⚡ A learning engine.
A modular Javascript machine learning and artificial intelligence SDK.
🐥 Current release: Alpha 1.0.0
📖 Read the Quick Start Guide
💻 API Docs
🚜 Roadmap
Ensure the latest version of node is used:
nvm use 9
npm install saij
Basic Engine
example:
import Engine from 'saij/Engine';
import Organ from 'saij/Organ';
import Layer from 'saij/Layer';
import Group from 'saij/Group';
import NeuralNet from 'saij/NeuralNet';
import Request from 'saij/Request';
const organA = new Organ({
layers: [
new Layer({
source: new NeuralNet({
type: 'LSTM', architecture: [1, 3, 1]
})
})
]
});
const organB = new Organ({
layers: [
new Layer({
source: new Request('www.some-url.com/stuff')
})
]
});
const organGroup = new Group([organA, organB]);
const engine = new Engine([organGroup]);
engine.setTarget('my-html-element');
engine.setInputTarget('my-input-html-element');
engine.start();
See Saij in action at https://apps.saij.io.
To run all unit tests use:
npm test
To run a single test use:
npm run jest src/saij/file-to-test
All tests must pass 100% coverage (and all use cases) before a merge to master.
Check coverage after a test using:
open build/cov/lcov-report/index.html
Check your code style using eslint:
npm run lint
To build the project use:
npm run build
Built files will be output to build/pkg
.
Package will be automatically published to NPM once it has passed CI (Travis), semantic versioning and it must be successfully merged into master.
A GitBook guide and tutorial documentation can be found at https://skan.gitbook.com/saij, including examples and support.
API documentation can be found at https://skan-io.github.io/saij/
API docs can also be accessed anytime locally using:
serve ./docs
Or you can rebuild the docs:
npm run docs
server ./docs
PRs accepted.
Commits must use Angular commit message standard. Pull requests try to use the pull request template provided.
See the contribute file!
Small note: If editing the README, please conform to the standard-readme specification.
Much of the core and structure of this project has been heavily inspired by OpenLayers. They're doing great work in GIS and mapping so thanks to all the contributors of that project, go check it out!
BSD-3-Clause © 2018 Skan.io