0% found this document useful (0 votes)
169 views50 pages

GenerativeAIBootcamp Presentation

This document provides an overview of a presentation deck on Azure AI Content Safety Service. It includes sections on the value proposition, features, technical capabilities, use cases, demos and customer stories. The intent is to help customers understand the capabilities and use cases of the service.

Uploaded by

Noyula
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
169 views50 pages

GenerativeAIBootcamp Presentation

This document provides an overview of a presentation deck on Azure AI Content Safety Service. It includes sections on the value proposition, features, technical capabilities, use cases, demos and customer stories. The intent is to help customers understand the capabilities and use cases of the service.

Uploaded by

Noyula
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 50

[Do not present this slide] Updated: November 2023

Azure AI Content Safety L300 Deck


Instructions: This deck is intended for use with BDM, LOB Analysts, Architects audiences. Use this deck to
present the story & details of Azure AI Content Safety Service. This includes an overview of the value proposition,
features, technical capabilities and use cases, demos, and customer stories. The intent is to help customers build an
in-depth understanding of the capabilities and use cases. Speaker notes are embedded throughout the deck..

You can find more case studies at aka.ms/aicasestudies and demos at aka.ms/aidemos.
Reach out to the AI Solution Sales team if you have any questions.
Generative AI Bootcamp

Leveling Up Your Generative AI in 3 hours or Less!


Learning Goals For Today’s Bootcamp

Build A Foundation of Understand the tools Build a Generative AI


Generative AI Knowledge available to build with application

Understanding new terms and Using tools available to building Applying the concepts into an
technologies and how they key Generative AI use cases application build
connect
Common Use Cases Deploying a model
Exploring Large Language Models
Prompt Engineering Foundations Integrating the model into an application
Comparing Different Model Types
Advanced Prompt Engineering Handling model responses
Improving the Results of a Model
12 Lessons - Everything You Need to Know to Build Generative AI Applications

Find the Repo Here: aka.ms/genai-beginners


Request Access to
Azure OpenAI Service
Apply Here:
https://azure.microsoft.com/products/ai-services/openai-service
Who am I?
Introduction to myself!
Who are you?
Time to find out!

https://www.menti.com/alo59fd8wc6i
Generative AI
Foundations
Understanding the core terms and technologies
Large Language Models (LLMs)
GPT-4 Open Source
Models

Davinci / Llama 2
GPT-4-32k
Babbage
OpenAI Falcon
Models
Dolly v2
Text GPT-3.5-
Embeddings
Ada
Turbo
Mistral
Whisper Dalle3
Understanding the Difference

Criteria Open Source LLMs Proprietary LLMs

Availability Publicly Available and can be used by anyone Owned by an organization, access limited.

Allowed to be inspected and customized for


Customization different use cases than original foundation Limited customization – fine tuning available in certain cases
model

May not be as performant as proprietary


Performance Often optimized for production use
models

Cost Free to use – may require hosting resources May require a subscription or payment based on use

Maintenance May not be maintained in long term Often maintained and updated by model owner
Understanding the Difference
Model Types

Embeddings Models Text Generation Models


Understanding the Difference
Model Types

Chat Completions Image Generation


Exploring the Models
Azure AI Studio - https://ai.azure.com
Exploring the Models
Azure AI Studio
Exploring the Models
Azure AI Studio
Comparing Models
Azure AI Studio
Improving Results

Img source: Four Ways that Enterprises Deploy LLMs | Fiddler AI Blog
Tools Available to Build
with Generative AI
What is out there and when to use it
Generative AI Applications
Common Use Cases

Text Search Image


Text Applications
Prompt Engineering
How does a Prompt Work?

LLMs sees prompt as a


sequence of tokens..
Prompt Engineering
How does a Prompt Work?

LLMs sees prompt as a


sequence of tokens..

https://platform.openai.com/tokenizer
Prompt Engineering
How does a Prompt Work?

Base LLMs will predict


the next token

https://ai.azure.com/playground/gpt-35-turbo
Prompt Engineering
How does a Prompt Work?

Instruction-tuned LLM
extends base behavior
for task
https://ai.azure.com/playground/gpt-35-turbo
Prompt Engineering
How does a Prompt Work?

System Message
•Define the model’s profile, capabilities,
and limitations for your scenario

•Define the model’s output format

•Provide example(s) to demonstrate the


intended behavior of the model

•Provide additional behavioral guardrails


Prompt Engineering
Building in Responsible AI with Metaprompting / System Message
Prompt Engineering
Chat With Your Data

Adding a Data Source


Prompt Engineering

Providing Examples
- Zero Shot Prompting

- Few-Shot Prompting

- Chain of Thought

- Tree of Thought
Model Temperature Temperature .7
Controlling Randomness
No Repetition / More Randomness
Temperature 0
Repetitive / Deterministic
Break – 10 Mins
Search Applications
Understanding Embeddings
How does a LLM find things?

Some
Data
Embedding Vector Store /
Embeddings Database
[1.76,0.33,]
Model

User
Query
Understanding Embeddings
How does a LLM find things?

YouTube
Transcripts
Embedding Vector Store /
Embeddings Database
[1.76,0.33,]
Model

What is
Azure ML?
Understanding Chunking
Fixed-Size Chunking Context Aware Chunking
Exploring Vector Databases
Image Generation
Applications
Image Generation Use Cases

Image Generation Editing Images Creating Variations


Model Accepts:
Model Accepts: Model Accepts:
Image – Original Image
Prompt - Description of Image – Original Image
Image Image Mask – Area for edits
N - # of variations
Size of Image - # of Pixels Prompt - Description of
edits Size of Image – # of Pixels
N - # of images generated
N - # of images generated
Temperature – Randomness
of output Temperature – Randomness
of output Only Available on DALLE-2
Open AI
Exploring Vector Databases
Build a Generative AI
Application
Let’s build something together!
Deploying a Model
Connecting to a Model – Sample Code
Connecting to a Model – Making a Request
Observing the Response
Integrating Responses – Function Calling
LLM
Skill: Beginner
Product: Azure

User: ”I am a beginner
Function
with
find_course(skill, product)
Azure AI”

LLM
API Call
I recommend
/course/{product}/{skill}
AI-900
Integrating Responses – Function Calling
The opportunity is yours to
lead the AI transformation
12 Lessons - Everything You Need to Know to Build Generative AI Applications

Find the Repo Here: aka.ms/genai-beginners

You might also like