Free · 159 Topics · No Signup
Generative AI Notes
Prompting and text-to-text models — ChatGPT, Bard and AI chatbots — written by Coding Now Tech Institute's mentors. Free to read, structured to actually help you learn.
Generative AI notes by Coding Now Tech Institute cover 159 topics — from ai prompt home to tool calling vs agent — each explained with short definitions, syntax and runnable code examples. They are 100% free, need no signup, and work as quick revision for college exams, Generative AI interviews and Coding Now Tech Institute's mentor-led Generative AI course in Pitampura, Delhi.
AI Prompt Tutorial
AI Prompt Tutorial chapters
AI text-to-text
AI text-to-text chapters
AI Fundamentals
What generative AI is, how it differs from AI/ML/DL, and how it works.
LLMs
How large language models are trained, run, and their real limitations.
Tokens
Tokenization and why it drives cost, context limits and latency.
Transformers
Attention, self-attention and the architecture behind modern LLMs.
Prompt Engineering
Structuring prompts, few-shot patterns, chaining and prompt security.
LLM APIs
Chat completions, sampling parameters, streaming and provider-agnostic API design.
Structured Output
Getting reliable JSON out of an LLM, and validating it.
Embeddings
Turning text into vectors, and measuring semantic similarity.
Vector Databases
Storing and searching embeddings at scale — indexes and hybrid search.
RAG
Retrieval-augmented generation — chunking, retrieval, and evaluating RAG systems.
Fine-tuning
When to fine-tune vs prompt vs RAG, and parameter-efficient methods like LoRA.
Multimodal AI
Vision-language models, speech, image/video generation, and multimodal RAG.
Tool Calling
Letting an LLM call functions/APIs, and validating those calls safely.
No notes found. Try a different search term, or browse all Generative AI notes.
AI Prompt Tutorial
2 of 2 topics publishedAI Prompt Home
In these tutorials, you get step-by-step guides on how to write AI prompts to get the best possible results from text-to-text and text-to-image generative AIs.
AI Prompt Intro
When talking about Artificial Intelligence (AI) in the tech industry, we refer to intelligent machines that can do jobs that previously needed human intelligenc
AI text-to-text
4 of 4 topics publishedAI Fundamentals
8 of 8 topics publishedWhat Is Generative AI?
What generative AI actually is, and how it differs from discriminative AI.
Generative AI vs AI
Why generative AI is one capability within the broader field of AI.
Generative AI vs Machine Learning
How generative AI relates to traditional machine learning techniques.
Generative AI vs Deep Learning
How generative AI relates to deep learning, and the transformer breakthrough.
Types of Generative AI
LLMs, diffusion models, GANs and the different generative AI architectures.
Generative AI Applications & Use Cases
Real production use cases, and why narrow tasks work better than open-ended assistants.
How Generative AI Works
The training-to-inference pipeline behind text generation, step by step.
Generative AI Limitations, Risks & Challenges
Technical limitations, operational risks and security/ethical challenges.
LLMs
13 of 13 topics publishedWhat Is an LLM?
What makes a language model "large," and what it actually learns from training.
How LLMs Work
The training vs inference split, and where prompting and fine-tuning fit in.
LLM Architecture
The decoder-only transformer pipeline, from tokens to next-token prediction.
LLM Training
The three-stage training pipeline: pretraining, instruction tuning, alignment.
Pretraining
Self-supervised learning on raw text, and the knowledge cutoff it creates.
Instruction Tuning
Teaching a base model to follow instructions and respond helpfully.
Alignment
RLHF and preference-based training toward helpful, honest, harmless behavior.
LLM Inference
Prefill vs decode, KV caching, and why output length drives latency.
Context Window
What counts toward it, why conversations "forget," and long-context tradeoffs.
LLM Parameters (and Model Weights)
What a parameter/weight actually is, and what "7B" or "70B" means.
