LangChain Mastery - Most Practical Course To Build AI Apps

Explore fundamental AI concepts such as large language models, machine learning, and artificial intelligence, and learn to build production-ready applications with ChatGPT, GPT-3.5 turbo, LangChain, and OpenAI.

Understand why we need models and how machine learning teaches computers to learn from data, covering features, labeling, training, testing, and prediction with examples like predictive maintenance and fraud detection.

Explore artificial intelligence, including artificial narrow intelligence and artificial general intelligence, and see how machine learning enables generative ai to create original content.

Learn how prompts steer ai apps by defining prompts, practicing prompt engineering, and shaping input and output tokens. Understand token costs and api call calculations for models like ChatGPT.

Explore how ChatGPT works behind the scenes, from pre-training and fine-tuning with prompts and feedback to a transformer with self-attention that encodes input into vectors and generates responses.

Explore supervised learning with labeled input-output pairs to learn mappings for tasks like image classification and spam detection, and unsupervised learning on unlabeled data to uncover patterns and clusters.

Write our first GenAI code Part-1 teaches you to call the OpenAI chat completion API from Python, using a virtual environment and dotenv to securely load API keys.

Learn to use the OpenAI Python client library to interact with models, including installation, key management, and prompts, and compare it with the raw API.

Learn how LangChain, an orchestration framework for developing applications powered by language models, simplifies building workflows. Compare it to raw APIs and libraries, and explore OpenAI UI and API integrations.

Learn to write your first LangChain code to use an OpenAI chat model, sending system and user messages to generate a response and format it as a string.

Discover why LangChain offers modular code management, readability, and scalable chain orchestration for AI apps, with built-in state management, parsers, security, and ecosystem benefits.

Set up your environment, configure api keys for Windows and Mac, and write your first generative ai code while exploring LangChain's benefits through coding and quizzes for building with llms.

Explore the typical LLM application workflow with three blocks: prompt, model, and output, and how prompts feed the model to generate formatted results like JSON.

Explore how chat models and foundation models power AI apps, using LangChain interfaces to connect to OpenAI, Google, and others, and how to craft, invoke, and format chat prompts.

Explore how foundation models differ from chat models by using the LM interface with OpenAI, highlighting string input and output, and why we won't use LM in this course.

Access open source models like llama 2 via hosted APIs such as replicate and together, without local installation, and configure keys, env vars, and llm parameters in VSCode.

Build an interactive financial concept explainer app in Python using a chat model, with continuous user prompts and a quit command, as shown in assignment one of LangChain mastery.

Learn to calculate LLM API call costs by using OpenAI's built-in cost function and custom pricing for proprietary models, including token and character-based pricing across OpenAI and Google.

Explore the LLM application workflow, differences between chat models and LLMs, and token usage; also cover open source models, an assignment on financial concepts, and optional interfaces and classes.

Explore prompts and output parsers in llm-powered apps, learning to craft effective prompts, use prompt templates, and dynamic prompts with f-strings for well-organized outputs.

Learn to build dynamic prompts using chat prompt templates, human message templates, and prompt templates with placeholders and input variables to format system and user messages for task-specific prompts.

Master few-shot prompting in LangChain Mastery with input-output examples and chat prompt templates. Define example lists, system and human messages, then apply these techniques via the chat api.

Build a travel app with few-shot templates, system and user prompts, and chat prompts to guide moving from source to destination.

Explore how context window and token limit shape model responses, and learn to manage inputs by prioritizing recent prompts, system instructions, and relevant examples within token budgets.

Explore four example selector types—similarity, m.m.r, length, and n gram overlap—and how they choose examples to fit the context window via vector embeddings and semantic search.

Explore how to structure language model outputs with LangChain output parsers, using Pydantic models for JSON and validation, and leverage format instructions and partial variables to produce reliable, formatted results.

Master crafting prompts and building prompt templates, employing few-shot examples, managing context window token limits, and using output parsers to refine responses for practical apps and quizzes.

Discover Streamlit, a lightweight open source Python library that lets data scientists build and share interactive machine learning and data science web apps without HTML, CSS, or JavaScript.

Integrate Streamlit into an existing financial concept app to create a dynamic, interactive AI assistant with a model selector, user input, and live UI updates.

Create a financial qa system interface in Streamlit that uses session state to manage a chat messages list with user and assistant roles, chat input, loader, and OpenAI language model.

Refactor a Streamlit financial concept explainer chat app into modular, readable code following the single responsibility principle, separation of concerns, session state management, and environment variables.

Develop a social media script writing app with Streamlit and Landscape that outputs a script title and full script from content type, topic, audience, duration, and format, updating dynamically.

Use Streamlit to rapidly build dynamic user interfaces for ai apps, refactoring projects while applying software engineering principles, and crafting practical tools like financial systems and social media script writers.

Explore LangChain chains to connect chat prompts, models, and parsers into a left-to-right workflow, enabling composability, memory concepts, external data integration, and robust error handling.

Learn runnable as the fundamental unit for LangChain chains, using a pipe operator to pass outputs between steps. Chain runnable objects to form prompts, models, and parsers.

Explore runnable sequence, a flexible workflow for multi-step operations, with runnable, chain, and lambda in real-world examples like text cleaning, transformation, and sentiment analysis.

Master sequential chains by wiring two components in one workflow: generate content from a topic, then generate hashtags from the content, showcasing a two-chain sequence in a practical LangChain workflow.

Explore how a pass-through in a chain keeps data flowing without modification, enabling future changes with function references and modular transformation steps.

Master chains by exploring runnable renewal sequences and sequential chains, including renewal pass through, and apply a hands-on video transcript analyzer app to build structured sequential workflows in AI apps.

Explore why memory matters in ai apps by showing how a memory module in a chain reads and writes past interactions, enabling chatbots to reference prior prompts and responses.

Learn to add memory in ai apps by storing and retrieving conversation history with a history placeholder, chat prompt templates, and an in-memory history class.

Master memory with sessions in LangChain by using an SQLite memory database and session IDs to persist and recall chat history across restarts.

Learn to manage memory with user sessions by appending user IDs to session IDs, enabling retrieval of chat histories across multiple users and sessions.

Develop AI applications that remember past conversations by leveraging session based memory and memory for user sessions, maintaining context across interactions for a more personalized experience.

Understand prompts and prompt engineering, and learn to craft inputs with context, examples, output format, and system instructions to guide AI models toward accurate, ethical, and relevant results.

Discover how to craft good prompts that produce accurate, relevant, factually correct, and logically consistent results. The lecture illustrates these criteria with examples and practical prompts.

Learn prompt engineering techniques to guide AI models toward desired outputs, including zero-shot, one-shot, few-shot, multi-step prompting, and chain of thought methods.

Explore prompt engineering techniques including zero-shot, one-shot, and few-shot prompting, chain-of-thought and self-consistency, prompt chaining, and negative prompting; learn how multiple reasoning paths and majority vote improve accuracy.

Master prompt engineering by learning Jupyter Notebook setup, exploring what makes a good prompt, benchmark techniques, and diverse methods to craft prompts that guide AI responses with precision and clarity.

Explore practical use cases for large language models, including text summarization, tagging and categorization, sentiment analysis, question answering, and information extraction, with real world examples and impact.

Explore the top use cases for llms across content generation, conversational agents, code generation and debugging, structured data analysis, and language translation, with practical application ideas.

Explore real world LLM use cases and over ten practical scenarios where LLMs drive value and innovation in the AI era.