Case Study

Diagnosing a Soft Prompt Training Issue

A machine learning engineer is tasked with adapting a large, pre-trained language model for a specialized legal document summarization task. They prepend a set of learnable, continuous prompt vectors to the input and train the system on a large dataset of legal documents and their corresponding summaries. After training, they observe two outcomes: the model performs very well on the legal summarization task, but its performance on general conversational tasks has significantly worsened. Based on the standard method for training these types of prompts, what is the most probable cause of this negative side effect?

0

1

Updated 2025-10-10

Contributors are:

Who are from:

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