1 of 17

Error as Signal: Stiffness-Aware Diffusion Sampling via Embedded Runge-Kutta Guidance

Inho Kong*, Sojin Lee*, Youngjoon Hong†, Hyunwoo J. Kim†

1

MLV Lab

2 of 17

2

Background

Gaussian Noise

Generated Data

Diffusion ODE formulation

Diffusion sampling can be formulated as solving an ordinary differential equation (ODE).

Sampling errors stem from two main sources: model prediction and integration.

MLV Lab

3 of 17

3

Model-induced Error

Solver-induced Error�

Overview

Inaccurate Score Estimation

Models are trained to estimate the score function via score matching loss.

Inaccurate ODE Integration�We rely on numerical ODE solvers since the diffusion ODE lacks an exact solution.

Extrapolate outputs from two models

(e.g., CFG, AG)

Extrapolate outputs from two solvers

(Our Method)

Target Problem

Referred to as

Mitigation Strategy

Conventional guidance methods primarily address model-induced errors.

ERK-Guid directly targets solver-induced local truncation errors to improve generation quality.

MLV Lab

4 of 17

4

Key Intuition (Drift Field)

Stiffness in diffusion ODEs arises when drift directions change rapidly.

A 2D drift field illustrates stiff and non-stiff regions.

Drift Field

non-stiff region

stiff region

MLV Lab

5 of 17

5

Key Intuition (Local Truncation Error)

Local Truncation Error (LTE)

Drift Field

In these stiff regions, a significantly larger LTE occurs during sampling.

MLV Lab

6 of 17

6

Key Intuition (Eigenvector Alignment)

Local Truncation Error (LTE)

Dominant Eigenvector Direction

We observe that in stiff regions, the LTE closely aligns with the dominant eigenvector of the Jacobian.

This alignment provides a reliable directional proxy to correct solver-induced errors.

MLV Lab

7 of 17

7

Method Formulation (Cost-free Estimators)

Calculating exact stiffness and eigenvectors via Jacobian-vector products is inefficient for practical sampling.

We leverage Embedded Runge-Kutta (ERK) pairs to obtain cost-free estimators.

Embedded Runge-Kutta pair

Stiffness Estimator

Dominant Eigenvector Estimator

MLV Lab

8 of 17

8

Method Formulation (ERK-Guid)

 

Adaptive Guidance Scale�determined by stiffness and hyperparameters

Step-size

Guidance Vectordrift differences between distinct solver states

MLV Lab

9 of 17

9

Experiments

Accuracy of the estimators

Our estimated stiffness has a strong correlation with the exact stiffness.

MLV Lab

10 of 17

10

Experiments

Accuracy of the estimators

The cosine similarity between ERK-Guid and the eigenvector is notably high in high-stiffness regions.

These results demonstrate that our estimator is sufficiently accurate.

MLV Lab

11 of 17

11

Experiments

ERK-Guid consistently improves all sample quality metrics without any extra inference cost.

Notably, the performance gains are most dramatic at low sampling steps where solver errors dominate.

Effectiveness of ERK-Guid

MLV Lab

12 of 17

12

Experiments

ERK-Guid provides an orthogonal guidance signal with existing methods such as CFG and Autoguidance. Integrating our method improves sample quality metrics across different sampling steps.

Guidance compatibility

MLV Lab

13 of 17

13

Experiments

ERK-Guid can be integrated into numerical solvers such as Heun, DPM-Solver, and DEIS.

Applying our method as a plug-and-play module improves generation quality across baseline solvers.

Plug-and-play adaptation

MLV Lab

14 of 17

14

Experiments

Comparison with adaptive step-size control

Conventional adaptive solvers resize the step size when stiffness exceeds a specific threshold.

While this approach improves performance, ERK-Guid achieves better sample quality without extra NFEs.

MLV Lab

15 of 17

15

Experiments

ERK-Guid calculates the guidance vector without additional network evaluations.

As a result, it introduces minimal inference overhead and maintains the same memory consumption

Wall-clock time (seconds per image) and Memory consumption

MLV Lab

16 of 17

16

Experiments

Qualitative result demonstrates how ERK-Guid helps reduce structural distortions.

It generates an image with much better preserved details than the baseline.

Qualitative result

DPM Solver

DPM Solver + ERK-Guid

MLV Lab

17 of 17

17

Summary

  • A New Perspective on Solver Errors: Repurposing local truncation errors (LTE) in stiff regions as informative guidance signals.

  • Cost-Free Estimation via ERK Pairs: Accurately estimating stiffness and the dominant eigenvector without any extra network evaluations.

  • Orthogonal & Plug-and-Play Guidance: Consistently enhancing sample quality and seamlessly integrating with existing models, solvers, and guidance schemes.

MLV Lab