Case Study

Refining a Specialized Legal LLM

A team has fine-tuned a large language model to summarize legal clauses from contracts. Despite this training, the model frequently includes extraneous information, such as historical context or related legal precedents, instead of sticking strictly to the summarization task. The team proposes two strategies for an additional round of training to correct this behavior. Evaluate the two strategies and justify which one is more likely to be effective.

0

1

Updated 2025-10-02

Contributors are:

Who are from:

Tags

Ch.4 Alignment - Foundations of Large Language Models

Foundations of Large Language Models

Foundations of Large Language Models Course

Computing Sciences

Evaluation in Bloom's Taxonomy

Cognitive Psychology

Psychology

Social Science

Empirical Science

Science