aleximmer/Laplace
A Python library for applying Laplace approximations to neural networks, enabling Bayesian posterior approximations and uncertainty quantification.
Not currently ranked — collecting fresh signals.
star history
The laplace package provides methods for applying Laplace approximations to entire neural networks, subnetworks, or just the last layer. It facilitates posterior approximations, marginal-likelihood estimation, and posterior predictive computations for deep learning models. The library supports various hessian factorizations, prior precision tuning methods, and predictive approaches, making Bayesian deep learning more accessible through a well-documented Python API.
Frequently asked
- What is aleximmer/Laplace?
- A Python library for applying Laplace approximations to neural networks, enabling Bayesian posterior approximations and uncertainty quantification.
- Is Laplace open source?
- Yes — aleximmer/Laplace is open source, released under the MIT license.
- What language is Laplace written in?
- aleximmer/Laplace is primarily written in Python.
- How popular is Laplace?
- aleximmer/Laplace has 548 stars on GitHub.
- Where can I find Laplace?
- aleximmer/Laplace is on GitHub at https://github.com/aleximmer/Laplace.