of Science and Technology (Guangzhou) · ShanghaiTech University · Shanghai Institute of Technical Physics, CAS · Westlake University · Zhejiang University · AIR, Tsinghua University

Abstract

This paper proposes a novel approach to address the challenge that pretrained VLA models often fail to effectively improve performance and reduce adaptation costs during standard supervised finetuning (SFT). Some advanced finetuning methods with auxiliary training objectives can improve performance and reduce the number of convergence steps. However, they typically incur significant computational overhead due to the additional losses from auxiliary tasks. To simultaneously achieve the enhanced capabilities of auxiliary training with the simplicity of standard SFT, we decouple the two objectives of auxiliary task training within the parameter space, namely, enhancing general capabilities and fitting task-specific action distributions. To deliver this goal, we only need to train the model to converge on a small-scale task set using two distinct training strategies. The difference between the resulting model parameters can then be interpreted as capability vectors provided by auxiliary tasks. These vectors are then merged with pretrained parameters to form a capability-enhanced meta model. Moreover, when standard SFT is augmented with a lightweight orthogonal regularization loss, the merged model attains performance comparable to auxiliary finetuned baselines with reduced computational overhead. Experimental results demonstrate that this approach is highly effective across diverse robot tasks. Project page: https://chris1220313648.github.io/Fast-dVLA/

Figures & tables

Figure 1: Speed/Success Rate trade-off. Left (Intra-comparison): Compared to other acceleration strategies for discrete diffusion VLAs (dVLAs), DD-VLA ( Liang et al., 2025b ) and Dream-VLA ( Ye et al., 2025a ) , our Fast-dVLA achieves a favorable success rate and speed. Here, BlockDiff denotes block diffusion ( Arriola et al., 2025 ) . Right (Inter-comparison): Our Fast-dVLA surpasses autoregressive methods, i.e. , $\pi_{0}$ -FAST ( Pertsch et al., 2025 ) . It also reaches parallel performance and inference frequency with state-of-the-art (SOTA) continuous flow-matching methods, i.e. , $\pi_{0.5}$ ( Intelligence et al., 2025 ) , while maintaining several inherent advantages of dVLAs. We report metrics on LIBERO ( Liu et al., 2023 ) .

Figure 2: Comparison among discrete decoding paradigms. Here, Forward per Sequence denotes the needed forward numbers for a full sequence output, Forward Speed denotes the decoding speed for each forward, and Speed per Sequence (i.e., Inference Speed) denotes the decoding speed for the full sequence output. Our Fast-dVLA requires significantly fewer forward passes and executes each pass efficiently, resulting in substantially faster inference.

Figure 3: Visualization of the decoding tendency of action tokens at different positions in Dream-VLA ( Ye et al., 2025a ) . Brighter regions indicate higher decoding probability. Despite using bidirectional attention, the model exhibits a clear left-to-right decoding tendency such that action tokens at earlier temporal positions are typically decoded in earlier diffusion iterations. Overall, the decoding process reveals an implicit block-wise AR pattern.

Panel of Figure 4 — Figure 4: KV cache similarity across diffusion iterations under block-diffusion decoding. We visualize the similarity of attention key–value (KV) states for the first action block across different denoising steps. (a): In native dVLA with bidirectional attention, the KV representations evolve across iterations, preventing effective reuse of cached states. (b): In contrast, after adapting dVLA to a block-wise attention architecture via asymmetric distillation, once all tokens in the first block are unmasked, the corresponding KV states remain fixed, enabling efficient KV cache reuse and substantially reducing the computational overhead in subsequent iterations.

Figure 5: Block-wise Attention.

Figure 6: Overview of the pipelined parallel decoding in Fast-dVLA . Blocks are processed concurrently in a dynamically growing pipeline. A new block is introduced when the tail block exceeds the addition threshold

\tau_{\text{add}}=\tfrac{2}{7}

, and becomes fully activated after its predecessor surpasses the activation threshold

\tau_{\text{act}}=\tfrac{4}{7}

Decoding Method	Success Rate $\uparrow$					Speed $\uparrow$
Decoding Method	Spatial	Goal	Object	Long	Avg.	(Tokens/s)
Dream-VLA ( Ye et al., 2025a )	0.902	0.920	0.880	0.720	0.856	98.8 ( ×1.0 )
+ Fast-dLLM	0.884	0.894	0.834	0.702	0.828	183.2 ( ×1.9 )
+ Block Diffusion	0.918	0.904	0.886	0.722	0.858	181.7 ( ×1.8 )
+ Fast-dVLA (ours)	0.912	0.920	0.902	0.746	0.870	313.1 ( ×3.2 )
Discrete Diffusion VLA ( Liang et al., 2025b )	0.972	0.986	0.974	0.920	0.963	152.1 ( ×1.5 )

Table 1: Comparison between various acceleration strategies on two base models in terms of task-wise success rates (SR) and inference speed on LIBERO.

