מריץ LLM בדפדפן

An LLM (Large Language Model) predicts the next token in order to generate text. This page runs DistilGPT-2 through Transformers.js and WebAssembly on your device: the model downloads once, then generation works offline with no API key and nothing sent to a server.