-
Notifications
You must be signed in to change notification settings - Fork 0
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
Showing
15 changed files
with
712 additions
and
18 deletions.
There are no files selected for viewing
104 changes: 104 additions & 0 deletions
104
chapters/11_training_llms/11_01_compute_memory/index.html
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,104 @@ | ||
<!DOCTYPE html> | ||
<html><head> | ||
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8"> | ||
<meta charset="utf-8"> | ||
<meta name="viewport" content="width=device-width, initial-scale=1.0"> | ||
<link href="https://cdn.jsdelivr.net/npm/[email protected]/dist/css/bootstrap.min.css" rel="stylesheet" integrity="sha384-EVSTQN3/azprG1Anm3QDgpJLIm9Nao0Yz1ztcQTwFspd3yD65VohhpuuCOmLASjC" crossorigin="anonymous"> | ||
<link rel="stylesheet" type="text/css" href="/dl4nlp/css/style.css"> | ||
|
||
|
||
<title>Deep Learning for Natural Language Processing (DL4NLP) | Chapter 11.01: Memory and compute requirements</title> | ||
|
||
|
||
<link rel="apple-touch-icon" sizes="180x180" href="/dl4nlp/apple-touch-icon.png"> | ||
<link rel="icon" type="image/png" sizes="32x32" href="/dl4nlp/favicon-32x32.png"> | ||
<link rel="icon" type="image/png" sizes="16x16" href="/dl4nlp/favicon-16x16.png"> | ||
<link rel="manifest" href="/dl4nlp/site.webmanifest"> | ||
<link rel="mask-icon" href="/dl4nlp/safari-pinned-tab.svg" color="#5bbad5"> | ||
<meta name="msapplication-TileColor" content="#da532c"> | ||
<meta name="theme-color" content="#ffffff"> | ||
|
||
</head><body> | ||
<img id="logo" src="/dl4nlp/dl4nlp.svg" /> | ||
|
||
<div id="nav-border" class="container"> | ||
<nav id="nav" class="nav justify-content-center"> | ||
|
||
<a class="nav-link" href="/dl4nlp"> | ||
|
||
Home | ||
</a> | ||
|
||
<a class="nav-link" href="/dl4nlp/chapters/"> | ||
|
||
Chapters | ||
</a> | ||
|
||
<a class="nav-link" href="/dl4nlp/appendix/"> | ||
|
||
Appendix | ||
</a> | ||
|
||
<a class="nav-link" href="/dl4nlp/exercises/"> | ||
|
||
Exercises | ||
</a> | ||
|
||
<a class="nav-link" href="/dl4nlp/references/"> | ||
|
||
References | ||
</a> | ||
|
||
<a class="nav-link" href="/dl4nlp/team/"> | ||
|
||
Team | ||
</a> | ||
|
||
</nav> | ||
</div><div id="content" class="container"> | ||
<h1>Chapter 11.01: Memory and compute requirements</h1> | ||
<p>Large language models (LLMs) require significant compute and memory resources due to their vast number of parameters and complex architectures. In this chapter you will learn about different contributions to compute requirements and how model size components influence memory requirements.</p> | ||
<h3 id="lecture-slides">Lecture Slides</h3> | ||
|
||
|
||
<script src="https://cdn.jsdelivr.net/npm/[email protected]/build/pdf.min.js" integrity="sha256-hEmjt7z3bB53X/awJyV81gmBLpVw2mj7EsvoJelZWow=" crossorigin="anonymous"></script> | ||
|
||
|
||
|
||
|
||
|
||
|
||
<a href="https://github.com/slds-lmu/lecture_dl4nlp/raw/main/slides/chapter11-training-llms/slides-111-compute-memory.pdf"> | ||
<button class="btn btn-primary" style="margin-bottom:3rem"> | ||
Download »slides-111-compute-memory.pdf« | ||
</button> | ||
</a> | ||
|
||
|
||
<ul class="section_skipper list-unstyled"> | ||
|
||
|
||
<li id="prev_in_section"><a class="btn btn-primary" href="https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/11_02_reduce_comp/">Chapter 11.02: How to reduce memory and compute? »</a></li> | ||
|
||
</ul> | ||
|
||
|
||
|
||
|
||
</div><footer class="bg-light text-center text-lg-start fixed-bottom"> | ||
<ul class="list-inline text-center"> | ||
<li class="list-inline-item">© 2022 Course Creator</li> | ||
|
||
<li class="list-inline-item"><a class="nav-link" href="https://slds-lmu.github.io/i2ml/" target="_blank">I2ML Course</a></li> | ||
|
||
<li class="list-inline-item"><a class="nav-link" href="https://github.com/slds-lmu/lecture_dl4nlp" target="_blank">Material Source Code</a></li> | ||
|
||
<li class="list-inline-item"><a class="nav-link" href="https://github.com/slds-lmu/dl4nlp" target="_blank">Website source code</a></li> | ||
