FLM — Fly Language Model

FLM is a research-oriented language model trained from scratch using a connectome-derived recurrent network based on fruit-fly neural wiring. It allows users to interact with and observe real recurrent activity, local learning, and model dynamics directly in the browser. FLM supports text generation with controllable parameters and provides detailed inspections of its neural states, making it an exploratory tool for experimental AI research rather than a general-purpose language model.