One file instead of five
Project settings, patterns, weights, training sessions, and rollback history live together in one Modern project file.
SignalWeave Modern
Train, test, inspect, and analyze neural networks from one project file.
SignalWeave Modern is the main SignalWeave experience: a cross-platform .NET 8 desktop and WebAssembly app for feed-forward and simple recurrent network work. It keeps network settings, patterns, weights, rollback checkpoints, and analysis in one workflow.
Searching for BasicProp? SignalWeave is the maintained cross-platform alternative. Modern is the recommended workflow, while Classic remains available as a smaller reference path for users who want the BasicProp-style surface.
Single project workflow
Desktop and browser builds
Weights, tests, clustering, time series, and surface analysis
Built for inspection
Move from training to tests, weights, hidden activations, clustering, time series, and surface plots without leaving the app.
Runs anywhere
Use the desktop build on Windows, Linux, or macOS, or open the same Modern workflow directly in the browser.
Use Cases
What people use SignalWeave Modern for
Modern is best when you want a visual, inspectable neural-network workflow rather than a bare library or notebook-only loop.
Teaching and demonstrations
Show how topology, training steps, weights, and outputs change without forcing students into a code-first workflow.
Small dataset experiments
Load a project, train in blocks, test patterns individually, roll back, and compare sessions without rebuilding your setup.
Pattern debugging
Inspect targets, outputs, deltas, hidden activations, and per-pattern pass/fail state to find exactly where a model is going wrong.
Weight and structure analysis
Inspect weight matrices, cluster outputs or hidden states, and understand how the model organizes your patterns internally.
Sequence and SRN behavior
Use time-series and hidden-state tools to inspect how recurrent behavior changes across ordered patterns and resets.
Browser-based sharing
Share a live app URL for quick demos and evaluations without asking users to install the desktop build first.
How To Start
Use the sample project in under five minutes
The fastest way to understand the app is to open the sample and walk one training and test cycle.
Open Modern
Launch the web app or download the desktop build for your OS.
Load the sample
Open seventeen-patterns-7x7-modern.swproj.json to load the saved network, patterns, and weights.
Train one block
Go to Training and click Train #1. The error graph and session history update as the run completes.
Inspect the result
Use Tests, Weights, and Analysis to inspect outputs, hidden activations, and learned structure.
Documentation
Start with the right page for your goal
SignalWeave now has public documentation pages on this domain so guides, manuals, and migration information can be indexed directly by search engines.
Quick Start
Use the sample project and walk one full training and testing cycle in minutes.
User Manual
Learn what the main sections and settings do and how the Modern workflow fits together.
BasicProp Alternative
Understand how SignalWeave maps to retired BasicProp use cases and why Modern is now the main line.
Screenshots
Modern workflow at a glance
The page focuses on Modern because that is the primary product line. Classic is still available, but it is a side workflow rather than the recommended starting point.
Downloads
Get Modern first. Use Classic only if you specifically need the BasicProp-style workflow.
Recommended
SignalWeave Modern
The project-file-first workflow for day-to-day use. Use this if you want the app that the landing page, sample guide, browser build, and current documentation are centered around.
FAQ
Common questions
What is SignalWeave Modern for?
It is a visual neural-network trainer and analysis workbench for feed-forward and simple recurrent networks. It is intended for people who want a usable application, not just a code library.
What makes Modern different from Classic?
Modern is organized around one project file and integrated tabs for training, tests, weights, patterns, and analysis. Classic preserves the BasicProp-style workflow and is useful mainly if you want that exact reference surface.
What is stored in the project file?
Modern stores the network definition, embedded patterns, current weights, training sessions, rollback checkpoints, and related workspace state together, so you do not need separate load/save actions for each item.
Can I use the app without installing anything?
Yes. The Modern workflow is available directly in the browser as a WebAssembly app at /app/. Desktop builds are also available for Windows, Linux, and macOS.
How do I start with a real example?
Download the sample project, open it in Modern, run Train #1, then use Tests, Weights, and Analysis to inspect what changed.
Is SignalWeave open source?
Yes. The full project, release history, and source are available on GitHub.