|
||
</ul> | ||
</footer> | ||
|
||
|
||
|
||
</body> | ||
</html> |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,106 @@ | ||
<!DOCTYPE html> | ||
<html><head> | ||
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8"> | ||
<meta charset="utf-8"> | ||
<meta name="viewport" content="width=device-width, initial-scale=1.0"> | ||
<link href="https://cdn.jsdelivr.net/npm/[email protected]/dist/css/bootstrap.min.css" rel="stylesheet" integrity="sha384-EVSTQN3/azprG1Anm3QDgpJLIm9Nao0Yz1ztcQTwFspd3yD65VohhpuuCOmLASjC" crossorigin="anonymous"> | ||
<link rel="stylesheet" type="text/css" href="/dl4nlp/css/style.css"> | ||
|
||
|
||
<title>Deep Learning for Natural Language Processing (DL4NLP) | Chapter 11.02: How to reduce memory and compute?</title> | ||
|
||
|
||
<link rel="apple-touch-icon" sizes="180x180" href="/dl4nlp/apple-touch-icon.png"> | ||
<link rel="icon" type="image/png" sizes="32x32" href="/dl4nlp/favicon-32x32.png"> | ||
<link rel="icon" type="image/png" sizes="16x16" href="/dl4nlp/favicon-16x16.png"> | ||
<link rel="manifest" href="/dl4nlp/site.webmanifest"> | ||
<link rel="mask-icon" href="/dl4nlp/safari-pinned-tab.svg" color="#5bbad5"> | ||
<meta name="msapplication-TileColor" content="#da532c"> | ||
<meta name="theme-color" content="#ffffff"> | ||
|
||
</head><body> | ||
<img id="logo" src="/dl4nlp/dl4nlp.svg" /> | ||
|
||
<div id="nav-border" class="container"> | ||
<nav id="nav" class="nav justify-content-center"> | ||
|
||
<a class="nav-link" href="/dl4nlp"> | ||
|
||
Home | ||
</a> | ||
|
||
<a class="nav-link" href="/dl4nlp/chapters/"> | ||
|
||
Chapters | ||
</a> | ||
|
||
<a class="nav-link" href="/dl4nlp/appendix/"> | ||
|
||
Appendix | ||
</a> | ||
|
||
<a class="nav-link" href="/dl4nlp/exercises/"> | ||
|
||
Exercises | ||
</a> | ||
|
||
<a class="nav-link" href="/dl4nlp/references/"> | ||
|
||
References | ||
</a> | ||
|
||
<a class="nav-link" href="/dl4nlp/team/"> | ||
|
||
Team | ||
</a> | ||
|
||
</nav> | ||
</div><div id="content" class="container"> | ||
<h1>Chapter 11.02: How to reduce memory and compute?</h1> | ||
<p>Here you will learn about ways to reduce the memory and compute requirements for big models. We introduce distributed training, where you make use of data- and tensor parallellism, and FlashAttention, a method to perform attention more efficiently.</p> | ||
<h3 id="lecture-slides">Lecture Slides</h3> | ||
|
||
|
||
<script src="https://cdn.jsdelivr.net/npm/[email protected]/build/pdf.min.js" integrity="sha256-hEmjt7z3bB53X/awJyV81gmBLpVw2mj7EsvoJelZWow=" crossorigin="anonymous"></script> | ||
|
||
|
||
|
||
|
||
|
||
|
||
<a href="https://github.com/slds-lmu/lecture_dl4nlp/raw/main/slides/chapter11-training-llms/slides-112-reduce-comp.pdf"> | ||
<button class="btn btn-primary" style="margin-bottom:3rem"> | ||
Download »slides-112-reduce-comp.pdf« | ||
</button> | ||
</a> | ||
|
||
|
||
<ul class="section_skipper list-unstyled"> | ||
|
||
<li id="next_in_section"><a class="btn btn-primary" href="https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/11_01_compute_memory/">« Chapter 11.01: Memory and compute requirements</a></li> | ||
|
||
|
||
<li id="prev_in_section"><a class="btn btn-primary" href="https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/11_03_scaling/">Chapter 11.03: Scaling Laws and Chinchilla »</a></li> | ||
|
||
</ul> | ||
|
||
|
||
|
||
|
||
</div><footer class="bg-light text-center text-lg-start fixed-bottom"> | ||
<ul class="list-inline text-center"> | ||
<li class="list-inline-item">© 2022 Course Creator</li> | ||
|
||
<li class="list-inline-item"><a class="nav-link" href="https://slds-lmu.github.io/i2ml/" target="_blank">I2ML Course</a></li> | ||
|
||
<li class="list-inline-item"><a class="nav-link" href="https://github.com/slds-lmu/lecture_dl4nlp" target="_blank">Material Source Code</a></li> | ||
|
||
<li class="list-inline-item"><a class="nav-link" href="https://github.com/slds-lmu/dl4nlp" target="_blank">Website source code</a></li> | ||
|
||
</ul> | ||
</footer> | ||
|
||
|
||
|
||
