GEN AI: From Basics to Advanced Level

Complete a seven-question assignment that recaps using large language models via APIs (cohere or openai) to build, train-test split data, and validate a regression-based model.

Join this assignment-driven lecture to use a language model to generate a new business idea for Bengaluru software professionals, and follow guided code setup, prompts, and practice questions.

Explore artificial neural networks, known as anns, the heart of llms today, and learn how weights replace betas, hidden layers transform data, and create non-linear decision boundaries for accurate classification.

Explore word to vector embeddings and how to convert words into vectors with word2vec. Learn preprocessing, vector size, context window, and embedding layers for text classification with TensorFlow.

Practice neural networks and word2vec for news article classification by completing a six-step assignment that configures, compiles, trains, and evaluates a document classification model to predict sports or politics.

present a solution approach for assignment 2 in the Gen AI course, detailing data prep, tokenization, and building a four-category news headline classifier with word embeddings.

Learn name entity recognition with hugging face pipelines, explore aggregation strategies and licenses, and fine-tune pre-trained transformers for domain-specific tasks.

Fine-tune a distilbert classifier on a five-class emotion dataset to boost accuracy from 25% to 88% using hugging face transformers, tokenization, and a 10,000-record sample.

Explore how LangChain connects large language models to private data using prompt templates and chains, enabling rapid AI app development with OpenAI, Cohere, or Hugging Face.

Build a three-step LangChain workflow: sentiment analysis chain, follow-up question chain, and a sequential chain, using prompts and a large language model via OpenAI or Cohere.

Build a two-chain LangChain solution to analyze sentiment and generate a follow-up question, using OpenAI or Cohere and a sequential chain, as demonstrated in assignment four.

Create a Wikipedia company profile app that inputs a company name, loads data with a Wikipedia document loader, and outputs founder, start date, revenue, employees, and a four-line summary.

Format the Wikipedia company profile for assignment five into a JSON output using output parsers, and extract founder, founding year, revenue, and number of employees with a pedantic output parser.

Explore memory components in LangChain to build context-aware chatbots using conversational buffer memory, conversational buffer window memory, and conversation entity memory, with rag and OpenAI integration.

LangChain agents and toolkits show how to fuse large language models with external tools, enabling reasoning and acting to address math and current-events limits.