Model Size
The capability-vs-cost tradeoff, and how to choose the right model size.
LLM Limitations
Knowledge cutoff, bounded context, and inconsistent multi-step reasoning.
LLM Hallucinations
Why models fabricate confident, wrong answers — and how to detect and reduce it.
Tokens
8 of 8 topics publishedTokens
The basic unit an LLM reads and writes, and why it matters practically.
Tokenization
How Byte Pair Encoding builds a vocabulary and splits text into tokens.
Token Count
Why you can't estimate tokens by word count, and how to count them precisely.
Context Window vs Token Limit
Two related but distinct constraints — model capacity vs an API output cap.
Input Tokens
Why input tokens often dominate cost in RAG and context-heavy applications.
Output Tokens
Why output tokens cost more and directly drive generation latency.
Token Cost
How LLM API pricing works, and the real cost levers available to you.
Token Optimization
Practical techniques to reduce token usage without hurting output quality.
Transformers
12 of 12 topics publishedTransformers – Complete Overview
The 2017 breakthrough that made today's LLMs computationally feasible.
Transformer Architecture
Inside one transformer block — attention, feed-forward, residuals and normalization.
Attention (in AI)
The core mechanism that lets a model weigh relevance between tokens.
Self-Attention
Attention applied within one sequence, and causal masking for generation.
Multi-Head Attention
Running several attention computations in parallel to capture different relationships.
Query, Key, Value in Attention
The actual mathematical mechanism behind attention, with the scaled dot-product formula.
Positional Encoding
Why attention needs word-order information injected back in.
Embeddings in Transformers
How tokens become vectors before entering the attention layers.
Encoder-Decoder Architecture
The original transformer design, cross-attention, and where it's still used.
Decoder-Only Transformers
The simpler, unified architecture behind most modern general-purpose LLMs.
Transformer vs RNN
Sequential vs parallel processing, and why long-range dependencies favor transformers.
Transformer vs LSTM
Why LSTM's gating improvements still didn't solve the sequential bottleneck.
Prompt Engineering
19 of 19 topics publishedPrompt Engineering – Complete Guide
What prompt engineering actually is, and the techniques covered in this section.
Prompt Structure
Separating instructions, context and task so the model parses intent clearly.
System Prompt
Setting persistent behavior and persona, and why it's not a security guarantee.
User Prompt
The per-turn request, and how applications construct it dynamically.
Context in Prompt
Supplying grounding information, and instructing the model to actually use it.
Few-Shot Prompting
Using multiple examples to lock in format and demonstrate variation.
Zero-Shot Prompting
Asking for a task with no examples, and when it works well.
One-Shot Prompting
Using exactly one example to establish a consistent output pattern.
Chain-of-Thought Prompting
Working through steps before a final answer to improve multi-step accuracy.
Reasoning Prompts
Self-consistency, sub-question decomposition, and verification patterns.
Role Prompting
Assigning a persona to shape tone and framing — and its real limits.
Task Decomposition (Prompting)
Breaking one large prompt into a sequence of smaller, testable ones.
Prompt Templates
Reusable, testable prompt structures instead of ad hoc prompt strings.
Prompt Chaining
Connecting multiple LLM calls in sequence, with real error-handling considerations.
Prompt Optimization
A systematic, testable process for improving prompt performance.
Prompt Evaluation
Building a test set and measuring prompt quality before shipping changes.
Prompt Injection
Direct vs indirect injection, and practical defense-in-depth strategies.
Prompt Security
The broader threat surface beyond injection — leakage and unsafe output use.
Prompt Engineering Best Practices
A practical checklist across structure, technique, evaluation, security and cost.
LLM APIs
13 of 13 topics publishedLLM API – Overview
The common request/response shape across most LLM API providers.
Chat Completions
The role-based message structure behind modern conversational LLM APIs.
Model Selection
A practical framework for choosing which model to call, and model routing.
Temperature
How temperature controls output randomness, with a worked example.
