Projects

Predicting AI bias using SAEs

Published:

A comparative analysis of how Sparse Autoencoders and MLP activations encode gender information inside LLMs.

Beyond Lang

Published:

Voice calling with real-time speech translation and transcription.