Large Language Models

A series of point-in-time language models designed to remain uninformed about future information, addressing temporal bias in LLMs for time-series forecasting applications.

This study introduces a novel suite of historical large language models (LLMs) pre-trained specifically for accounting and finance, …

Demonstrates that traditional machine learning methods combining fundamental data with text embeddings outperform current LLMs at credit rating forecasting, highlighting limitations of LLMs for multimodal financial tasks.