Learning tfjs + umap-js
Just one of the things I'm learning. https://github.com/hchiam/learning
Get UMAP output coordinates with umap-js and visualization powered by chart.js.
Great explanation of SNE vs t-SNE (vs UMAP - which sounds better than t-SNE): https://towardsdatascience.com/visualizing-your-embeddings-4c79332581a9 - basically UMAP is better than t-SNE is better than SNE (which is related to the order they were invented). My other repo that uses tfjs-tsne can be found here: https://github.com/hchiam/learning-tfjs-tsne
Learn more about TensorFlow at https://github.com/hchiam/learning-tensorflow
Starting by testing out this repo
Run yarn global add parcel or npm install -g parcel, and then:
Using yarn: (triple-click to select all)
git clone https://github.com/hchiam/learning-tfjs-umap.git && cd learning-tfjs-umap && yarn;
yarn dev;Or with npm: (triple-click to select all)
git clone https://github.com/hchiam/learning-tfjs-umap.git && cd learning-tfjs-umap && npm install;
npm run dev;notes to self
First run yarn dev to create the /dist folder, and then run this:
yarn deploy