Example of a Generic Prompt Template for Information Extraction
A prompt template is used to guide instruction-tuned Large Language Models (LLMs) in performing information extraction tasks. By structuring the input, the template directs the model to extract specific data. A generic example of such a template is:
'You will be provided with a text. Your task is to {*task-description*}
Text: {*text*}'
Here, {*task-description*} is the instruction for the specific task, and {*text*} represents the unstructured data to be processed.
0
1
Tags
Ch.3 Prompting - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Related
Example of a Generic Prompt Template for Information Extraction
A developer is using a large language model that has been specifically adapted to follow commands for pulling structured data from text. The goal is to extract the names of companies and the specific products they launched from the following news snippet:
'Yesterday's tech summit was a showcase of innovation. Innovate Inc. unveiled its new QuantumLeap Processor, while competitor FutureTech debuted its advanced AI-Driven Assistant. Both products are expected to hit the market by year's end.'
Which of the following commands is most likely to produce the desired structured output of company-product pairs?
Crafting an Extraction Command
Diagnosing Inconsistent Extraction Performance
Learn After
Example of a Task Description for Keyword Extraction
Example of a Task Description for Event Extraction
Example of a Task Description for Coreference and Anaphora Resolution
Consider the following instruction given to a language model: 'You will be provided with a text. Your task is to identify and list all the dates mentioned.' Which portion of this instruction constitutes the generic, reusable template that could be adapted for other information extraction goals?
Applying a Prompt Template for a New Task
A data science team needs to create a system that can perform several different information extraction tasks, such as identifying company names, extracting key dates, and listing product features from various documents. Which of the following prompt structures provides the most effective and reusable foundation for these different tasks?