Mechanism of Knowledge Internalization via Fine-Tuning
A software company fine-tunes a general-purpose language model using its extensive database of past customer support tickets and their resolutions. The goal is to create a chatbot that can answer new, unique customer queries. After the process, the chatbot can successfully resolve a user's issue even when the user describes their problem using phrasing not seen in the training data. Based on how this training process works, explain the fundamental change that occurred within the model that allows it to handle these novel queries effectively.
0
1
Tags
Ch.3 Prompting - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Analysis in Bloom's Taxonomy
Cognitive Psychology
Psychology
Social Science
Empirical Science
Science
Related
A development team fine-tunes a large, general-purpose language model to act as a specialized chatbot for a financial services company. The training data consists exclusively of question-answer pairs about stock trading, portfolio management, and market analysis. After fine-tuning, the team observes that while the model provides excellent, detailed answers to financial questions, it now struggles to answer simple, general knowledge questions (e.g., 'What is the tallest mountain in the world?') that it could easily answer before the process. Which of the following statements provides the most accurate evaluation of this outcome?
Mechanism of Knowledge Internalization via Fine-Tuning
Analyzing a Failed Fine-Tuning Strategy