score_sde_pytorch and score_sde
These are parallel implementations of the same method in different frameworks—PyTorch and JAX respectively—making them competitors for the same use case rather than complementary tools.
About score_sde_pytorch
yang-song/score_sde_pytorch
PyTorch implementation for Score-Based Generative Modeling through Stochastic Differential Equations (ICLR 2021, Oral)
Implements a unified SDE framework for score-based generative modeling with support for multiple architectures (NCSN++, DDPM++, NCSN, NCSNv2, DDPM) and training/evaluation pipelines for image generation tasks. Enables exact likelihood computation, conditional generation (inpainting, colorization, class-conditional), and latent code manipulation through reversible stochastic processes. Integrates with Hugging Face Diffusers library for easy inference and features modular, extensible design for custom SDEs, predictors, and correctors.
About score_sde
yang-song/score_sde
Official code for Score-Based Generative Modeling through Stochastic Differential Equations (ICLR 2021, Oral)
Unifies score-based generative models through continuous-time SDEs, enabling unified training of NCSN, NCSNv2, DDPM variants, and new NCSN++/DDPM++ architectures. Supports exact likelihood computation, latent code manipulation, and conditional generation (inpainting, colorization, class-conditional) through flexible Predictor-Corrector sampling. Implemented in JAX with modular abstractions for SDEs, predictors, and correctors, allowing straightforward extensions while maintaining compatibility across sampling and evaluation methods.
Scores updated daily from GitHub, PyPI, and npm data. How scores work