Getting Started

Installation

The latest version of JAI-SDK can be installed from pip as follows:

pip install jai-sdk --user

Nowadays, JAI supports python 3.7+.

Getting your auth key

JAI requires an auth key to organize and secure collections. You can quickly generate your free-forever auth-key by running the command below:

>>> from jai import get_auth_key
>>> get_auth_key(email='email@mail.com', firstName='Jai', lastName='Z')

Attention

Your auth key will be sent to your e-mail, so please make sure to use a valid address and check your spam folder.

How does it work?

With JAI, you can train models in the cloud and run inference on your trained models. Besides, you can achieve all your models through a REST API endpoint.

First, you can set your auth key into an environment variable or use a .env file or .ini file. Please check the section How to configure your auth key for more information.

Bellow an example of the content of the .env file:

JAI_AUTH="xXxxxXXxXXxXXxXXxXXxXXxXXxxx"

In the below example, we’ll show how to train a simple supervised model (regression) using the California housing dataset, run a prediction from this model, and call this prediction directly from the REST API.

>>> import pandas as pd
>>> from jai import Jai
>>> from sklearn.datasets import fetch_california_housing
...
>>> # Load dataset
>>> data, labels = fetch_california_housing(as_frame=True, return_X_y=True)
>>> model_data = pd.concat([data, labels], axis=1)
...
>>> # Instanciating JAI class
>>> j = Jai()
...
>>> # Send data to JAI for feature extraction
>>> j.fit(
...     name='california_supervised',   # JAI collection name
...     data=model_data,    # Data to be processed
...     db_type='Supervised',   # Your training type ('Supervised', 'SelfSupervised' etc)
...     verbose=2,
...     hyperparams={
...         'learning_rate': 3e-4,
...         'pretraining_ratio': 0.8
...     },
...     label={
...         'task': 'regression',
...         'label_name': 'MedHouseVal'
...     },
...     overwrite=True)
...
>>> # Run prediction
>>> j.predict(name='california_supervised', data=data)

In this example, you could train a supervised model with the California housing dataset and run a prediction with some data.

JAI supports many other training models, like self-supervised model training. Besides, it also can train on different data types, like text and images. You can find a complete list of the model types supported by JAI on The Fit Method.

What to do next?

Visit Jai in 5 Minutes to get a more complex and detailed example of how to use JAI correctly.

Read about The Fit Method if you want a complete overview of what models JAI can train and what you can do to get your better model.