Allows to design deep-learnings models for various problems, training them in the browser and sharing them with others.
This project is still work in progress. That means some important features are still missing:
- Deployment to Cloud (AWS or GCE)
- OpenAuth integration
- Sync models (inclusive weights) and share
- Generic architecture in order to support multiple scenarios
Finished features:
- Mnist scenario
- Inspecting weights and activations
- Editing graphs
A large database of handwritten digits that is commonly used for training various image processing systems.
Demo go to https://lochbrunner.github.io/demos/social-ai and navigate to the mnist example. Press the Load button, after you can step through the samples press Train. Move the nodes from the upper left corner and view the activations after pressing the Predict button.
An abstract strategy board game for two players, in which the aim is to surround more territory than the opponent. Coming soon!
The aim of this scenario is to create a little go bot with TensorFlow.js and live debugging possibility.
To install the dependencies run
yarn download-training-dataThis script downloads and extracts the latest sgf files from badukmovies.com
Work in progress!
A simulation of an ant colony. Model the brain of each individual ant like in serious games.
Coming soon!
Finding the way to a symbolic deduction from the problem to the given target solution.
Coming soon!
Make sure you have nodejs and optional yarn installed.
Run
yarn installRun
yarn build