TensorCI API Client

The TensorCI API Client is the easiest way to fetch predictions from your trained TensorCI models.

Installation

Assuming you have Python already, this library can be installed with pip:

$ pip install tensorci_client

Fetch Model Predictions

Once you’ve deployed a trained model to your team’s TensorCI API cluster, fetching that model’s predictions involves the following 3 steps:

1.Obtaining API Credentials

Each TensorCI project comes with its own Client ID and Client Secret to secure its hosted model predictions. These credentials can be found on the TensorCI Dashboard at https://app.tensorci.com/<TEAM_SLUG>/<PROJECT_SLUG>/settings.

2. Setting API Credentials as Environment Variables

Once you’ve obtained your API credentials, set the following environment variables in the Python project you wish to fetch model predictions from:

  • TENSORCI_CLIENT_ID
  • TENSORCI_CLIENT_SECRET
  • TENSORCI_TEAM
  • TENSORCI_PROJECT

3. Initializing the TensorCI Client

Once you’ve set the above environment variables, model predictions can be fetched in the following manner:

from tensorci import TensorCI

client = TensorCI()

unseen_features = {
  'key1': 'val1',
  'key2': 'val2'
}

prediction = client.predict(data=unseen_features)

print('Got Prediction: {}'.format(prediction))