Module 1: Introduction to NLP
Introduction to NLP: covers what NLP is, its history, applications and challenges. NLP Techniques: covers common techniques such as tokenization, part-of-speech tagging, named entity recognition and sentiment analysis with examples of their use. NLP Tools: introduces popular NLP tools such as NLTK, spaCy with examples of how to use them for basic NLP tasks.
Module 2: Foundational Knowledge of Transformers & LLM System Design
Transformers Foundational Knowledge: covers fundamental concepts of transformers, including self-attention, multi-head attention, and positional encoding. Introduction to Fundamental Concepts of ML System Design: covers the basics of designing machine learning systems, including data collection and preprocessing, model selection and training, and performance evaluation.
Module 3: Semantic Search
In this module, we will learn about retrieval systems and their significance in information retrieval. Discover popular methods such as Sparse vs. Dense Vectors, Euclidean Distance, Cosine Similarity, Approximate Nearest Neighbors (ANN), and practical coding using FAISS to achieve fast and precise search results.
Module 4: Creating a search engine from scratch
Building a Semantic Search Model: covers the basics of semantic search models, their architecture, and how they work. The session will focus on building a semantic search model for hotels using various natural language processing techniques. Deployment on a Serverless Inference: discusses the benefits of serverless computing and how to deploy the semantic search model on a serverless platform like Huggingface. Preprocessing of Hotel Data: covers the preprocessing of the hotel data using techniques like text cleaning, tokenization, stemming, and lemmatization, and how to convert the hotel data into embeddings that can be used by the semantic search model. Evaluation of the Model: discusses the different evaluation metrics used for semantic search models and how to measure the performance of the model using these metrics. This session will also cover techniques for improving the model's performance and optimizing the search speed. Discussion on Query Intent Models.
Module 5: The Generation Part of LLMs
In this module, we'll explore the fundamentals of RAG and their real-world applications, as well as dive into Langchain's concept of chunking and agents, seamlessly connecting retrievals to Gen AI.
Module 6: Prompt-tuning, fine-tuning and local LLMS
In this module, we will learn how to effectively engineer prompts, fine-tune language models, leverage the PEFT approach, and measure the success of their efforts using appropriate validation metrics.
