Share this page!

Enter URL or ID to Unroll

You can paste full URL like: https://x.com/threadreaderapp/status/1644127596119195649
or just the ID like: 1644127596119195649

How to get URL link on X (Twitter) App

  1. On the Twitter thread, click on or icon on the bottom
  2. Click again on or Share Via icon
  3. Click on Copy Link to Tweet
  4. Paste it above and click "Unroll Thread"!
  5. More info at Twitter Help
Chaitanya K. Joshi Profile picture
Chaitanya K. Joshi
@chaitjo
PhD student @Cambridge_CL. Research Intern @AIatMeta and @PrescientDesign. Interested in Deep Learning for biomolecule design.
Mar 10 • 11 tweets • 4 min read
Introducing All-atom Diffusion Transformers

— towards Foundation Models for generative chemistry, from my internship with the FAIR Chemistry team @OpenCatalyst @AIatMeta

There are a couple ML ideas which I think are new and exciting in here 👇 Image 📄 Link to paper PDF: arxiv.org/abs/2503.03965

💻 Open source code on GitHub: github.com/facebookresear…
Dec 13, 2023 • 4 tweets • 2 min read
We just released a "Hitchhiker's Guide" for getting started with deep neural nets for 3D structural biology & chemistry -- we hope it will be useful for newcomers! 🤗

Learn about the core architectures powering protein design, material discovery, molecular simulations, and more! And I learnt so much from my passionate and kind co-authors on this long journey of writing something together without targeting any venues or aiming to publish 💙

We focussed purely on explaining everything simply and clearly -- let us know how to improve it!
Jan 24, 2023 • 7 tweets • 6 min read
🚀Excited+nervous to share our latest work on understanding geometric GNNs for biomolecules, materials, etc.

"On the Expressive Power of Geometric GNNs" with @crisbodnar @SimMat20 @TacoCohen @pl219_Cambridge

PDF: arxiv.org/abs/2301.09308
Code: github.com/chaitjo/geomet…
Findings👇 Axes of geometric GNN expressivity: (1) Scalarisation body o How powerful are geometric GNNs? How do design choices influence expressivity?

💡Key idea: Geometric graph isomorphism + Geometric WL framework --> upper bound on geom. GNN expressivity.

Standard GNN tools (WL) are inapplicable due to new physical symmetries (roto-translation).
Mar 15, 2022 • 9 tweets • 3 min read
🚨 New blogpost alert v2:

"Recent Advances in Efficient and Scalable Graph Neural Networks"

Read on for an overview of the toolbox enabling Graph Neural Networks to scale to real-world graphs and real-time applications! 👇

chaitjo.com/post/efficient… Training and deploying GNNs to handle real-world graph data poses several theoretical and engineering challenges:
1. Giant Graphs – Memory Limitations
2. Sparse Computations – Hardware Limitations
3. Graph Subsampling – Reliability Limitations Image
Oct 3, 2021 • 12 tweets • 3 min read
Are you applying for a PhD in Machine Learning, Artificial Intelligence, and beyond?

Here's a thread of high-quality resources that helped me understand the process + craft my application better. 👇 Tim Dettmers' guide to ML PhD applications:

timdettmers.com/2018/11/26/phd…
Sep 14, 2020 • 10 tweets • 4 min read
Sparse Transformers / Graph Neural Networks as a general architecture across a variety of domains: a thread on success stories 👇 Firstly (and this is the most well known), large-scale Transformers have seemingly replaced RNNs in commercial NLP pipelines as they scale better: jalammar.github.io/illustrated-be…
Feb 28, 2020 • 6 tweets • 3 min read
Excited to share a blog post on the connection between #Transformers for NLP and #GraphNeuralNetworks (GNNs or GCNs).

graphdeeplearning.github.io/post/transform… Image The key idea: Sentences are fully-connected graphs of words, and Transformers are very similar to Graph Attention Networks (GATs) which use multi-head attention to aggregate features from their neighborhood nodes (i.e., words).
cc. @PetarV_93 Image