A repo to show, keep and tinker with Bayesian optimization in python. This repo is not to be used as something to build upon, but rather a folder of code snippets and resources, e.g. links, articles, videos etc. to show how to use concept, and easily snag for other projects.
- Using Ax (https://ax.dev/), the high level no-frills (no configuration) system
- .. (to come?)
This is not a standalone code, but just a bag of code snippets and other artifacts helping anyone to get started with Bayesian Optimization in python. Feel free to contribute with a pull request!
It is reccomendet to use conda. , and read this: For Ax, read this: For BoTorch, read this: https://github.com/pytorch/botorch/blob/main/README.md For PyTorch, read this: https://pytorch.org/get-started/locally/
For conda environment: Use conda environment file, and then install BoTorch using conda (conda install botorch -c pytorch -c gpytorch), and subsequently ax using pip (pip3 install ax-platform)
Ax:
- Reading: https://ax.dev/docs/api.html, https://ax.dev/tutorials/gpei_hartmann_loop.html
- Watching:
BoTorch:
-
On Ax and BoTorch: https://youtu.be/2c8YX0E8Qhw (Facebook launch), https://youtu.be/X_-npBRGRX4 (GP summer school)
-
On Bayesian Optimization (and Gaussian Processes)
- Full Playlist: https://www.youtube.com/playlist?list=PLqxVbTmGzmPvl2XwJ_k46BaJBbII1d4G9 https://youtu.be/_SC5_2vkgbA
- Gold Nuggets: (Marc Disenroth) https://youtu.be/_SC5_2vkgbA
-
On PyTorch and GPyTorch: [ToDo ..]
- To use just run any python file as a script, it should work. Let me know if it does not
No test have been set up for the examples.
Written by Johan Ludde Wessén jlwessen@kth.se, October 2021
🏛 Website: https://www.kth.se/profile/jlwessen