Title

Causal Integration in Foundation Models for Mechanistic Understanding

Challenge

Despite their enormous popularity, the mechanisms that LLMs use to perform tasks and the related failure modes are poorly understood.

Description

Research methods for incorporating causal knowledge or discovering learnt causal structure within large language models (LLMs) and diffusion models, aiming to improve our understanding of these models, generate mechanistic hypotheses and seamless integration of observational and interventional data during training and inference.

Skills Required

Causality, LLMs, machine learning theory

Skills to be Developed

Causal inference, representation learning

Relevant Background Reading

1. https://openreview.net/pdf?id=Idusfje4-Wq,
2. https://arxiv.org/abs/2412.12095
3. https://arxiv.org/abs/2310.20307