Top-p (Nucleus Sampling)
How top-p restricts the candidate token pool, and how it differs from temperature.
Max Output Tokens
Capping response length, and why checking the finish reason matters.
Streaming
Delivering output incrementally, and what it does (and doesn't) improve.
Batch Processing
Lower-cost, non-real-time processing for bulk LLM workloads.
Rate Limits
Designing around API rate limits proactively, not just reacting to errors.
Retries
Exponential backoff, jitter, and which errors are actually worth retrying.
API Error Handling
Handling malformed output, content rejections and empty responses gracefully.
LLM API Cost
The full cost picture beyond simple token pricing, and practical controls.
LLM API Security
Credential handling, data privacy, and access control for LLM integrations.
Structured Output
7 of 7 topics publishedStructured Output – Overview
Why predictable, parseable output matters for production LLM apps.
JSON Output
Practical prompting techniques for reliable JSON, with defensive parsing.
JSON Schema for LLM Output
Formally specifying expected structure, and using it to guide generation.
Structured Generation
Provider-enforced schema conformance at the generation level.
Schema Validation
Validating LLM output as untrusted input, with real handling strategies.
Structured Output vs Free Text
When rigid structure helps, and when it hurts a genuinely conversational response.
Function Calling vs Structured Output
Two related but distinct patterns — a final answer vs a request to act.
Embeddings
9 of 9 topics publishedEmbeddings – Complete Overview
The text-to-vector-to-similarity pipeline that powers semantic search and RAG.
Text Embeddings
Generating embeddings in practice, and what they're used for.
Embedding Models
How embedding models differ from LLMs and from each other.
Vector Representations
How meaning is encoded as position in a high-dimensional space.
Semantic Similarity
Measuring meaning-based closeness, and where it can still fall short.
Cosine Similarity
The standard formula for comparing embedding vectors, with a worked example.
Embedding Dimensions
What vector size means practically, and the storage/quality tradeoff.
Embedding Search
The full semantic search flow, from setup to query-time retrieval.
Embedding Best Practices
Practical guidance covering consistency, chunking, scale and evaluation.
Vector Databases
9 of 9 topics publishedVector Database – Overview
What vector databases do, and a conceptual landscape of tools.
Vector Search
The database-level similarity search operation, exact vs approximate.
Nearest Neighbor Search
The underlying computational problem vector databases solve.
Approximate Nearest Neighbor (ANN)
Trading a small accuracy margin for dramatically faster search at scale.
Vector Index
The pre-built structure that makes fast similarity search possible.
Vector Database vs SQL
Different query strengths, and why they're usually used together.
Vector Database vs Document Database
Overlapping but distinct strengths, and a practical decision angle.
Metadata Filtering
Combining exact-match filters with similarity search — critical for security.
Hybrid Search
Combining semantic and keyword search for the best of both.
RAG
27 of 27 topics publishedRAG (Retrieval-Augmented Generation) – Overview
The full RAG pipeline, and why grounding answers beats relying on memorized knowledge.
RAG Architecture
The offline ingestion vs online query split, and every component involved.
RAG Pipeline (with Code)
A concrete, working ingestion and query pipeline implementation.
Document Chunking
Why documents are split before embedding, and the core size tradeoff.
Chunking Strategies
Fixed-size, semantic, and structure-aware chunking compared.
Chunk Size
The tradeoff behind this key parameter, and how to choose it empirically.
Chunk Overlap
Preventing information loss at chunk boundaries.
Vector Retrieval
The RAG-specific retrieval step, and choosing top_k.
Hybrid RAG
Combining semantic and keyword retrieval within a RAG pipeline.
Reranking
A second, more precise relevance pass over initial retrieval candidates.
Context Retrieval (Context Assembly)
Ordering, formatting and fitting retrieved chunks before generation.
RAG Prompt
The prompt template that drives grounded, citable, hallucination-resistant answers.
