Netron

Netron is a viewer for neural network, deep learning and machine learning models. It opens a model file and draws its graph so developers and researchers can inspect the layers, operators and structure of a model, which is useful when debugging, converting or documenting models. The viewer supports ONNX, TensorFlow Lite, PyTorch, torch.export, ExecuTorch, TorchScript, TensorFlow, Core ML, OpenVINO, Keras, Caffe, Darknet, Safetensors and NumPy files, with experimental support for MLIR, JAX, GGUF, RKNN, ncnn, MNN, PaddlePaddle and scikit learn. It runs in the browser at netron.app through an Open Model button, so a file can be examined without installing anything. Desktop versions are available for macOS, Linux and Windows, and a Python package starts the viewer from the command line on a given file. The tool is for machine learning engineers, students and anyone who needs to understand a model produced by another framework. It is a viewer only and does not train, convert or run models. The site shows no pricing and no plans. The page carries a copyright notice for its author, Lutz Roeder, and the project is distributed through its public repository.