Of course, PySyft has the ability to run in its own environment. But if you would like to train FL models in the browser, you must resort to using some ML framework like TensorFlow.js.
Syft.js is a microlibrary built on top of TensorFlow.js, allowing for training and prediction of PySyft models in the browser.
PySyft acts as the parent node, instructing child nodes (Syft.js clients running in a website on users' browsers) of what tensors to add to a list, remove from a list, and operate against.
Link to full documentation here
If you're using a package manage like NPM:
npm install --save syft.js @tensorflow/tfjs-core
Or if Yarn is your cup of tea:
yarn add syft.js @tensorflow/tfjs-core
When using a package manager, TensorFlow.js will be automatically installed. If you're not using a package manager, you can also include Syft.js within a <script>
tag (see example below).
Note: If you're training or predicting with another syft.js client running somewhere else (or in another browser tab) then it's highly suggested you include the WebRTC adapter shim inside of your web application.
<script src="https://cdn.jsdelivr.net/npm/@tensorflow/[email protected]/dist/tf.min.js"></script>
<script src="https://webrtc.github.io/adapter/adapter-latest.js"></script>
<!-- If using WebRTC -->
<script src="https://unpkg.com/syft.js@latest"></script>
For integration into your client-side application, please check out our guide.
For further API documentation, please check that out here.
- Fork and clone
- Run
npm install
- Run
npm start
- Do your work.
- Push to your fork
- Submit a PR to openmined/syft.js
It's important to note that each of the examples are self-sustaining projects. This means that they run independent of the npm start
command run at the root syft.js directory. In order to run the demos, please do the following:
- Make sure that grid.js is running and seeded with data.
- Open a terminal to the root syft.js directory.
- Run
npm install
and thennpm start
. - Open a new terminal tab and
cd
into one of the examples. - Run
npm install
and thennpm start
inside of that example's directory. This should open a new browser tab for that example. - Leave both tasks going as you proceed with normal development. The example will automatically refresh in your browser every time you make a change to syft.js or to the example itself.
We're accepting PR's for testing at the moment to improve our overall code coverage. In terms of core functionality, we're considering the current version of Syft.js feature complete until a further roadmap is designated.