Convert plain English into 3D physics simulations
A prototype that uses LLMs to parse natural language descriptions and generate 3D rigid-body physics simulations. Users describe objects, forces, and initial conditions in plain English, and the system creates and plays a short simulation. It targets educators, students, or hobbyists who want to quickly visualize physics scenarios without coding.
Related
Replay coding-agent sessions on a 3D codebase map
Mindwalk replays coding-agent sessions on a 3D map of your codebase, letting developers visualize and debug agent actions spatially. It uses LLMs to power the coding agent whose sessions are replayed, helping developers understand agent behavior and codebase structure.
Interactive Jacobian-Lens visualizer and live steerer for GGUF models on llama.cpp
A tool that visualizes and steers GGUF models using the Jacobian lens technique, built on llama.cpp. It allows users to observe and modify model internals in real time, solving the lack of such tools for GGUF format models.
3D library of personal book quotes
A desktop-only 3D library where users can walk around and pick up books, each containing a quote saved from a book read since 2017. LLMs were used to port the plain-text quote collection into the 3D environment, enabling a novel, immersive way to revisit personal reading highlights.
Recreation of MU/TH/UR 6000 UI from Alien using Fable and Claude
A fan recreation of the fictional MU/TH/UR 6000 interface from the Alien franchise, built using Fable (a generative UI tool) and Claude. The creator, a React developer, used natural language prompts to generate the UI with minimal coding, demonstrating how LLMs can assist in rapid prototyping of complex animated interfaces.
GGUFun: play Snake and maze on Ollama with hand-crafted GGUF models
GGUFun is a project that creates hand-crafted GGUF model files to run deterministic games like Snake and a simple maze on Ollama. The models are built manually without training, using custom weights to compute game logic and respond with fixed outputs. It demonstrates that LLM inference engines can be repurposed for non-language tasks.