Hi there! I'm Aman, a PhD student in Computation and Neural Systems at Caltech in Matt Thomson's lab. I'm interested in understanding (math) and building (engineering) intelligence.
How can simple components come together to create an intelligent whole?
I am pursuing these questions via deep learning (LLMs, differentiable programs) and information theory. In the past I worked on theoretical neuroscience, computational biology, chemical computation, and signal processing (see Publications).
For collaborations, reviews, or just to chat, email me at abhargav [at] caltech [dot] edu.
We just dropped our main show on "What's the Magic Word? A Control Theory of LLM Prompting". This is some of the bonus footage we didn't release on YT. With @ABhargava2000 @witkowski_cam @Sz_Looi pic.twitter.com/3zGp5ddKFL
— Machine Learning Street Talk (@MLStreetTalk) May 27, 2024
A few recent videos from my research
Posts
On Abstraction and Long-Time Horizon Prediction
Neural Cellular Automata, Active Inference, and the Mystery of Biological Computation
Toward a Control Theory of LLMs
How to Play With LLMs
subscribe via RSS