A comprehensive guide to learning LLMs (Foundational Models)

Sre Chakra Yeddula

·

Follow

4 min read

·

Jun 14, 2023

--

Listen

Share

The LLM (Foundational Models) space has seen tremendous and rapid growth. With so much information out there. It might feel like being lost at sea. Having gone through this exercise myself, I wanted to make this easier for folks starting the journey (or even in the midst of it). I used this foolproof method of consuming the right information and ended up publishing books, artworks, Podcasts and even an LLM powered consumer facing app ranked #40 on the app store. Follow my entire journey here.

I have been going through various training modules, including this one by Google Cloud and this one by Databricks and other learning paths. I crunched all the information to provide an easy way for folks to understand the space; with curated videos (some of them unlisted!) and resources from popular channels and providers, I have laid them all into one massive mind map.

Although the current version of it focuses on text modality, I intend to expand this to include additional modalities over the below link for the entire interactive mind map.

LLM Learning MindMap: Lucidspark

Here is a print friendly view of all the resources.

Learning LLMs (Foundational Models)

1. Base Knowledge / Concepts:

What is AI, ML and NLP

Introduction to ML and AI — MFML Part 1 — YouTube
What is NLP (Natural Language Processing)? — YouTube

Introduction to Natural Language Processing (NLP)

NLP 2012 Dan Jurafsky and Chris Manning (1.1) Intro to NLP — YouTube
NLTK with Python 3 for Natural Language Processing — YouTube

Understanding Text Representation and Feature Extraction

• Bag of Words, TF-IDF, Word Embeddings (Word2Vec, GloVe):

NLP Concepts: TF-IDF, Bag-of-Words Explained with Examples — YouTube Calculate TF-IDF in NLP (Simple Example)

• Understanding Word2Vec and its implementation

Lecture 2 | Word Vector Representations: word2vec — YouTube

Use case example.

• Understanding Text Classification (Naive Bayes, SVM, Neural Networks)

Text Classification Using Naive Bayes | Naive Bayes Algorithm In Machine Learning | Simplilearn — YouTube
Support Vector Machine (SVM) in 2 minutes — YouTube Neural Networks Explained in 5 minutes — YouTube

Sequence learning models

• Introduction to RNNs and LSTMs

Recurrent Neural Networks (RNN) and Long Short-Term Memory (LSTM) — YouTube Recurrent Neural Networks (RNNs), Clearly Explained!!! — YouTube

• Introduction to Sequence Learning and Attention Mechanisms

Stanford CS224N: NLP with Deep Learning | Winter 2019 | Lecture 8 — Translation, Seq2Seq, Attention — YouTube
Stanford CS224N NLP with Deep Learning | Winter 2021 | Lecture 7 — Translation, Seq2Seq, Attention — YouTube

2. LLMs (Foundational Models) 101:
Introduction to Transformer Models
Transformers, explained: Understand the model behind GPT, BERT, and T5 — YouTube
Illustrated Guide to Transformers Neural Network: A step by step explanation — YouTube

• Attention Mechanism Deep dive.

Transformer Neural Networks — EXPLAINED! (Attention is all you need) — YouTube
Attention Mechanism: Overview — YouTube

• Encoder-Decoder Deep dive.

Encoder-Decoder Architecture: Overview — YouTube

Lab:

Encoder-Decoder Architecture: Lab Walkthrough — YouTube

• Tokenization, Embedding and encoding

What are Transformer Models and How do they Work? — YouTube Transformer Models (cohere.com)

Intro to BERT (early LLM example)

BERT Neural Network — EXPLAINED! — YouTube
BERT Research — Ep. 1 — Key Concepts & Sources — YouTube

PreTraining an LLM

NLP Demystified 15: Transformers From Scratch + Pre-training and Transfer Learning With BERT/GPT — YouTube

Finetuning LLMs

Finetuning Large Language Models — by Sebastian Raschka

3. Building a text generation model from scratch:

• N-gram Language model — beginner

Generating Sentences with n-grams using Python — YouTube

• with RNNs and Seq2seq — Intermediate

Text generation with an RNN | TensorFlow
Neural machine translation with attention | Text | TensorFlow

• Transformer from scratch — Advanced

Neural machine translation with a Transformer and Keras | Text | TensorFlow
Short — Pytorch Transformers from Scratch (Attention is all you need) — YouTube
Long duration — Transformers from scratch

4. Using LLMs

Use vs build ($) — Pretrained Vs Train Vs Finetune LLMs

Pretraining vs Fine-tuning vs In-context Learning of LLM (GPT-x) EXPLAINED | Ultimate Guide ($) — YouTube

Using 3rd party models

• Proprietary

-OpenAI

Getting Started with OpenAI API and GPT-3 | Beginner Python Tutorial — YouTube
Introduction — OpenAI API

• Open Source

🤗 Transformers (huggingface.co)
Training Sentiment Model Using BERT and Serving it with Flask API — YouTube

5. Deploy LLMs in production

• Deploy Model

Azure — Use endpoints for inference — Azure Machine Learning | Microsoft Learn

AWS + Huggingface — Exporting 🤗 Transformers Models (huggingface.co)

Google Cloud — Build, tune, and deploy foundation models with Vertex AI — YouTube

GitHub — GoogleCloudPlatform/llm-pipeline-examples

• Deploy Database (Vector DBs)

Vector Databases and Large Language Models // Samuel Partee // LLMs in Production Conference — YouTube

• Evaluate Model

LLMOps (LLM Bootcamp) — YouTube

• Deploy + tools (langchain)

LangChain 101: Quickstart Guide — YouTube

Hope this gives you a good pathway to develop your skills. Happy learning.

Originally published at https://www.linkedin.com.

BECOME a WRITER at MLearning.ai //Try These FREE ML Tools Today!