Skip to main content

Programmatic parameter sweeps for pydantic.

Project description

pydantic-sweep

pydantic_sweep is a library to programmatically, safely and flexibly define complex parameter sweeps over pydantic models in Python.

PyPI - Version PyPI - Python Version GitHub branch check runs GitHub License

Highlights:

  • Specify parameter sweeps in Python
  • Flexibility: specify complex parameter combinations by chaining simple functional operations
  • Safety checks for parameter combinations (get meaningful errors early)
  • pydantic field validation
  • Conversion between json/yaml/python code representation of models

For example, the following code will instantiate models with (x=5, sub=Sub1(s=1) and (x=6, sub=Sub1(s=2) and try each of those with seed values of seed=43 and seed=44, leading to four different configurations:

import pydantic_sweep as ps

class Sub1(ps.BaseModel):
    s: int = 5

class Sub2(ps.BaseModel):
    y: str = "hi"

class Model(ps.BaseModel):
    seed: int = 42
    x: int = 5
    sub: Sub1 | Sub2

# We want to test across two seeds
configs = ps.config_product(
    ps.field("seed", [43, 44]),
    # And two specific Sub1 and `x` combinations
    ps.config_zip(
        ps.field("x", [ps.DefaultValue, 6]),
        ps.field("sub.s", [1, 2]),
    )
)
# This includes safety checks that Sub1 / Sub2 are uniquely specified
models = ps.initialize(Model, configs)

# The code above is equivalent to
models_manual = [
    Model(seed=43, sub=Sub1(s=1)),
    Model(seed=43, x=6, sub=Sub1(s=2)),
    Model(seed=44, sub=Sub1(s=1)),
    Model(seed=44, x=6, sub=Sub1(s=2)),
]
assert models == models_manual

# We can also check that we didn't accidentally duplicate a setting
ps.check_unique(models)

While in this toy example, manually specifying the combinations may still be viable, the library allows infinitely combining different configs and sub-models, making it a powerful tool for large-scale experiment definition. To learn mode about the capabilities of the library please visit the documentation page.

Installation

You can install the library by checking out the repo and running

pip install 'pydantic-sweep'

License

The main code-base is licensed under MPL-2.0 a weak copy-left license that allows commercial use. See the license file for the exact clauses and this FAQ for a high-level description.

An exception from this are the documentation in the docs and example folders as well as this README file, which are licensed under the 0BSD: a highly permissive license that does not require attribution. That way, you are free to copy & paste example code into your use-case. See here for a high-level description.

Project details


Download files

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

Source Distribution

pydantic_sweep-0.3.8.tar.gz (18.3 kB view details)

Uploaded Source

Built Distribution

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

pydantic_sweep-0.3.8-py3-none-any.whl (22.5 kB view details)

Uploaded Python 3

File details

Details for the file pydantic_sweep-0.3.8.tar.gz.

File metadata

  • Download URL: pydantic_sweep-0.3.8.tar.gz
  • Upload date:
  • Size: 18.3 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for pydantic_sweep-0.3.8.tar.gz
Algorithm Hash digest
SHA256 e1f194cc801692d628a369e93c1a47c5cac9cd316b36e48074324fc432404e6c
MD5 7e9690123c30b4df9e7047c13c89c94d
BLAKE2b-256 d368132fb796d6d3ff05a0785d94db7537fc1b43fa2767f9c776ea9eff0397c3

See more details on using hashes here.

Provenance

The following attestation bundles were made for pydantic_sweep-0.3.8.tar.gz:

Publisher: release.yml on befelix/pydantic_sweep

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file pydantic_sweep-0.3.8-py3-none-any.whl.

File metadata

  • Download URL: pydantic_sweep-0.3.8-py3-none-any.whl
  • Upload date:
  • Size: 22.5 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for pydantic_sweep-0.3.8-py3-none-any.whl
Algorithm Hash digest
SHA256 3df1f72258f50b66ddd83e5c9a603e00f1436dfa726dfe3abaeb25f2b4b17b2a
MD5 825d918143d555b51e10891f33f4488c
BLAKE2b-256 6b7300ddd60ca0e25fbd782b192a5d40d7d439c71e5a89ed314bcc6ab161a708

See more details on using hashes here.

Provenance

The following attestation bundles were made for pydantic_sweep-0.3.8-py3-none-any.whl:

Publisher: release.yml on befelix/pydantic_sweep

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

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