In a new ICLR 2023 paper @gkreiman, @DimaKrotov, @alxndrdavies, Deepak Singh and I extend upon a mapping between the cerebellum and Transformers to create a modified multi-layered perceptron that beats continual learning baselines.

Paper: arxiv.org/abs/2303.11934

Thread: 1/18 These modifications come from modeling inhibitory interneurons in the cerebellar circuit not accounted for by the original biological mapping.

Competitive inhibition may help explain why biological brains are so much better than artificial ones at continual learning.

2/18

Hmm maybe driverless cars still have a ways to go…

Documenting a series of epic failures by a driverless @Cruise car to handle a firetruck in SF tonight. This caused a minor but still dangerous traffic jam a human driver would have easily avoided creating. 🧵 The Cruise car was in front of a firetruck that started its sirens. Even though it was far ahead w/ many cars in between, the Cruise stopped and turned to the side of the road. It then would jolt forwards but not move much while every other car kept driving. This was mistake #1.

Very excited to have been involved in this SARS-CoV-2 research that introduces new assays to detect viral gene immune suppression capabilities and even discovers a potential new gene overlapping with Spike.

Read the pre-print here: biorxiv.org/content/10.110…

Thread 🧵 1/n Key paper contributions:
1. Many SARS-CoV-2 proteins are discovered to have enhanced immune suppression compared to other coronaviruses incl. SARS-CoV-1 and MERS.
2. A potential gene is discovered overlapping the Spike gene that shows immune suppression capabilities.

2/n