Skip to main content
Deno 2 is finally here 🎉️
Learn more
Module

x/netsaur/packages/core/mod.ts>NeuralNetwork

Powerful machine learning, accelerated by WebGPU
Go to Latest
interface NeuralNetwork
Re-export
import { type NeuralNetwork } from "https://deno.land/x/netsaur@0.4.0-patch/packages/core/mod.ts";

Base Neural Network Structure. All Neural Networks should implement this.

Properties

backend: Backend

The backend used by the Neural Network.

The configuration of the Neural Network.

Methods

train(
datasets: DataSet[],
epochs?: number,
rate?: number,
): void

The train method is a function that trains a neural network using a set of training data. It takes in an array of DataSet objects, the number of epochs to train for, and the learning rate. The method modifies the weights and biases of the network to minimize the cost function and improve its accuracy on the training data.

 network.train([{
   inputs: tensor2D([
     [0, 0],
     [1, 0],
     [0, 1],
     [1, 1],
   ]),
   outputs: tensor2D([[0], [1], [1], [0]]),
 }]);
predict(data: Tensor<Rank>): Promise<Tensor<Rank>>

The predict method is a function that takes in a Tensor object representing the input to the neural network and returns a Promise that resolves to a Tensor object representing the output of the network. This method is used to make predictions on new data after the network has been trained.

const prediction = await net.predict(tensor1D([0, 0]));
console.log(prediction.data[0]);
save(): Uint8Array

The save method saves the network to a Uint8Array. This method is used to save the network after it has been trained.

const modelData = network.save();
Deno.writeFileSync("model.st", modelData);
saveFile(path: string): void

The saveFile method takes in a string representing the path to a file to the safetensors format and saves the network to that file. This method is used to save the network after it has been trained.

network.saveFile("model.st");