</body> | ||
</html> |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -1 +1 @@ | ||
<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Chapter 11: Training Large Language Models on Deep Learning for Natural Language Processing (DL4NLP)</title><link>https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/</link><description>Recent content in Chapter 11: Training Large Language Models on Deep Learning for Natural Language Processing (DL4NLP)</description><generator>Hugo</generator><language>en-us</language><atom:link href="https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/index.xml" rel="self" type="application/rss+xml"/><item><title>Chapter 11.01: LLMs: Parameters, Data, Hardware, Scaling</title><link>https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/11_01_compute_scaling_chinchilla/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/11_01_compute_scaling_chinchilla/</guid><description><p>In this chapter you will learn how to calculate the number of parameters in the Transformer, understand Transformer computation and memory load, learn about Flash Attentions and understand Scaling Laws and Chinchilla.</p></description></item><item><title>Chapter 11.02: LLM Optimization</title><link>https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/11_02_x_optimize/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/11_02_x_optimize/</guid><description><p>In this Chapter we discuss ways to optimize the performance of Large Language Models (LLMs) with methods such as Prompt engineering or methods beyond that.</p></description></item></channel></rss> | ||
<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Chapter 11: Training Large Language Models on Deep Learning for Natural Language Processing (DL4NLP)</title><link>https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/</link><description>Recent content in Chapter 11: Training Large Language Models on Deep Learning for Natural Language Processing (DL4NLP)</description><generator>Hugo</generator><language>en-us</language><atom:link href="https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/index.xml" rel="self" type="application/rss+xml"/><item><title>Chapter 11.01: Memory and compute requirements</title><link>https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/11_01_compute_memory/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/11_01_compute_memory/</guid><description><p>Large language models (LLMs) require significant compute and memory resources due to their vast number of parameters and complex architectures. In this chapter you will learn about different contributions to compute requirements and how model size components influence memory requirements.</p></description></item><item><title>Chapter 11.02: How to reduce memory and compute?</title><link>https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/11_02_reduce_comp/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/11_02_reduce_comp/</guid><description><p>Here you will learn about ways to reduce the memory and compute requirements for big models. We introduce distributed training, where you make use of data- and tensor parallellism, and FlashAttention, a method to perform attention more efficiently.</p></description></item><item><title>Chapter 11.03: Scaling Laws and Chinchilla</title><link>https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/11_03_scaling/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/11_03_scaling/</guid><description><p>In this chapter we introduce various scaling laws and chinchilla.</p></description></item><item><title>Chapter 11.04: LLM Optimization</title><link>https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/11_04_x_optimize/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://slds-lmu.github.io/dl4nlp/chapters/11_training_llms/11_04_x_optimize/</guid><description><p>In this Chapter we discuss ways to optimize the performance of Large Language Models (LLMs) with methods such as Prompt engineering or methods beyond that.</p></description></item></channel></rss> |
Oops, something went wrong.