Learn Before
A company has a large, pre-trained language model and needs to quickly deploy it for two distinct new tasks: summarizing legal documents and generating marketing copy. Instead of creating two separate, retrained versions of the model, they decide to guide the original model's behavior using specific, task-oriented instructions for each request. What is the fundamental reason this approach is considered highly efficient in terms of computational resources and time?
0
1
Tags
Ch.1 Pre-training - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Ch.2 Generative Models - Foundations of Large Language Models
Analysis in Bloom's Taxonomy
Cognitive Psychology
Psychology
Social Science
Empirical Science
Science
Related
Paradigm Shift in NLP due to Prompting
User Customization of LLMs via Prompt Design
Efficient Model Adaptation for a Startup
A company has a large, pre-trained language model and needs to quickly deploy it for two distinct new tasks: summarizing legal documents and generating marketing copy. Instead of creating two separate, retrained versions of the model, they decide to guide the original model's behavior using specific, task-oriented instructions for each request. What is the fundamental reason this approach is considered highly efficient in terms of computational resources and time?
The primary reason that adapting a pre-trained language model using task-specific instructions is considered highly efficient is because this method involves making minor, incremental updates to the model's internal weights with each new instruction.