Decoding Method	Task	Tasks Completed in a Row $\uparrow$					Avg.	Speed $\uparrow$
Decoding Method	Task	1/5	2/5	3/5	4/5	5/5	Len. $\uparrow$	(Tokens/s)
UD-VLA	ABCD $\rightarrow$ D	0.992	0.968	0.936	0.904	0.840	4.64	67.3 ( ×1.0 )
+ Fast-dLLM	ABCD $\rightarrow$ D	0.972	0.920	0.858	0.808	0.762	4.32	132.5 ( ×2.0 )
+ Block Diffusion	ABCD $\rightarrow$ D	0.988	0.944	0.894	0.862	0.804	4.50	129.5 ( ×1.9 )
+ Fast-dVLA (ours)	ABCD $\rightarrow$ D	0.984	0.952	0.922	0.870	0.812	4.54	186.7 ( ×2.8 )

Table 2: Comparison between various acceleration strategies on UD-VLA in terms of tasks completed in a row, average sentence length, and inference speed on CALVIN.

Method	Task	Tasks Completed in a Row					Avg. Len. $\uparrow$
Method	Task	1/5	2/5	3/5	4/5	5/5	Avg. Len. $\uparrow$
RT-1 ( Brohan et al., 2023 )	ABCD $\rightarrow$ D	0.844	0.617	0.438	0.323	0.227	2.45
LLaDA-VLA ( Wen et al., 2025c )	ABCD $\rightarrow$ D	0.956	0.878	0.795	0.739	0.645	4.01
Deer ( Yue et al., 2024b )	ABCD $\rightarrow$ D	0.982	0.902	0.821	0.759	0.670	4.13
GR-1 ( Wu et al., 2024 )	ABCD $\rightarrow$ D	0.949	0.896	0.844	0.789	0.731	4.21
ReconVLA ( Song et al., 2025c )	ABCD $\rightarrow$ D	0.980	0.900	0.845	0.785	0.705	4.23

Table 3: Comprehensive evaluation of long-horizon manipulation on the CALVIN benchmark. UniVLA ∗ denotes the variant without historical frames for fair comparison.

Method	Spoon on Towel		Carrot on Plate		Stack Green Block		Eggplant in Basket		Average
Method	Grasp	Success	Grasp	Success	Grasp	Success	Grasp	Success	Success	Speed
RoboVLM ( Liu et al., 2025a )	37.5	20.8	33.3	25.0	8.3	8.3	0.0	0.0	13.5	-
SpatialVLA ( Qu et al., 2025 )	20.8	16.7	29.2	25.0	62.5	29.2	100.0	100.0	42.7	-
OpenVLA-OFT ( Kim et al., 2025a )	50.0	12.5	41.7	4.2	70.8	20.8	91.7	37.5	18.8	-
$\pi_{0}$ ( Black et al., 2024 )	45.8	29.1	25.0	0.0	50.0	16.6	91.6	62.5	27.1	-
$\pi_{0}$ -FAST ( Pertsch et al., 2025 )	62.5	29.1	58.5	21.9	54.0	10.8	83.3	66.6	32.1	107.5

Table 4: Evaluation on WidowX Robot tasks in SimplerEnv. We report the Grasping Success Rate (Grasp) and Task Success Rate (Success) in percentages (%). Besides, we further report the decode speed (Speed) of all VLAs with discrete output, which is calculated in tokens per second.

Figure 7: Real-world experiment results. We report (a) successful grasps per minute ( $\uparrow$ ); (b) success rates ( $\uparrow$ ) and completion times ( $\downarrow$ ); (c) execution frequency ( $\uparrow$ ).

Figure 8: Action Mean Squared Error (MSE) of the dVLA at varying training steps on LIBERO. The MSE of our asymmetric distillation exhibits the fastest decline, indicating the most rapid convergence speed.

Figure 9: Ablation study on the confidence threshold $\tau_{\text{conf}}$ of Fast-dVLA based on UD-VLA.

Block	Success Rate	Speedup
Multiples	74.7%	4.01 $\times$
Ramdom	73.3%	3.95 $\times$

Table 5: Choice of block size on LIBERO-Long based on Dream-VLA. Here, multiples denotes that the value is the multiples of the dimensionality of action, while random denotes random numbers for the block size.

