Adapting Large Language Models through fine tuning techniques for application in social human-robot interaction

The research area of LLMs, while very recent, is evolving rapidly in many different ways. In this work, we explore the application of LLMs in the context of humanoid robotics,trough Abel, a humanoid robot. I created a custom dataset of conversation. I selected LLaMA 3, a large language model developed by Meta, with 8B parameters, as the base model for fine-tuning. I adapted LLaMA 3 to understand and respond to the unique interactions expected of a humanoid robot like Abel. This involved training the model on the custom dataset and optimizing key parameters to achieve the best possible performance. By evaluating the final model against responses generated by external GPT model.