Zero to Hero in NLP, LLM & GenAI

Zero to Hero in NLP, LLM & Generative AI

Greetings, everyone!

We've all witnessed the remarkable surge of NLP-based applications in recent times, particularly following the introduction of ChatGPT. The global enthusiasm surrounding the realms of AI and ML, with a special focus on Natural Language Processing, is Tangible. What's truly fascinatingabout NLP is its ability to closely emulatehuman understanding of language, making it a standout category within AI and ML.

My goal is to simplify and explain the concepts of NLP,LLM & GenAI for all of you, bridging the gap between users and the fascinating machinery that powers these technologies.

1 : Understanding the Complexities of Human Language for NLP,LLM & GenAI

WHY Human Language is difficult for Machine or Natural Language Processing?

2 : Text Data Pre-processing Pipeline

Text-Data Pre-processing pipeline for Tokenization | How text data gets converted into TOKENS

3 : Understanding Embedding

How Language Model Learns the Complexity of Human Language through Embedding

4 : How Embedding Vector gets generated

How Embedding Vector is generated | Pre-training Embedding Models | Learning Word2Vec and Skip-gram

5 : Exploaring pre-trained embedding models

What are available pre-trained Embedding models | Open Source Embedding models | Embedding model repository

6 : Exploaring Embedding use cases and Implementing Question Answering System

Embedding is the backbone of almost all NLP task like Semantic Search, Clustering, Recommendations, Anomaly detection, Question Answering, Classification and many more. In this video we are going to implement Question Answering System for PDF document using:

⭐ LangChain, an open-source library for loading , chunking and semantic search

⭐ Sentence Transformer for open-source embedding model

⭐ Chroma Db - to store embedding vector, again an open source.

⭐ OpenAI's gpt3.5 Turbo generative model to generate the final answer.