LLM4BUS

This is a repository of my projects from the Stanford Continuing Ed course "TECH16: LLM's for Business with Python". The course is taught by Charlie (Cathal) Flanagan.

HW 1: Use the OpenAI GPT-3.5 LLM to summarize some text. Also, have the model compare differences between two texts.

I used the OpenAI GPT-3.5 LLM to summarize some research notes created by my 3rd grader for her assignment. I generated two different types of summaries (one concise, and the other verbose), and used the model to compare the resulting summaries.

HW 2: Using an LLM to performa a classification task

I tackled an NLP classification task: classifying whether a news headline was serious or satirical. An interesting element was to use prompt engineering to cajole the model into outputing JSON corresponding to a particular schema.

HW 3: Document summarization using LangChain, and simple RAG using LlamaIndex

I used LangChain to summarize a 10K filing(a Word Document), and investigated the differences between "stuff" vs. "map-reduce" summarization methods. For the second part of the homework, I created a webscaper to scrape fan web pages, and used LlamaIndex to create a Knowledge Base out of the scraped data.