Extracting Contrastive Temporal Knowledge using Large Language Models

This repository contains a fine-tuned model developed for my bachelor thesis on "Extracting Contrastive Temporal Knowledge using Large Language Models" using a method called "Contrastive Decoding".

Model Information:

• Base Model: LLama2-8B was fine-tuned
• Dataset: The dataset contained articles from the "New York Times" spanning from the year 2000 until 2007. Very short articles were excluded from the dataset to ensure quality and relevance.

Additional Resources

You can find all the scripts and additional resources used for my thesis in my GitHub repository: LLM-Temporal-Difference-Learning

Acknowledgments

I would like to thank the DataScience team at the University of Innsbruck for their support and guidance during the development of this model.