adzcai / iclr-23-viz

Visualization of the ICLR 2023 submissions using allenai/specter embeddings and TSNE.

Home Page:iclr-23-viz.vercel.app

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

iclr-23-viz

Visualization of paper submissions for the ICLR 2023 conference.

The visualization was made by:

  1. Fetching data from
  2. Passing the paper title, keywords, abstract, and category through the SPECTER language model
  3. Using the first token as the sequence embedding
  4. Running TSNE to reduce down to 2 dimensions. (The plots look nicer and less scattered than PCA.)

Powered by create-svelte.

Deployed with Vercel.

Developing

npm install
npm run dev -- --open

Building

To create a production version of your app:

npm run build

You can preview the production build with npm run preview.

About

Visualization of the ICLR 2023 submissions using allenai/specter embeddings and TSNE.

iclr-23-viz.vercel.app

License:MIT License


Languages

Language:Svelte 81.8%Language:JavaScript 7.0%Language:TypeScript 6.7%Language:HTML 2.3%Language:CSS 2.2%