Skip to main content
Pre-release

This release is a pre-release and may not be stable for production use.

logo

PyGlove: Manipulating Python Programs

PyPI version codecov pytest

Getting started | Installation | Examples | Reference docs

What is PyGlove

PyGlove is a general-purpose library for Python object manipulation. It introduces symbolic object-oriented programming to Python, allowing direct manipulation of objects that makes meta-programs much easier to write. It has been used to handle complex machine learning scenarios, such as AutoML, as well as facilitating daily programming tasks with extra flexibility.

PyGlove is lightweight and has very few dependencies beyond the Python interpreter. It provides:

  • A mutable symbolic object model for Python;
  • A rich set of operations for Python object manipulation;
  • A solution for automatic search of better Python programs, including:
    • An easy-to-use API for dropping search into an arbitrary pre-existing Python program;
    • A set of powerful search primitives for defining the search space;
    • A library of search algorithms ready to use, and a framework for developing new search algorithms;
    • An API to interface with any distributed infrastructure (e.g. Open Source Vizier) for such search.

It's commonly used in:

  • Automated machine learning (AutoML);
  • Evolutionary computing;
  • Machine learning for large teams (evolving and sharing ML code, reusing ML techniques, etc.);
  • Daily programming tasks in Python (advanced binding capabilities, mutability, etc.).

PyGlove has been published at NeurIPS 2020. It is widely used within Alphabet, including Google Research, Google Cloud, Youtube and Waymo.

PyGlove is developed by Daiyi Peng and colleagues at Google Brain.

Hello PyGlove

import pyglove as pg

@pg.symbolize
class Hello:
  def __init__(self, subject):
    self._greeting = f'Hello, {subject}!'

  def greet(self):
    print(self._greeting)


hello = Hello('World')
hello.greet()

Hello, World!

hello.rebind(subject='PyGlove')
hello.greet()

Hello, PyGlove!

hello.rebind(subject=pg.oneof(['World', 'PyGlove']))
for h in pg.iter(hello):
  h.greet()

Hello, World!
Hello, PyGlove!

Install

pip install pyglove

Or install nightly build with:

pip install pyglove --pre

Examples

Citing PyGlove

@inproceedings{peng2020pyglove,
  title={PyGlove: Symbolic programming for automated machine learning},
  author={Peng, Daiyi and Dong, Xuanyi and Real, Esteban and Tan, Mingxing and Lu, Yifeng and Bender, Gabriel and Liu, Hanxiao and Kraft, Adam and Liang, Chen and Le, Quoc},
  booktitle={Advances in Neural Information Processing Systems (NeurIPS)},
  volume={33},
  pages={96--108},
  year={2020}
}

Disclaimer: this is not an officially supported Google product.

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pyglove-0.5.0.dev202608120834.tar.gz (565.9 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

pyglove-0.5.0.dev202608120834-py3-none-any.whl (736.0 kB view details)

Uploaded Python 3

File details

Details for the file pyglove-0.5.0.dev202608120834.tar.gz.

File metadata

  • Download URL: pyglove-0.5.0.dev202608120834.tar.gz
  • Upload date:
  • Size: 565.9 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/7.0.0 CPython/3.14.6

File hashes

Hashes for pyglove-0.5.0.dev202608120834.tar.gz
Algorithm Hash digest
SHA256 e4c9e80f41897783ac016eaa2aa70108ab2add6231a9fc697d062cc0e5cd61ef
MD5 c6e1901f3e0081d8bd660b26a4ce6a54
BLAKE2b-256 be64a82703e7287060dfb58ddde1eae7fc884743210ef3dbda5e52aeb076dafd

See more details on using hashes here.

File details

Details for the file pyglove-0.5.0.dev202608120834-py3-none-any.whl.

File metadata

File hashes

Hashes for pyglove-0.5.0.dev202608120834-py3-none-any.whl
Algorithm Hash digest
SHA256 15fe1ec35689ac81cdc58118c09bb0a39e1665a62313105b9a658829bc8f6187
MD5 b2953fae817be773643f832126b64b1a
BLAKE2b-256 629ba4f43ff51da3fd4ef1b9d7ade0a9da28e9e2bca7af00de9fb0d5272b8aa0

See more details on using hashes here.

Release history Release notifications | RSS feed

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page