-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathindex.html
173 lines (157 loc) · 9.44 KB
/
index.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0, maximum-scale=1.0, user-scalable=no">
<title>Open Source AI</title>
<link rel="stylesheet" href="dist/reset.css">
<link rel="stylesheet" href="dist/reveal.css">
<link rel="stylesheet" href="dist/theme/black.css">
<!-- Theme used for syntax highlighted code -->
<link rel="stylesheet" href="plugin/highlight/monokai.css">
</head>
<body>
<div class="reveal">
<div class="slides">
<section>
<h1>Open Source AI</h1>
<br />
<p>by JJ Asghar | <a target="_blank" href="https://bsky.app/profile/jjasghar.bsky.social">@jjasghar</a> | <a target="_blank" href="mailto:[email protected]">[email protected]</a></p>
</section>
<section>
<h1>But first Granite!</h1>
<small>I need you to scan this qrcode to prove to my boss you're at my talk ❤️</small>
<img src="./images/granite-main.png" />
<aside class="notes">
</aside>
</section>
<section>
<section><p>What is an open source model?</p></section>
<section>
<p>The largest problem is that AI models are binaries, and the "source code" isn't just the data, it's so much more.</p>
</section>
<section><img src="./images/mof.jpg" /></section>
<section>
<p>tl;dr: Everyone is debating what "open source" means and we can't seem to come to an agreement, but the Model Openness Framework seems to be the most supported.</p>
</section>
</section>
<section>
<section>
<img src="https://github.com/instructlab/.github/blob/main/assets/instructlab-banner.png?raw=true" />
<p>What is InstructLab and what does it solve?</p>
<aside class="notes">
You have learned that this is the difference between knowing you need a car or transport, to a sports car or a truck.
</aside>
</section>
<section><p>What is fine tuning?</p></section>
<section><p><bold>tl;dr</bold> InstructLab is an opinionated workflow for fine tuning models.</p></section>
<section><iframe src="https://docs.google.com/presentation/d/e/2PACX-1vS2K5wSUhi6WWtWvV5M3JazDi_eSNlZDq_6WNrTQl8zddE-vtb9EUudjzeB_14MQPx2NK2Ed_xnYo9i/embed?start=false&loop=true&delayms=60000" frameborder="0" width="1440" height="839" allowfullscreen="true" mozallowfullscreen="true" webkitallowfullscreen="true"></iframe></section>
<section><p>A <a target="_blank" href="https://github.com/instructlab/taxonomy/compare/main...cdoern:taxonomy:moo-deng">qna.yaml</a> example</p></section>
<section><p>A <a target="_blank" href="https://github.com/cdoern/knowledge/blob/main/pop_culture/moodeng.md">source markdown document</a> example</p></section>
</section>
<section>
<section><p>RAG vs Fine Tuning</p></section>
<section>
<p>Imagine an AI model as an Librarian</p>
<aside class="notes">
It might know that George Washington's home is Mount Veron, Virginia.
But not know that George Washington died of a throat infection.
</aside>
</section>
<section><p>RAG: Retrieval-Augmented Generation or the "library"</p></section>
<section><p>Fine Tuning goes into the "brain", so sending the librarian "back to school"</p></section>
</section>
<section>
<section><p>Agents and Agentic Workflows</p></section>
<section><p>Let's talk quickly about agentic frameworks...</p></section>
<section><p>What is an "agent?"</p></section>
<section>
<p>A few major frameworks people are graviting towards:</p>
<ol>
<li><a target="_blank" href="https://i-am-bee.github.io/beeai-framework/#/">Bee Framework</a></li>
<li><a target="_blank" href="https://www.crewai.com">crewai</a> (a great crash course <a href="https://www.youtube.com/watch?v=ONKOXwucLvE">video</a>)</li>
<li><a target="_blank" href="https://github.com/microsoft/autogen">autogen</a></li>
</ol>
</section>
</section>
<section>
<section>
<p>But what model should we tune?</p>
<aside class="notes">
You have learned that this is the difference between knowing you need a car or transport, to a sports car or a truck.
