Learn Before
Instruction Following as a Prerequisite for Prompting
While using templates to prompt a Large Language Model is an effective way to direct its behavior to perform various tasks without adapting its underlying structure, the success of this technique fundamentally depends on the model's inherent capacity to accurately recognize and follow the given instructions or questions.
0
1
Tags
Ch.2 Generative Models - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Related
Prompt Engineering
Example of a Prompt for Machine Translation
A user wants a language model to generate a summary of a long article, but specifically for a non-expert audience. They provide the model with the following input: 'Summarize the attached article in simple, easy-to-understand language, as if you were explaining it to a middle school student.' What is the primary function of this input technique?
Analyzing Model Inputs
The Purpose of Prompting
Instruction Following as a Prerequisite for Prompting
Learn After
A developer is building a tool that allows users to generate custom email responses by providing specific commands, such as 'Write a polite refusal to this meeting invitation.' They have two language models to choose from:
- Model X: A very large model trained on a vast library of books and websites, excelling at generating fluent, human-like text.
- Model Y: A smaller model specifically trained on a dataset of commands paired with their correct outputs.
Which model is the more suitable choice for this tool, and why?
Enabling Instruction Following via Pre-training
Diagnosing a Language Model's Output
Analyzing Model Failure