Neural Network vs OxViz

Learning and using neural networks in an intuitive and natural way! Intuitive and interactive tutorials. A large number of charts and interactive components are used to make learning more intuitive. Visualized Neural Network Lab. Open the black box of neural networks with simplified datasets and full visualization, gain more intuition. Learning with mission like gaming. In the missions, you will run into many key concepts in neural networks, complete the challenges and acquire a deep understanding. Put neural network lab in your pocket. Variety of datasets: 2D and 3D datasets both with regression and classification. Get the dataset just with a tapping. Do experiments at anytime anywhere. Good visualized neural network knowledges. Key concepts and dynamics are showed with charts and animations. Activations, regularizations, loss functions, classifications, Embedding, CNN1d, CNN2d, RNN and more to come. Visualized Deep Learning Models. Learn the most classic models in an intuitive way that has never been seen before, and quickly master cutting-edge technology. Through modular diagrams, understand how data changes from input to output and how it changes at each step. There are detailed documents for each module, and even reference papers, source codes and animations. Visual neural network model editor. Quickly build a model by dragging your fingers. What you see is what you get: Through real-time calculation and display of the output of each step, the construction of the model has never been so intuitive and efficient; real-time display of various errors, including parameters/input types and value compatibility, etc., abandoning inefficient debugging. MORE FEATURES COMING!!!

OxViz — Learn ONNX models by exploring them visually OxViz is an interactive ONNX model explorer for iPhone, iPad, and Mac that teaches you how models are built and how data flows through them. Import an .onnx file and learn model structure step‑by‑step: visualize the graph, inspect operators and tensors, trace data paths, and produce clear, shareable documentation — all locally on your device. What you can do: - Visualize model architecture: see layers, connections, and data flow in a clean native interface. - Read model internals: inspect operators, tensor shapes, initializers, attributes, parameter counts, opsets, and metadata to understand what each component does. - Trace behavior interactively: follow upstream and downstream flows from any node to learn how inputs transform into outputs. - Diagnose common issues: run the Model Health Report to identify missing shapes, disconnected nodes, duplicate names, unresolved inputs, and unusual topology so you can learn problem patterns. - Create learning materials: export graphs and editable diagrams (PDF, SVG, PNG, .drawio) and generate a professional Model Health Report PDF for class notes, presentations, or handouts. Designed for learners and practitioners - OxViz is built for ML students, educators, researchers, and developers who want a hands‑on way to learn model design and debugging without sending models to a server. - Use it to study exported onnx models, prepare presentation examples, compare expected vs actual architectures, or document designs for reports and assignments. Privacy-first, local exploration All model analysis runs on your device — no account, no uploads — so you can safely explore proprietary or sensitive models while learning. Features at a glance - Import and inspect ONNX model graphs. - View operators, inputs, outputs, tensor shapes, attributes, initializers, and parameter counts. - Search and navigate large graphs quickly. - Trace flow from any node to reveal model behavior. - Model Health Report that explains common structural problems. - Export editable diagrams and professional report PDFs for teaching or documentation. - Native iPhone and iPad interface, with compatible Mac use through the iPad app. - Automatic graph layout optimized for clarity. (using dagre.js / MIT license) Free to try; one‑time in‑app purchase unlocks unlimited imports, exports, and Pro features.