Netron

Netron, built by independent developer Lutz Roeder, is a free, open-source viewer for neural network, deep learning, and machine learning models — letting you open a model file and visually explore its computational graph rather than reading raw weights or config files. It renders graph structures showing operations and their connections, includes an "Open Model…" flow for loading files directly, and supports a wide range of model formats used across the ML ecosystem. Being open source with an active GitHub repository (rather than a closed, hosted tool) means you can audit or extend it yourself. Completely free. Good fit for an ML engineer who wants to visually inspect a model's architecture — debugging a conversion, understanding a pretrained model, or documenting a network structure — without writing custom visualization code.