### 🧹 Clean up The repository has been cleaned up with the help of @MArpogaus. We are now using `ruff` and `pre-commit` to ensure code quality and consistency (#35, #38). We additionally dropped the legacy `setup.py` for the better `pyproject.toml` (25183d9937c81d87d2f24f2d76e6bccd82413ac0). ### ♻️ Refactor The `zuko.flows.core` components have been moved to `zuko.lazy` and `zuko.flows.mixture` components have been moved to `zuko.mixtures` (86816c4131f218464b739103d2b54ae7890453fa). This change should facilitate the addition of parameterized distributions which are not normalizing flows. ### ✨ What's new * New bounded Bernstein transformation (#37) by @MArpogaus and @oduerr * Support for adjacency matrices in autoregressive models (#54) by @adrianjav * Support for non-full covariance matrices in GMM (#50) by @dominik-strutz * Refreshed tutorials (#48) by @psteinb **Full Changelog**: https://github.com/probabilists/zuko/compare/1.1.0...1.4.0

### ✨ What's new * New [VAE tutorial](https://zuko.readthedocs.io/en/stable/tutorials/vae.html) using the MNIST dataset (8812e04507bf27d4fb9346acd9174459c097fc62) * Add support for unconditional univariate flows (#34) * New Bernstein polynomial flow (#32 and #33) by @oduerr and @MArpogaus **Full Changelog**: https://github.com/probabilists/zuko/compare/1.0.0...1.1.0

### 💥 Breaking news Zuko's repository has been transferred to the [probabilists](https://github.com/probabilists) organization. ### ♻️ Refactor To facilitate the maintenance of existing flows and the addition of new flows, the `zuko.flows` module has been refactored into many submodules (#24). As part of this refactor, the `DistributionModule` and `TransformModule` have been renamed to `LazyDistribution` and `LazyTransform` to clarify the purpose of these classes and avoid confusion with [Pyro](https://github.com/pyro-ppl/pyro) components. ### ✨ What's new * New general coupling transformation (#23) by @simonschnake * New gaussianization flow (f7e4f85d66cf648b9f9f72af85891556d46879f5) * New tolerance parameters for `odeint` (bc4323fb217d98794988e8a06e383aa801be1b50) * Use independent networks in neural transformations (c600dadb0cfbb275269193073527dfb83da6c013) * New [tutorials](https://zuko.readthedocs.io/en/stable/tutorials.html) in the documentation (#27) **Full Changelog**: https://github.com/probabilists/zuko/compare/0.2.0...1.0.0

### 💥 Breaking news We decided to drop the support for older PyTorch versions (1.11 and below) as it made it much more difficult to implement some features, especially ones relying on automatic differentiation. PyTorch 1.12 has been around for about a year and is compatible with all CUDA drivers since 10.2. ### ✨ What's new * Add `rsample_and_log_prob` method to generate samples with their log-probability (#18, #19) * Add LU-factorized linear transformation inspired by the `nflows`'s `LULinear` (9c742332012377ce3fd8892cd95268198b1c89d2) ### 🐛 Bug fixes * Fix `_call` not implemented error `call_and_ladj` (#14) **Full Changelog**: https://github.com/francois-rozet/zuko/compare/0.1.4...0.2.0

First official release of Zuko :partying_face: ### ✨ What's new * Refactor `NormalizingFlow` class to improve efficiency `log_prob` efficiency (a755a6994b625e5bc1fb5bcad4d85bb69f5332d6, 3e9c68d54d78f825319fa9712fd7f6a33be407ac) * New continuous normalizing flow (CNF) (a755a6994b625e5bc1fb5bcad4d85bb69f5332d6) * Documentation hosted on Read the Docs (882b8213c5bc93cd63084e7dd382c8ae7250a8f4) * New neural circular spline flow (NCSF) (#12) * New Gaussian mixture model (GMM) (995efc220fa43611265ba35482e8e90c5efd68b0) **Full Changelog**: https://github.com/francois-rozet/zuko/compare/0.0.6...0.1.4