I employed GPT-4 to construct a synthetic dataset that enabled the development of a robust classifier by training BERT for identifying surprise in customer interactions.
I built a classifier by training FinBERT using a synthetic dataset developed using GPT-4 to measure attribution patterns in IPO prospectuses.