Portfolio
November 2024 (May take up to 50 seconds to spin up instance)
October 2024 (May take up to 50 seconds to spin up instance)
Explore the latent space of LLMs by crafting your own completions.
This site is meant to help LLM novices learn more about how LLMs generate completions, by allowing them to interact with tokens direct from the model.
Key Technologies:
Full-stack web application built with React and Node.js
React D3 Tree powers the tree component, with custom node rendering and add new node functionality
OpenAI's API generates additional tokens as users interact with the tree
Interested in learning more?
Can you tell a real human from a chatbot? Try your hand here.
This interactive web application lets users engage in a chat where the challenge is to guess whether they’re chatting with a human or an LLM bot.
Key Technologies:
Full-stack web application built in React and Node.js
Socket.io to facilitates real-time, bidirectional communication
OpenAI's API generates dynamic chatbot responses based on user input
User authentication and data storage are powered by Supabase to enable persistent user sessions and to allow users to access their chat history