Machine learning pipelines

Thoughts on learning to "fail fast" in research. When you have a new idea, and the project is worthwhile only if P, then you need to check P first, not last. Examples: 1. You can only establish causality with a certain kind of data. 🧵

As I struggled to deal with the impact of COVID on my family members in India, I got delayed by a day for submitting my reviews for a conference & I got a message from a senior reviewer with the blurb below. My humble request to everyone - pls don't say this to anyone ever! [1/n]

An awesome example of how using NLP techniques with evolutionary sequences can let you model genetic variation of proteins (without alignments!) and design novel nanobodies!! I worked on this back in college in with the wonderful Debora Marks lab- cool to see it published now :)

I couldn't resist. With contributions from Max Kleiman-Weiner

another creative tool by the brilliant Joel Simon giving artists a fun new way to 'paint' on images with ML-driven manipulations!!

In Nature Reviews Cancer, one of great success stories in #RareCancers, by one of the principal architects Dr. Sanford Simon Rockefeller University: Fighting rare cancers: lessons from fibrolamellar hepatocellular carcinoma. (Photograph of Dr. Simon & Elana via: cancer.gov/research/areas…)

Excited to share my first paper from my PhD research in Chris Garcia's lab! We determined structures of the IL-21 receptor complex and engineered IL-21 variants that enhanced vaccine responses in tonsil organoids with reduced autoantibody induction. (1/9) cell.com/cell-reports/f…

Announcing 𝐏𝐮𝐩𝐩𝐞𝐭𝐬 𝐀𝐩𝐩 A collaborative puppet theater you control with your hands 🙌 It was an experiment in creative tools based on principles of craft, play and collaboration. This was a Morphogen R&D project from 2021-2022. A thread🧵

I'm excited to share our Nature Methods primer on language models for biological research. Elana Simon, James Zou, and I explain how to use language models, both natural language models and biological language models, to accelerate biological research. nature.com/articles/s4159…

Very excited to share our primer on LLM for biology Nature Methods! We discuss how to use: ⚡️human & protein LMs (eg #ESM) ⚡️#singlecell & #multimodel LMs ⚡️when to fine-tune, transfer learn etc. Paper nature.com/articles/s4159… Colab tutorial colab.research.google.com/drive/1zIIRGeq… Written for a

I recently spoke with Rob Wiblin 80,000 Hours about Anthropic’s Responsible Scaling Policy for safely developing AI systems. I particularly enjoyed discussing the trade-offs of RSPs, and how safety-conscious researchers might want to consider them when working on AI capabilities.

We’re excited to introduce @ChaiDiscovery and release Chai-1, a foundation model for molecular structure prediction that performs at the state-of-the-art across a variety of drug discovery tasks We're releasing inference code, weights & a web interface: chaidiscovery.com/blog/introduct…