HiddenLayer SDK Python (Beta)

Hiddenlayer is a Python SDK that provides a simple and efficient way to interact with the Hiddenlayer API, a platform for building and deploying machine learning models. The SDK allows developers to easily protect models using the Hiddenlayer API, without having to write complex code or manage the underlying infrastructure.

Contents

• Installation
• Getting started
• Code examples
• Interface stability

Installation

Install from PyPi using pip, a package manager for Python.

pip install hiddenlayer-sdk

Scanning models on different platforms such as AWS S3 and Huggingface are supported. You can install the dependenices yourself or via optional dependencies:

pip install hiddenlayer-sdk[aws]
pip install hiddenlayer-sdk[azure]
pip install hiddenlayer-sdk[hf]

Getting Started

Once you've installed the hiddenlayer package, you can instantiate the HiddenlayerServiceClient for the SaaS platform as follows:

from hiddenlayer import HiddenlayerServiceClient

hl_client = HiddenlayerServiceClient(
  host="https://api.us.hiddenlayer.ai",
  api_id=..., # Your Hiddenlayer API Client ID
  api_key=... # Your Hiddenalyer API Secret Key
)

Scanning Models

hl_client.model_scanner.scan_file(
  model_name="name_of_the_model",
  model_path="path/to/model/file.pkl"
)

If you would like to specify the version associated with the model, you can pass that in as well. Note that the model version must not already exist.

hl_client.model_scanner.scan_file(
  model_name="name_of_the_model",
  model_version=1,
  model_path="path/to/model/file.pkl"
)

Scanning Folders

hl_client.model_scanner.scan_folder(
  model_name="name_of_the_model",
  path="path/to/model/"
)

If you would like to specify the version associated with the model, you can pass that in as well. Note that the model version must not already exist.

hl_client.model_scanner.scan_folder(
  model_name="name_of_the_model",
  model_version=1,
  path="path/to/model/"
)

Using AIDR for Predictive Models

Note: This is only supported using the SaaS version of the platform.

model = hl_client.model.create_model(model_name="example_model")

hl_client.aidr_predictive.submit_vectors(
  model_id=model.sensor_id,
  requestor_id="custom-source-id",
  input_vectors=X,
  output=y
)

Code Examples

Code examples can be found in the repo here

Interface Stability

Hiddenlayer is actively working on stabilizing the Hiddenlayer SDK for Python's interfaces. You are highly encouraged to pin the exact dependency version.