Skip to main content

> ML_ALGORITHM // SCORE-BASED-GENERATIVE-MODELING-SDE_v1.0

Score-Based Generative Modeling through Stochastic Differential Equations

Unifying mathematical foundation for diffusion models that models generative perturbation and denoising as continuous forward and reverse-time SDEs.

Diffusion & Score-Based Generative Modelsdeep-generativeblack-boxlarge (>100k)
Back to All Algorithms
Computational Complexity
Training Complexity:O(epochs * batch_size * score_net_pass)
Inference Complexity:O(predictor_corrector_steps * score_net_pass)
Hardware Profile
CPU Friendly:No
Requires GPU:Yes
Memory Footprint:high
Interpretability & Data
Interpretability Tier:black-box
Training Data Needs:large (>100k)

Interpretability Assessment

Predictor-corrector sampling combines numerical SDE integration with Langevin MCMC error correction.

Suitable Tasks & Supported Modalities

Suitable Tasks:
image generationinverse problems mri
Supported Modalities:
imagemedical-imaging

Implementing Libraries

PyTorchLinux Foundation / PyTorch Foundation · v2.4.1
View Spec
diffusers

Foundational Literature

Score-Based Generative Modeling through Stochastic Differential EquationsYang Song, Jascha Sohl-Dickstein (2020) · International Conference on Learning Representations (ICLR)
Common Pitfalls & Warnings
  • Numerical SDE solver error accumulation when using naive Euler-Maruyama solvers with insufficient discretization steps