Training on AI for Generative AI for developers

Generative AI for developers

Generative AI for Developers

Pre requisite:  Programming skills any language or understanding of Programming.  Having Python programming is big plus.

Course Outline

  1. Brief overview of Generative AI
    1. Open AI Eco system overview
    2. Industry applications and recent developments.
  2. Evolution of Generative AI – “Attention all you Need” and other relevant developments
  3. Prompt engineering and various prompt.
    1. Writing the relevant and Effective prompts using different techniques.
  4. Introduction to LLM for Text applications
  5. Introduction to LLM for Image applications

Understand the working of multimodal models like Stable Diffusion: Denoising, Delusion, Autoencoders, Contrastive Learning, Shared Embedding Spaces

Apply image prompting techniques on Dall-E and Midjourney to generate desired

product images using various stable diffusion methods and prompt parameters such as style, ratios, seeds, FPS.

Read, load and embed large datasets and tables to read your data with GPT/Copilot

Understand and apply the fundamentals of style, design and photography to improve image quality and accuracy with prompt iteration and few-shot prompting

Apply self-consistency, seeding and standardised formatting in prompting to create

consistent styles and designs across hundreds of product images

Generate product descriptions along with images using various instructor-tuned models and APIs

  1. Using ChatGPT 3.5/4.0 using API
  2. Understand the working of LLMs like GPT3 that power ChatGPT: Attention Mechanisms, Transformers, Reinforcement Learning, RLHF among others
  3. RAG (Retrieval Augmentation Generation)

 

  1. Integrate LLM chat models over the searched embeddings to respond to the customer
  2. Experiment with different vector stores, search and index algorithms and LLMs to improve the chatbot
  1. Langchain
  2. LlamaIndex
  3. Vector store/Vector Databases
    1. Understand the working of embeddings and how they help in semantic search
    2. A customer facing chatbot that answers questions by scanning organisation’s custom data
    3. Create and analyse embeddings for semantic search
    4. Create embeddings for large documents by creating chunks
    5. Create a Q/A system that fetches answer using similarity search over embeddings
    6. Scale the Q/A system by making use of vector stores like Pinecone/Chroma
    7. Embed, index large documents and search in Vector store

 

  1. Brief overview of AWS Bedrock and Gen AI eco system or Google Vertex AI/Gemini
  2. Scaling of LLM application
  3. Design of the LLM applications.

 

Assignments:

  1. Exploring the 5 Gen AI applications build by Industry/Corporates?
  2. Prompt design for 5 NLP/Text tasks
  3. Prompt Design for 5 Images tasks
  4. Simple Gen AI application using ChatGPT API.
  5. Building Helpmate or Product Search Application using Open AI
  6. Exercise using Text Embeddings using Vector Stores
  7. Exercise using Text loading/Embedding and search on Vector databases Chrome DB/Pinecone or equivalent.
  8. Building Application using RAG (building Context Aware Application)
  9. Scaling the LLM application
  10. Creating Generative Application using AWS Bedrock/Google Vertex AI.

Project:

The Capstone project similar to the one of the following–

  1. Building Chatbot
  2. Conversation Application
  3. Product search Engine/Recommender
  4. Building Enterprise level Google Like Search
  5. Building Tutors Schools/Colleges

 

 

Categories: ,