🔍 I’m currently reverse engineering thicc neural nets at Aleph-Alpha.
🤖 I like to understand how neural networks do what they do. Given below are 2 of my recent projects which I loved working on:
- AtMan: Understanding Transformer Predictions Through Memory Efficient Attention Manipulation
- DORA: Exploring outlier representations in Deep Neural Networks
🎨 I spent some time in the past trying to reverse engineer vision models which eventually led to torch-dreams.
🛠️ I’ve also been an intern at RunwayML where we worked on building next-gen video tools with deep-learning.
Get in touch