LLM Tutorials and Learning Projects

This repository contains tutorials and learning projects to experiment with concepts in Large Language Models (LLMs). All projects are implemented in Jupyter Notebooks.

---

Tutorials and Projects

1. Introduction to LLMs

Notebook: introduction_to_llms.ipynb

• Objective: Learn how to use the ChatGPT API through practice tasks.
• Tasks:
  • Simple Translation: A basic translation task.
  • Controlling Outputs from Few-Shot Learning: Experiment with guiding LLM outputs using few-shot examples.

---

2. Basic RAG Pipeline

Notebook: basic_rag_pipeline.ipynb

• Objective: Build a simple Retrieval-Augmented Generation (RAG) pipeline.
• Steps:
  • Downloaded an article and performed text chunking.
  • Converted text into embeddings using OpenAI.
  • Retrieved embeddings using cosine similarity stored in a DataFrame.
  • Performed inference using:
    • Google GenAI (gemini-1.5-flash)
    • OpenAI (gpt-3.5-turbo)

---

3. LangChain Application

Notebook: langchain_application.ipynb

• Objective: Explore simple LangChain applications.
• Tasks:
  • "Hello World" level inferencing using HumanMessage Prompt templating with OpenAI (gpt-3.5-turbo).
  • Summarization of a PDF document using PyPDFLoader.

---

4. News Article Summarizer

Notebook: news_article_summarizer_langchain.ipynb

• Objective: Summarize news articles using LangChain.
• Steps:
  • Scraped news articles.
  • Used HumanMessage prompt templating from LangChain.
  • Generated bullet-point summaries with proper prompts.

5. Llama Index Overview

Notebook: llama_index_overview.ipynb

• Objective: Understand and utilize LlamaIndex.
• Steps:
  • Used download_loader for WikipediaReader.
  • Loaded and analyzed data with NLP and AI tools.

---

6. Customer Support QA Chatbot

Notebook: customer_support_qa_chatbot.ipynb

• Objective: Build a chatbot for customer support.
• Steps:
  • Downloaded articles using SeleniumURLLoader.
  • Stored data in a vector storage database using FAISS.
  • Queried the vector datastore and formatted output with PromptTemplate.

7. YouTube Video Summarizer

Notebook: youtube_video_summarizer.ipynb

• Objective: Summarize YouTube videos using Whisper and LangChain.
• Steps:
  • Downloaded and transcribed YouTube videos using Whisper.
  • Summarized content using SummarizationChain.
  • Split content into chunks, converted to embeddings, and stored in a vector database.
  • Prepared prompts and queried the database.
  • Generated summaries using RetrievalQA.

How to Use

1. Clone this repository:

   git clone <repository_url>
   cd <repository_name>

2. Open the notebooks with Jupyter:

   jupyter notebook

---