Hello world

gpt4all-helloworld-nodejs

Setup guide by cloning this repository

cd into this directory
npm installcCX
node index.js

Setup this from scratch - Ignore this if you cloned this repository

1. Still need to download a model
2. run the command below

cd C:/Users/<username>/Desktop
mkdir gpt4all-helloworld-nodejs
npm init

install the gpt4all nodejs binding

https://docs.gpt4all.io/gpt4all_typescript.html

npm install gpt4all@alpha

Download a model

Not really needed.. gpt4all@alpha will be downloading a valid model from hugging face automatically if one doesn't exists. The only problem is that the command prompt doesn't tell you how much it has already downloaded... which creates huge anxiety...

The best way for me to download a model seems to be installing in chat client from https://gpt4all.io/index.html. The chat client lists all chat models and their respective features.

After downloading a model in whatever ways, copy the model to this project directory.

Create index.js

Just create a file under the project directory.

Add these lines,

const { createCompletion, loadModel, createEmbedding } = require ('gpt4all')

async function testing(){
  const model = await loadModel('ggml-model-gpt4all-falcon-q4_0', { 
    verbose: true,
    modelPath: "C:/Users/jason/Desktop/gpt4all-nodejs_yannick/gpt4all-nodejs"
  });

  console.log("Model loaded successfully")
  console.log(model)
  const answers = await createCompletion(model, [
      { role : 'system', content: 'You are meant to be annoying and unhelpful.'  },
      { role : 'user', content: 'What is 1 + 1?'  },
  ]);

  answers.choices.forEach((x) => { 
    console.log(x)
  })
}

testing()