RAG Evaluation
Measuring retrieval and generation quality as two separate stages.
RAG Hallucinations
Why RAG reduces but doesn't eliminate hallucination, with real failure patterns.
RAG Failure Modes
A diagnostic table of common RAG failures and where to look for each.
RAG Security
Access control on retrieved content, and indirect injection via documents.
RAG vs Fine-Tuning
Knowledge injection vs behavior change — a practical decision framework.
RAG vs Long Context
Why retrieval still matters even as context windows grow larger.
Advanced RAG
Query transformation, multi-query retrieval, and self-correction loops.
Document Processing for RAG
Extracting clean, usable text — the critical first step of ingestion.
PDF for RAG
Why PDFs are deceptively hard to extract cleanly, and how to approach it.
Document Parsing
Plain text vs structure-aware parsing, and why structure matters.
OCR for RAG
Extracting text from scanned documents, and the errors OCR introduces.
HTML for RAG
Separating real content from navigation and boilerplate on web pages.
Markdown for RAG
Why Markdown's explicit structure makes it the easiest source format.
Document Cleaning
Removing artifacts and boilerplate before chunking and embedding.
Metadata in RAG
What to capture at ingestion time, and why it can't be added retroactively.
Fine-tuning
9 of 9 topics publishedFine-Tuning – Overview
What fine-tuning is good and not good at, and where it fits.
Fine-Tuning vs Prompting
A practical escalation path, and when prompting has genuinely plateaued.
Supervised Fine-Tuning
The standard example-based fine-tuning approach, with real training pairs.
LoRA
Training small adapter matrices instead of the full model's weights.
QLoRA
Combining LoRA with a quantized base model for even lower memory use.
Parameter-Efficient Fine-Tuning (PEFT)
The broader category of techniques LoRA and QLoRA belong to.
Fine-Tuning Dataset
Why dataset quality matters more than volume, and how to build one.
Fine-Tuning Evaluation
Comparing against the baseline, and watching for overfitting.
Fine-Tuning Best Practices
A project checklist covering validation, dataset, technique and evaluation.
Multimodal AI
10 of 10 topics publishedMultimodal AI – Overview
The difference between a multimodal model and a multimodal system.
Vision-Language Models
Models that reason about images and text together, with real limitations.
Image Understanding
Practical image tasks, prompting discipline, and cost considerations.
Document Vision
Reading forms and invoices — OCR-then-LLM vs direct vision models.
Audio AI – Overview
The STT-LLM-TTS voice pipeline, and where latency and errors compound.
Speech-to-Text (STT)
Transcription accuracy factors, and streaming vs batch processing.
Text-to-Speech (TTS)
Synthesizing natural speech, and preparing text for how it will sound.
Image Generation
Creating images from text descriptions, and real current limitations.
Video Generation
Why video generation is harder than image generation, with realistic expectations.
Multimodal RAG
Retrieving and reasoning over images, not just text.
Tool Calling
9 of 9 topics publishedTool Calling – Complete Guide
The full request-validate-execute-respond flow, end to end.
Function Calling
How this term relates to "tool calling" — largely interchangeable.
How Tool Calling Works
A technical, step-by-step look at the full mechanism with code.
Tool Schema
Describing a tool so the model can use it correctly and reliably.
Tool Selection
How the model decides which tool to call, and what drives accuracy.
Tool Result
Formatting execution results for the model, including error handling.
Tool Calling Security
Validation, permissions and least privilege for every tool call.
Tool Calling vs API
How tool calling adds a natural-language decision layer over existing APIs.
Tool Calling vs Agent
A single tool call vs an adaptive, multi-step agentic loop.
Ready to go from notes to a real career?
Join Coding Now Tech Institute's Generative AI course — live mentorship, hands-on projects, and 100% placement support in Delhi NCR.
Enroll Now — Free Demo AvailableGenerative AI Notes – FAQs
What students search before reading Generative AI notes.