Papers/2610.20636
🧪 Test?View on arXiv

FlowBudget: Training-Free Sampling-Budget Estimation for Flow Matching Models

Yasaman D. T., Author 2, Author 3, Author 4, Author 5

samplingflow modelsbudget estimationnumerical methods
2610.20636
Builder Relevance
80%
1h ago

Abstract

FLOWBUDGET provides a training-free method to estimate the sampling budget for flow matching models without the need for extensive empirical evaluations.

Reality Card

Core Claim

FLOWBUDGET can estimate the number of function evaluations (NFEs) required for sampling from flow matching models without generating samples at multiple budgets.

Method / Result

FLOWBUDGET normalizes the leading one-step discretization error and maps numerical tolerance to a step budget using the solver order.

Limitations

The method relies on the accuracy of interpolation probes and may not generalize across all flow model architectures.

Paper to code

Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.

No verified implementation link has been attached yet. AIBuzzHub will keep this panel separate from unverified search results.
← Back to all papers