You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
An experiment-driven laboratory for understanding LLM sampling strategies. Explore how Temperature, Top-k, and Top-p influence token probabilities and generation diversity through reproducible simulations, visualizations, and real Groq API experiments.
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.
KANTMICROGAMES: tiny pixel games, one HTML5 index file, no advertisement, no user tracking, no accounts, and full privacy. This is an Ai/LLM assisted experiment and project in which various models from various providers and various tools are used to build a real project.
Pack enormous amounts of text, code, and datasets into indexed RGB image/PDF archives that AI tools can search and read through sandbox code. Effectively enabling complete backup or analysis of enormous datasets, diff entire filesystems by chaining with visual analysis, inject your own script and have the LLM rebuild it - your own virtual machine.
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
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.
Code for Randomized PipeDream experiments comparing PipeDream-style pipeline training with randomized stale block-gradient methods on simple objectives and small LLM benchmarks.