Alexa Tartaglini | about

Stanford CS PhD student. Researching interpretability and multimodality.

Alexa Tartaglini | about Towards a cognitive science for neural networks I’m a second year Computer Science PhD student @ Stanford, advised by Christopher Potts and Judith Fan . My primary research interests are deep neural network interpretability and multimodal reasoning , drawing inspiration from cognitive science. The questions I’m the most excited about lie at the intersection of vision, language, and thought. How do neural networks see the world, and why do they continue to struggle on some of the most fundamental tasks studied by cognitive scientists? // bio // Prior to starting my PhD,

saved by

• Elizabeth Qiu
• Hamidah Oderinwale

related reading

• Distill — Latest articles about machine learningdistill.pub
• A Pragmatic Vision for Interpretability — AI Alignment Forumalignmentforum.org
• Transformer Circuits Threadtransformer-circuits.pub
• The Building Blocks of Interpretabilitydistill.pub
• An Ambitious Vision for Interpretability — AI Alignment Forumalignmentforum.org
• Multimodal Neurons in Artificial Neural Networksdistill.pub
• Home - colah's blogcolah.github.io
• On Optimism for Interpretabilitygoodfire.ai
• On neural scaling and the quanta hypothesisericjmichaud.com
• A short note on interpretability and mindsericjmichaud.com
• Interpretability — LessWronglesswrong.com
• What is the purpose of interpretability?ericjmichaud.com