GPT Next Token Generator, RAG, and Function Calling Demos

This repository includes four demos as follows:

Next Token Generator: This demo illustrates how large language models (LLMs) like GPT-4o predict the next token by adjusting the temperature and top_p parameters. The next token prediction demo is deployed using Streamlit and can be accessed from the link below:
Prompt Engineering: In this demo, the students learn how to use prompt engineering techniques to guid the model in retrieving desired output:
Retrieval Augmented Generation (RAG): This example shows how to run a basic RAG pipeline from scratch. Open the demo in Google Colab from the link below:
Function Calling: This demo provides an example of a basic function calling, which could be accessed in the Google Colab from the link below:

Features

Generate responses using OpenAI's GPT-4o model.
Adjust temperature and top_p parameters to see their impact on token prediction.
Visualize log probabilities of predicted tokens.
Parse and retrieve text chunks from a PDF document to respond to a domain specific query.
Create a custom agent to avoid hallucination and improve the capabilities of LLMs

In order to run the basic_rag_demo the API key to gpt-40 could be created/retrieved form the link below: OpenAI API Key

Note: Once you logged into OpenAI platform, find the API key from the sidebar:

This project is licensed under the MIT License.