llm-experiments

experiment in storing face embeddings in sqlite and searching the filesystem for face matches

Orchestrator-worker multi-agent pattern: one planner, three independent workers with identical prompts but different sampling configs (temperature/top_p). Isolates what sampling alone changes in model behavior — conservative vs balanced vs creative brainstorming.

Pi extension: plan mode, subagent delegation, and Exa web research

This repository documents a seven-round investigation into whether alternative computational geometries can improve upon the standard dot-product attention mechanism used in transformer language models.

This project explores the use of a Large Language Model (LLM) to evaluate image sharpness through a paired comparison experiment. The goal is to establish Just Noticeable Difference (JND) levels for sharpness perception

Experiments in Non-Euclidean Geometry for Language Models

Code for Randomized PipeDream experiments comparing PipeDream-style pipeline training with randomized stale block-gradient methods on simple objectives and small LLM benchmarks.

Can Large Language Models detect hidden deterministic patterns without being told they exist?

A hands-on Generative AI learning environment featuring practical experiments and projects for building AI-powered applications.