$\tau_{\text{add}}$	$\tau_{\text{act}}$	avg. len. $\uparrow$	Speed $\uparrow$
0.4	0.4	4.42	253.1
0.4	0.6	4.46	248.3
0.5	0.5	4.44	204.8
0.5	0.7	4.54	186.7
0.6	0.6	4.52	182.3
0.6	0.8	4.57	160.7

Table S1: Ablation study on block expansion threshold

\tau_{\text{add}}

and block activation threshold

\tau_{\text{act}}

on UD-VLA.

Radical Decoding	Speed (Tokens/s)	Avg. Len.
log2	186.67	4.54
log3	164.42	4.57
log4	144.71	4.58

Table S2: Comparison with various base models in terms of inference speed, average length, and execution frequency.

Method	Spatial	Object	Goal	Long	Average
Octo ( Octo Team et al., 2024 )	78.9%	85.7%	84.6%	51.1%	75.1%
SpatialVLA ( Qu et al., 2025 )	88.2%	89.9%	78.6%	55.5%	78.1%
CoT-VLA ( Zhao et al., 2025 )	87.5%	91.6%	87.6%	69.0%	81.1%
WorldVLA ( Cen et al., 2025 )	87.6%	96.2%	83.4%	60.0%	81.8%
ThinkAct ( Huang et al., 2025 )	88.3%	91.4%	87.1%	70.9%	84.4%
$\pi_{0}$ -FAST ( Pertsch et al., 2025 )	96.4%	96.8%	88.6%	60.2%	85.5%

Table S3 : Evaluation and comparison on the LIBERO benchmark.

Explore similar work

CapVector: Learning Transferable Capability Vectors in Parametric Space for Vision-Language-Action Models
May 11, 2026Wenxuan Song, Han Zhao, Fuhao Li +7Task Vectors VLM Adaptation
1HKUST (GZ) · 2Zhejiang University · 3Westlake University +2
BlockVLA: Accelerating Autoregressive VLA via Block Diffusion Finetuning
May 13, 2026Ruiheng Wang, Shuanghao Bai, Haoran Zhang +2Robotic Manipulation Vision-Language-Action Models
Xi'an Jiaotong University
dVLA-RL: Reinforcement Learning over Denoising Trajectories for Discrete Diffusion Vision-Language-Action Models
Jun 22, 2026Yuhao Wu, Yitian Liu, Weijie Shen +13Vision-Language-Action Models VLMs for Robotics
ScaleLab at SJTU · 1Shanghai Jiao Tong University · 3Baidu AI Cloud +3

May 11, 2026J/K move · Enter open · S save

CapVector: Learning Transferable Capability Vectors in Parametric Space for Vision-Language-Action Models

Wenxuan Song, Han Zhao, Fuhao Li, Ziyang Zhou+6

1HKUST (GZ) · 2Zhejiang University · 3Westlake University+2

Task Vectors VLM Adaptation Transfer Learning Vision-Language-Action Models Model Merging

Open paper AskPDF arXiv

This paper proposes a novel approach to address the challenge that pretrained VLA models often fail to effectively improve performance and reduce adaptation costs during standard supervised finetuning (SFT). Some advanced finetuning methods with auxiliary training objectives can improve performance and reduce the number of convergence steps. However, they typically incur significant computational overhead due to the additional losses from auxiliary objectives. To simultaneously achieve the enhanced capabilities of auxiliary training with the simplicity of standard SFT, we decouple the two objectives of auxiliary-objective SFT within the parameter space, namely, enhancing general capabilities and fitting task-specific action distributions. To deliver the goal, we only need to train the model to converge on a small-scale task set using two distinct training strategies, resulting in two finetuned models. The parameters' difference between the two models can then be interpreted as capability vectors provided by auxiliary objectives. These vectors are then merged with pretrained parameters to form a capability-enhanced meta model. Moreover, when standard SFT is augmented with a lightweight orthogonal regularization loss, the merged model attains performance comparable to auxiliary finetuned baselines with reduced computational overhead. Internal and external experiments demonstrate that our capability vectors (1) are effective and versatile across diverse models, (2) can generalize to novel environments and embodiments out of the box.

Abstract

Explore similar work

CapVector: Learning Transferable Capability Vectors in Parametric Space for Vision-Language-Action Models

BlockVLA: Accelerating Autoregressive VLA via Block Diffusion Finetuning

dVLA-RL: Reinforcement Learning over Denoising Trajectories for Discrete Diffusion Vision-Language-Action Models