Large Language Models (LLMs) are machine learning models trained on vast amount of textual data to generate and understand human-like language. These models can perform a wide range of natural language processing tasks from text generation to sentiment analysis and summarisation.
Foundation
Covers the fundamental concepts needed to understand Large Language Models, including language models, foundation models, prompts and key LLM parameters.
ML & Deep Learning Basics
Machine learning and deep learning provide the fundamental concepts behind how LLMs are trained, including neural networks, optimization, loss functions and model generalization.
- Neural Networks
- Activation Functions
- Loss Functions
- Gradient Descent
- Optimization
- Training, Validation & Test Splits
- Underfitting & Overfitting
NLP Fundamentals
Natural Language Processing provides the techniques needed to represent, process and understand human language, including tokenization, embeddings and sequence modeling.
- Tokenization
- Byte Pair Encoding
- Tokens and Context Windows
- Word Embedding
- Recurrent Neural Networks
- Seq2Seq Models
Transformers
Transformers are the foundational architecture behind most modern LLMs that rely on attention mechanisms to process the entire sequence of the data simultaneously.
- Introduction
- Embedding Layers
- Positional Encoding
- Attention Mechanism
- Self-Attention
- Masked Attention
- Multi-Head Attention
- Cross-Attention
- Feed-Forward Neural Network
- Layer Normalization
- Encoder-Decoder Model
- LLMs vs. Transformers
- Transformers from Scratch using TensorFlow
- Transformers from Scratch using PyTorch
LLM Pretraining
LLM pretraining learns language patterns from large amounts of text using self-supervised learning and language modeling techniques.
Prompting Techniques
Prompting techniques guide LLMs to generate desired outputs by carefully designing instructions, examples and reasoning strategies for different tasks.
- Introduction
- Zero-Shot Prompting
- Few-Shot Prompting
- Chain-of-Thought (CoT) Prompting
- Self-Consistency Prompting
- Zero-Shot Chain-of-Thought Prompting
- ReAct (Reasoning + Acting) Prompting
- Prompt Chaining
- Role-Based prompting
- Tree of Thought (ToT) prompting
Fine-Tuning
Fine-tuning adapts a pretrained LLM to specific tasks or domains by further training it on task-specific data using techniques such as SFT, instruction tuning, LoRA and RLHF.
- Introduction
- Supervised Fine-Tuning (SFT)
- Instruction Tuning
- Parameter-Efficient Fine-Tuning (PEFT)
- LoRA (Low-Rank Adaptation)
- QLoRA (Quantized Low-Rank Adaptation)
- Prompt Tuning
- Reinforcement Learning from Human Feedback (RLHF)
- Fine-Tune an LLM from Hugging Face
- LLM Distillation
Retrieval-Augmented Generation (RAG)
This section explains how RAG combines information retrieval with language models to generate responses using external knowledge sources.
- Introduction
- Fine tuning vs RAG
- Embedding models
- Chunking
- Vector Database
- Dense Passage Retrieval (DPR)
- RAG Pipeline
- Retrieval-Augmented Prompting
- Agentic RAG
- Mutlimodal RAG
Agents & Tool Use
LLM agents extend language models beyond text generation by enabling them to use tools, access external information, maintain memory and perform multi-step tasks.
Evaluation & Safety
LLM evaluation and safety techniques help measure model performance and address challenges such as hallucinations, prompt injection, harmful outputs and responsible AI concerns.
- LLM Evaluation
- LLM Hallucinations
- Perplexity
- BLEU and ROUGE
- Benchmarks
- LLM-as-a-judge
- Prompt Injection
- Guardrails
- Ethical Implications
- Responsible AI
Popular Models
Popular LLMs use different architectures, training approaches and capabilities to support tasks such as text generation, reasoning, coding and multimodal understanding.
Projects
Practical projects demonstrate how LLMs can be applied to build real-world applications such as chatbots, RAG systems, summarizers, sentiment analysis systems and translation tools.
- Building Chatbot using Gemini
- Building Chatbot using OpenAI
- RAG using Llama 3
- PDF Summarizer using RAG
- Sentiment analysis using BERT
- Text2text generation using Hugging Face Model
- Machine Translation using Transformer
- Customer Help Bot with RAG
For interview preparation, you can refer to this article - Generative AI and LLM Interview Question with Answer