Constellation client API

The Constellation client API allows developers to interact with the inference engine using the models configured for each project. Inference is the process of running data inputs on a machine-learning model and generating an output, or otherwise known as a prediction.

Before you use the Constellation client API, you need to:

• Sign up for a Cloudflare account Open external link .

• Enable Constellation by logging into the Cloudflare dashboard > Workers & Pages > Constellation.

• Create a Constellation project and configure the binding.

• Import the @cloudflare/constellation library in your code:

     import { Tensor, run } from "@cloudflare/constellation";
  

​​ Tensor class

Tensors are essentially multidimensional numerical arrays used to represent any kind of data, like a piece of text, an image, or a time series. TensorFlow popularized the use of Tensors Open external link in machine learning (hence the name). Other frameworks and runtimes have since followed the same concept.

Constellation also uses Tensors for model input.

Tensors have a data type, a shape, the data, and a name.

export type TensorTypes = "float32" | "float64" | "int32" | "int64" | "string" | "bool";

export type Tensor = {
    type: TensorTypes;
    shape: Array<number>;
    value: Array<any> | Array<Array<any>>;
    name?: string;
};

type

Defines the type of data represented in the Tensor. Options are: float32, float64, int32, int64, string or bool.

shape

Tensors store multidimensional data. The shape of the data can be a scalar, a vector, a 2D matrix or multiple-axes matrixes. Some examples:

• [] - scalar data
• [3] - vector with 3 elements
• [3, 2] - two-axes 3x2 matrix
• [3, 2, 2] - three-axis 2x2 matrix

Refer to the TensorFlow documentation Open external link for more information about shapes.

value

This is tensor’s data. Example tensor values could include:

• scalar: 4
• vector: [1, 2, 3]
• two-axes 3x2 matrix: [[1,2], [2,4], [5,6]]
• three-axes 3x2 matrix [ [[1, 2], [3, 4]], [[5, 6], [7, 8]], [[9, 10], [11, 12]] ]

name

Naming a tensor is optional, it can be a useful key for mapping operations when building the tensor inputs.

​​ Tensor examples

​​ A single string of text

   new Tensor(
       "string",
       [1, 1],
       ["_You Could Qualify For A Exclusive Offer."]
   );

​​ A two-element vector

   new Tensor("int32", [2], [1, 2]);

​​ A 3x2 float64 matrix

   new Tensor("float64", [3, 2], [ [0.1, 0.2], [0.3, 0.4], [0.5, 0.6] ]);

​​ Methods

​​ await project.run(project, model, input)

Runs the inference engine on a model and input.

import { Tensor, run } from '@cloudflare/constellation';

const input = [
    new Tensor("int32", [2], [1, 2]),
    new Tensor("float64", [3, 2], [0.1, 0.2, 0.3, 0.4, 0.5, 0.6]),
    new Tensor("string", [1, 2], []),
]

const output = await run(env.PROJECT, modelId, input)

• input is a list of tensors.
• modelId is the model ID inside the project. Use Wrangler to list the model ID associated with your project.
• PROJECT is your Constellation project ID.