1.0.2 • Published 9 months ago

is-true-that v1.0.2

Weekly downloads
-
License
MIT
Repository
github
Last release
9 months ago

Is True That

is-true-that is the most serious library you'll ever need to determine if something is true. Or not. Seriously. Given an object and a natural language condition, it evaluates whether it is true or false.

npm downloads testing npm version Beta In Progress

What is this?

Free, funny and (maybe) useful truth testing tool

This library is designed to bring a bit of humor to your coding experience while providing a straightforward way to assess truthiness in your JavaScript applications.

In short, you can test the veracity of anything you like.

Important: This is a beta release. Accuracy is not guaranteed, and computing speed is relatively slow. Please refer to the Timing section for details.

Installation

npm install is-true-that

Usage

This library is easy to import and use. The basic usage involves the isTrueThat function, where you pass an object and a natural language statement, receiving true if the condition holds, and false otherwise.

const isTrueThat = require('is-true-that');

console.log(isTrueThat({ food: 'pizza', toppings: ['cheese', 'pepperoni'] }, 'is it a pizza with pepperoni')); // true
console.log(isTrueThat({ food: 'pizza', toppings: ['cheese', 'pepperoni'] }, 'is it a pizza with pepperoni but without cheese and tomatoes')); // false
console.log(isTrueThat(7, 'is it a prime number')); // true
console.log(isTrueThat(1000, 'is this the number of soldiers in a Roman legion?')); // false
console.log(isTrueThat({ animal: 'snake', length: 5 }, 'is it 3+2 meters long?')); // true

For more advanced usage, you can configure LLM (Large Language Model) options, timing, verbosity, and capabilities:

import { isTrueThat, initModel } from 'is-true-that';

async function main() {
    const options = {
        verbose: true,
        modelName: 'your-model-name-here'
    };

    // Initialize a model
    options.model = await initModel(options);

    // These requests share the same chat, speeding up computation
    console.log(await isTrueThat({ food: 'pizza', toppings: ['cheese', 'pepperoni'] }, 'is it a pizza with pepperoni', options)); // true
    console.log(await isTrueThat({ food: 'pizza', toppings: ['cheese', 'pepperoni'] }, 'is it a pizza with pepperoni but without cheese and tomatoes', options)); // false
    console.log(await isTrueThat(7, 'is it a prime number', options)); // true
    console.log(await isTrueThat(1000, 'is this the number of soldiers in a Roman legion?', options)); // false
    console.log(await isTrueThat({ animal: 'snake', length: 5 }, 'is it 3+2 meters long?', options)); // true
}

main();

Caution: Each isTrueThat function call initializes a new model or chat session if they are not provided in the options.

Why?

Sometimes you just need to know if something is true. And sometimes, you need a good laugh.

On a more serious note, in certain situations, such as dynamic user inputs, you may require a quick truth assessment based on natural language queries without having to implement complex logic. LLMs can be useful for evaluating such truths!

Features

  • 82% Accuracy: Generally tells you if something is true or not.
  • Lightweight: No unnecessary bloat (basic model requires 2GB of disk space and up to 4GB of RAM).
  • Humorous: Because why not?

Timing

We ran 150 test cases using npm test and found that the average execution time is 16,834ms with a standard deviation of 5,230ms. Longer execution times occur when the context window exceeds 1,000 tokens, which can take up to 16 seconds to recover (typically every 5-10 requests). This can be mitigated by clearing the chat or increasing the number of context tokens.

Sample Test Results

Test CaseDescriptionExpected ResultActual ResultPass/Fail
1Pizza with pepperonitruetruePass
2Pizza with pepperoni but without cheese and tomatoesfalsefalsePass
3Prime number check for 7truetruePass
4Roman legion soldiers countfalsefalsePass
5Snake length checktruetruePass

These results demonstrate the accuracy and reliability of the is-true-that library in various scenarios.

Models

Is True That uses orca-mini-3b-gguf2-q4_0.gguf as base model which is a performant but non-commercial model offered with gpt4all. It weights 2GB and will be installed once and then cached. It requires 4GB of RAM to be executed locally.

For more consistent results, we suggests Llama 3 8B Instruct or heavier models. More information at gpt4all-models.

Other

Contributing

  1. Fork the repository!
  2. Create a feature branch: git checkout -b my-new-feature
  3. Commit your changes: git commit -am 'Add some feature'
  4. Push to the branch: git push origin my-new-feature
  5. Submit a pull request!

License

MIT


Made with ❤️ and a sense of humor.