Train local. See everything.

Create runs

Configure training runs with epoch presets, GPU/CPU selection, batch size, learning rate, and optimizer.

Live monitoring

Real-time loss charts, live log streaming, progress tracking, and the ability to cancel at any time.

Hyperparameter sweeps

Run multiple experiments with grid search across learning rates, batch sizes, and optimizers.

Run inspection

Browse completed runs with full metrics, logs, and artifact inspection. Filter by status.

Local-first

Everything runs on your machine. All artifacts saved to disk for inspection and reproducibility.

Reliability gauntlets

Repeatable test suite for queueing, pause/resume, cancellation, crash recovery, and fair scheduling.

From Release (recommended)

# 1. Download latest .msix from GitHub Releases
# 2. Double-click to install
# 3. Launch from Start Menu

# Then: select a workspace folder
# and start training

From Source

git clone https://github.com/mcp-tool-shop-org/runforge-desktop
cd runforge-desktop
dotnet run --project src/RunForgeDesktop/RunForgeDesktop.csproj

# Run tests
dotnet test

Install

dotnet add package RunForgeDesktop.Core

What it includes

// Core domain models and services for ML training:
// - Run lifecycle management
// - Hyperparameter sweeps
// - Live monitoring models
// - Artifact inspection types