</aside>
</section>
<section><img src="./images/hugging-face-models.png" /><p>This screenshot was taken March 10, 2025... 1.4 million seems a lot.</p></section>
</section>
<section>
<section>
<p>And this brings us back to Granite</p>
<img src="./images/granite-logo.png" />
</section>
<section>
<iframe width="560" height="315" src="https://www.youtube.com/embed/cVDv9apGTXo?si=1HxxJdHEzcbConh8" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen></iframe>
<ul>
<li><a target="_blank" href="https://www.ibm.com/granite/docs/models/granite/">Granite 3.2</a></li>
<li><a target="_blank" href="https://www.ibm.com/granite/docs/models/vision/">Granite Vision</a></li>
<li><a target="_blank" href="https://www.ibm.com/granite/docs/models/guardian/">Granite Guardian</a></li>
<li><a target="_blank" href="https://www.ibm.com/granite/docs/models/embedding/">Granite Embedding</a></li>
<li><a target="_blank" href="https://www.ibm.com/granite/docs/models/time-series/">Granite Time Series</a></li>
<li><a target="_blank" href="https://www.ibm.com/granite/docs/models/code/">Granite Code</a></li>
</ul>
</section>
</section>
<section>
<section><p>Use Cases for Granite</p></section>
<section><p><a target="_blank" href="https://github.com/jjasghar/ai-abstract-classifier">ai-abstract-classifier</a>: Run abstracts through a local LLM to give you some possible scores on what you are looking for.</p></section>
<section>
<p>Leverage the <a target="_blank" href="https://www.ibm.com/granite/docs/models/time-series/">Granite Time Series</a> model to project capacity planning </p>
<small>I think I've figured out how to push <a target="_blank" href="https://nagios.org">Nagios data</a> into the model to project possible alerts...I'm hoping to have a link here when it's ready to show off...</small>
</section>
<section>
<p>Leverage <a target="_blank" href="https://www.ibm.com/granite/docs/models/granite/">Granite 3.2</a> like this to create stories for some kids. This is a picture of me teaching AI at a STEM Night at an school.</p>
<img src="./images/IMG_2011.jpg" />
</section>
<section>
<p>Or even better come join me in the Granite Lounge to help me figure out the process of using <a target="_blank" href="https://www.ibm.com/granite/docs/models/time-series/">Granite Time Series</a> model to project my insulin pump supplies and insulin usage</p>
<br />
<small>I have the data, and I'm making progress, but it would be awesome to learn with you!</small>
</section>
</section>
<section>
<section><p>But JJ, how can I play with Granite?</p></section>
<section>
<p><a target="_blank" href="https://ibm.github.io/opensource-ai-workshop/">Open Source Workshop</a></p>
<img src="./images/opensource-workshop.png"/ >
<p>Spend some time learning how to get Granite and what is required to run it locally, securely, and privately on your laptop.</p>
</section>
<section>
<p><a target="_blank" href="https://ibm.github.io/granite-workshop/">Granite Workshop</a></p>
<img src="./images/granite-workshop.png"/ >
<p>Learn how to leverage the different Granite models to integrate with applications and developer focused work.</p>
</section>
</section>
<section>
<p>Here is a qrcode to this presentation</p>
<img src="./images/jj-github-link.png" />
<p>Thank you!</p>
<p>JJ Asghar | <a target="_blank" href="https://bsky.app/profile/jjasghar.bsky.social">@jjasghar</a> | <a target="_blank" href="mailto:[email protected]">[email protected]</a></p>
</section>
</div>
</div>
<script src="dist/reveal.js"></script>
<script src="plugin/notes/notes.js"></script>
<script src="plugin/markdown/markdown.js"></script>
<script src="plugin/highlight/highlight.js"></script>
<script>
// More info about initialization & config:
// - https://revealjs.com/initialization/
// - https://revealjs.com/config/
Reveal.initialize({
hash: true,
slideNumber: true,
// Learn about plugins: https://revealjs.com/plugins/
plugins: [ RevealMarkdown, RevealHighlight, RevealNotes ]
});
</script>
</body>
</html>