• What are LLMs, applications
• Difference and evolution from traditional NLPs, RNN, LSTMs, and Transformers
  Transformer Explainer
• Prompt Engineering Techniques (Advantages of a well-written prompt)
• Prompt Demonstration:
  Google AI Studio
• Tokenization:
  OpenAI Tokenizer
• Types of Tokenization (WordPiece, Byte-Pair Encoding, SentencePiece) and Demonstration:
  Tokenization Demo

Preprocess Text

• Preprocess resume text
• Tokenize and create embeddings

• Create an account and access token
• Understand the utilities in the platform

HF API call

• Access HF API to get feedback on different resumes
• Add custom instructions to see the performance

• Why Fine-Tune?
• Process of Fine-Tuning
• LoRA (Low-Rank Adaptation)

• Fine-Tune a T5 Model for Resume Analysis
• Push onto Hugging Face Hub

• Create a Simple UI using Streamlit
• Integrate Fine-Tuned Model to Generate Responses
• Deploy App

• Discuss questions from the previous day
• Conduct a quiz session

• What is RAG and its advantages?
• Comparison: RAG vs. Fine-Tuning
• RAG Architecture
• What are Vector Databases?
• Different Types of Searches
• FAISS and how it works

• RAG Implementation
• Integrate RAG and FAISS with the Fine-Tuned Model

• Assignment:

  • Develop an app to summarize lengthy documents
  • Take a document as input and choose the summary format (bullet points, short paragraphs)
  • Fine-tune a suitable LLM and integrate with a simple UI

• Open Discussion

• Interview Question Bank

Deck