Diagnosing Fine-Tuning Issues with Instruction Formatting
Given the following case study about a fine-tuning project, identify the most likely flaw in the team's approach to providing instructions and propose a specific change to the instruction format to improve the model's performance.
0
1
Tags
Ch.3 Prompting - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Ch.1 Pre-training - Foundations of Large Language Models
Analysis in Bloom's Taxonomy
Cognitive Psychology
Psychology
Social Science
Empirical Science
Science
Related
A development team is adapting a single pre-trained language model to handle two new, distinct functions. The first function is to classify customer feedback emails into one of three categories: 'Positive', 'Negative', or 'Neutral'. The second function is to generate a formal, two-paragraph apology letter in response to a customer complaint, adhering to a specific company tone and style. The team is preparing the training data and must decide how to format the textual instructions for the model. Which of the following strategies for formatting the instructions is most effective for this scenario?
Diagnosing Fine-Tuning Issues with Instruction Formatting
Comparing Instruction Formatting Strategies in Model Fine-Tuning