Reader small image

You're reading from  The Deep Learning Architect's Handbook

Product typeBook
Published inDec 2023
PublisherPackt
ISBN-139781803243795
Edition1st Edition
Right arrow
Author (1)
Ee Kin Chin
Ee Kin Chin
author image
Ee Kin Chin

Ee Kin Chin is a Senior Deep Learning Engineer at DataRobot. He holds a Bachelor of Engineering (Honours) in Electronics with a major in Telecommunications. Ee Kin is an expert in the field of Deep Learning, Data Science, Machine Learning, Artificial Intelligence, Supervised Learning, Unsupervised Learning, Python, Keras, Pytorch, and related technologies. He has a proven track record of delivering successful projects in these areas and is dedicated to staying up to date with the latest advancements in the field.
Read more about Ee Kin Chin

Right arrow

Handling knowledge for LLM solutions

Domain knowledge is key in creating LLM solutions as it provides the background information and understanding they need to solve specific problems. This ultimately ensures the answers or actions the solutions come up with are on point and helpful. Domain knowledge needs to be included as context either as part of parametric memory, non-parametric memory, or a combination of both. Parametric memory refers to the parameters that are learned in an LLM. Non-parametric memory refers to an external library of knowledge, such as a list of documents, articles, or excerpts, that can be selectively chosen to be injected as part of the LLM context. This process is also referred to as an in-context learning method, knowledge retrieval, or information retrieval.

Non-parametric external knowledge can be provided to an LLM through either of the following options:

  • As a latent conditioner in the cross-attention mechanism: Latent conditioning involves...
lock icon
The rest of the page is locked
Previous PageNext Page
You have been reading a chapter from
The Deep Learning Architect's Handbook
Published in: Dec 2023Publisher: PacktISBN-13: 9781803243795

Author (1)

author image
Ee Kin Chin

Ee Kin Chin is a Senior Deep Learning Engineer at DataRobot. He holds a Bachelor of Engineering (Honours) in Electronics with a major in Telecommunications. Ee Kin is an expert in the field of Deep Learning, Data Science, Machine Learning, Artificial Intelligence, Supervised Learning, Unsupervised Learning, Python, Keras, Pytorch, and related technologies. He has a proven track record of delivering successful projects in these areas and is dedicated to staying up to date with the latest advancements in the field.
Read more about Ee Kin Chin