pynavio 0.3.1

Python lib for navio

pynavio

Python lib for navio

• Free software: MIT license

Features

• Pynavio.Client is a navio API client that enables users to upload models and data, deploy and retrain models etc.

• Pynavio.mlflow.to_navio function calls mlflow.pyfunc.save_model function, saving a model zip file as required by navio.

  • it enables inferring the conda environment (with pip requirements) and adding extra pip dependencies to the inferred environment

  • it enables adding sys dependencies to the navio model

  • it also validates the models with Pynavio.mlflow.ModelValidator by default

• Pynavio.mlflow.ModelValidator is a class that validates the model (prediction/example request/MLmodel metadata schema checks, warnings related nested types/big model sizes)

• Pynavio.infer_external_dependencies is a helper function that infers the external dependencies based on the file path. Please refer to its doc string for limitations.

• Pynavio.infer_imported_code_path is a helper function that infers the imported code paths based on the file path and the root path. Please refer to its doc string for limitations.

• Pynavio.make_example_request generates a request schema for a navio model from data.

Documentation

The official documentation is hosted on https://navio.craftworks.io : https://navio.craftworks.io/docs/guides/pynavio/

Examples

To build all example models, use:

$ cd scripts && make

Credits

This package was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.

Examples/models uses code parts from Open Source project mjain72/Condition-monitoring-of-hydraulic-systems-using-xgboost-modeling.

Examples/models uses code parts from Open Source project https://www.kaggle.com/maciejautuch/car-price-prediction

History

0.0.1 (2021-09-07)

• First release on PyPI.

0.0.2 (2021-12-29)

• allows user provide modules as pip strings

• brings possibility to infer package dependencies(pip strings) using the infer_external_dependencies()

• handle data frame input for example request

• updates the default conda channels: remove anaconda and pytorch and add conda-forge

• adds and updates the dependencies

• brings gpu related changes

• brings example models (tabular etc.)

• fixes the issue with pigar running for 10 minutes when the input path contains venv

0.1.0 (2022-01-24)

• adds model helpers

• brings more example models: pump leakage model

• removes the need to download the data from kaggle, refers to local data dir for data instead

• adds credits to README, adds LICENSE

0.1.1 (2022-02-07)

• adds credits for kaggle notebook’s code parts usage

• brings more example models: car price model

0.1.4 (2022-08-25)

• API Client

• adds possibility to infer the dependencies used in the file only

• extra docs for example models

• moves the data dir out into examples

• updates dependencies

• fixes the missing import of the currently available models

• fixes mlflow==1.27.0 circular import error

• fixes pynavio FileNotFound error mlflow-1.26.0

• fixes _fetch_data example request path according to mlflow-1.26.0

• sets protobuf version to fix failing tests

• fixes to_navio() to support kedro path

0.1.5 (2022-09-15)

• fixes the documentation link

• attempts to remove the build badge

• removes installation for development section form readme

• fixes repository on the pypi page

• adds history

0.1.6 (2022-09-15)

• final touches/refactoring to make the repo ready for becoming public

0.1.7 (2022-11-16)

• fixes mlflow version to be <2.0 as it introduces breaking changes in the api

• wraps navio’s model deletion api

0.2.0 (2022-11-30)

• resolves mlflow2 incompatibility

• brings model validation checks:prediction, example request and MLmodel metadata schema validation

• makes check_model_serving function public

• adds jsonschema to requirements

• bump tensorflow from 2.9.1 to 2.9.3

• bump pillow from 9.2.0 to 9.3.0

• refactors example tests

0.2.1 (2022-12-07)

• loosens version requirement for jsonschema

• improves and fixes typos in error message texts

• prevents the exception traceback from prediction_call when checking if it is used, to not confuse the user

• brings a descriptive error if model path in code_path

• brings a descriptive error if code_path is not a list

0.2.2 (2023-04-03)

• bump tensorflow from 2.9.3 to 2.11.1

• updates torch version and adds onnx as requirement

• fixes torch, pyarrow and onnxrutime versions

• fixes the prediction schema to only expect a list as prediction

• bump ipython from 8.4.0 to 8.10.0

• improves the docstring of to_navio function

• bump wheel from 0.37.1 to 0.38.1

0.2.3 (2023-08-08)

• adds warnings related to limitations for nested inputs and big model.zip size

• make ModelValidator a public class

• adds docstrings to ModelValidator

• makes model validation optional in to_navio(), so one can disable it

• adds (pynavio model validation) to the validation messages so it is clear where they come from

• adds messages for pynavio model validation checks failed(how to disable)/succeeded(how to check model serving)

• updates readme with model validation info

0.2.4 (2023-08-22)

• add possibility for user to add sys dependencies to the navio model

0.3.0 (2024-01-17)

• allows for a conda environment with pip requirements to be inferred

• adds possibility to add extra pip dependencies to the inferred environment

0.3.1 (2024-01-23)

• minor reformatting fix