Scribd
Upload
9 views8 pages

Week 3 Day 1

This document outlines the objectives and content for Week 3 Day 1 of a course on LLM Engineering, focusing on open-source tools and the HuggingFace platform. Participants will learn to describe transformers, use Frontier Model APIs, and code with Google Colab. The week aims to prepare students for advanced tasks involving open-source models and various AI applications.

Uploaded by

Ishank Reddy
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
9 views8 pages

Week 3 Day 1

This document outlines the objectives and content for Week 3 Day 1 of a course on LLM Engineering, focusing on open-source tools and the HuggingFace platform. Participants will learn to describe transformers, use Frontier Model APIs, and code with Google Colab. The week aims to prepare students for advanced tasks involving open-source models and various AI applications.

Uploaded by

Ishank Reddy
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/ 8

LLM Engineering

MASTER AI & LARGE LANGUAGE MODELS


WEEK 3 DAY 1

Welcome to Open Source week

What you can now do

Describe transformers and explain key terminology

Confidently code with Frontier Model APIs

Build a multi-model AI Assistant with Tools

After today you will be able to

Describe the HuggingFace platform and everything it offers

Look at Models, Datasets and Spaces

Use Google Colab to code on a high spec GPU runtime


Reminder of the 8 weeks to mastery

Week 3 Week 8
Week 3 - Open Source The Finale - mastering
with HuggingFace Week 4 LLM Engineering
LLM ENGINEER
Selecting LLMs and
Code Generation

Week 2
Frontier Models with UIs,
Agentization and Multi-
Week 7
Week 1 modality
Week 6 Fine-tuning an Open-
What a time to be Source Model
working with LLMs Fine-tuning a Frontier
Week 5 Model

RAG and Question


Answering - Creating an
YOU STARTED Expert
HERE
HuggingFace Platform
The ubiquitous platform for LLM Engineers

Models Datasets Spaces


Over 800,000 Open Source Models of all A treasure trove of Apps, many built in Gradio,
shapes and sizes 200,000 datasets including Leaderboards
HuggingFace Libraries
And the astonishing leg up we get from them

hub datasets transformers

peft trl accelerate


GOOGLE COLAB

Code on a powerful GPU box

Google Colab

Run a Jupyter notebook in the cloud with a powerful runtime

Collaborate with others

Integrate with other Google services

Runtimes

1 CPU based

2 Lower spec GPU runtimes for lower cost

3 Higher spec GPU runtimes for resource intensive runs


GOOGLE COLAB

GPU power in action

pipe = FluxPipeline.from_pretrained("black-forest-
labs/FLUX.1-schnell",
torch_dtype=torch.bfloat16).to("cuda")
generator = torch.Generator(device="cuda").manual_seed(0)
prompt = "A futuristic class full of students learning AI
coding in the surreal style of Salvador Dali"

image = pipe(
prompt,
guidance_scale=0.0,
num_inference_steps=4,
max_sequence_length=256,
generator=generator
).images[0]

image.save("surreal.png")
PROGRESS

Ready for open-source coding

What you can now do

Confidently code with Frontier Models

Build a multi-modal AI Assistant with Tools

Navigate the HuggingPlace platform; run code on Colab

After next time you will be able to run open-source models!

Understand the 2 different levels of HuggingFace API

Use pipelines for a wide variety of AI tasks

Use pipelines to generate text, images and audio

You might also like