Using Large Language Models for Qualitative Analysis can Introduce Serious Bias

no code implementations • 29 Sep 2023 • Julian Ashwin, Aditya Chhabra, Vijayendra Rao

Therefore, given that some high quality annotations are necessary in order to asses whether an LLM introduces bias, we argue that it is probably preferable to train a bespoke model on these annotations than it is to use an LLM for annotation.

Bayesian Topic Regression for Causal Inference

1 code implementation • EMNLP 2021 • Maximilian Ahrens, Julian Ashwin, Jan-Peter Calliess, Vu Nguyen

To this end, we combine a supervised Bayesian topic model with a Bayesian regression framework and perform supervised representation learning for the text features jointly with the regression parameter training, respecting the Frisch-Waugh-Lovell theorem.

Causal Inference regression +1