PipeRider SDK For Python

The PipeRiper SDK for Python enables you to access and manipulate project resources in your PipeRider account.

Installation

pip install piperider-python-sdk

Configuration

To use the SDK you will need to obtain an API key. This can be done from the User Settings page of your account.

https://app.piperider.io/settings

Configure your API key using one of the following methods.

Store API key in an environment variable

Users can set the API key as an environment variable.

export PIPERIDER_API_KEY=<user_api_key>

Configure API key using the set_api_key function

Set your API key using the set_api_key function.

import PipeRider
PipeRider.set_api_key('api-key')

Usage

Import the PipeRider module.

import PipeRider

Create a project.

project = PipeRider.create_project('my-project')

Use an existing project.

project = PipeRider.project_by_uid('6f9f6eb9-8657-4761-9e32-116f1b8cfb82')

Create a run and save the settings for a machine learning project.

with project.runs.create(name='Awesome Run') as run:
    run.config = {
        'learning_rate': 0.02,
        'architecture': 'CNN',
        'dataset': 'TKNV-users',
    }

    run.params = {
        'batch_size': 64,
        'epoch': 100,
        'learning_rate': 0.005
    }

Add a dataset. This will enable you to more easily trace your model lineage.

run.add_dataset('golden-dataset')

Comments can also be added to the timeline.

project.comment('it is a good idea.')

If the run achieves the desired performance it can be registered as a model by declaring it a win.

run.win()