Web Dashboard¶

Overview¶

FRAMEWORM includes a web-based dashboard for experiment tracking, model management, and training monitoring.

Launch Dashboard¶

frameworm dashboard --port 8080

Then open http://localhost:8080 in your browser.

Features¶

1. Dashboard Home¶

Overview of all experiments
System resource usage
Recent activity

2. Experiments¶

List and filter experiments
View experiment details
Compare multiple experiments
Visualize metrics

3. Models¶

Model registry
Export models (TorchScript, ONNX)
Deploy models
Model metadata

4. Training Monitor¶

Start/stop training
Real-time metrics
Training logs
Progress tracking

API¶

The dashboard runs on FastAPI backend.

API documentation: http://localhost:8080/docs

Key Endpoints¶

GET /api/experiments # List experiments GET /api/experiments/{id} # Get experiment POST /api/experiments # Create experiment GET /api/models # List models POST /api/models/export # Export model GET /api/system/status # System status WS /api/training/stream # Real-time training

Development¶

Frontend Development¶

cd frameworm/ui/frontend
npm install
npm start  # Development server on port 3000

Backend Development¶

python -m frameworm.ui.api

Build Production Bundle¶

./frameworm/ui/build_frontend.sh

Configuration¶

Set environment variables:

export FRAMEWORM_EXPERIMENT_DIR=/path/to/experiments
export FRAMEWORM_CHECKPOINT_DIR=/path/to/checkpoints

Screenshots¶

(Add screenshots of dashboard pages)

Troubleshooting¶

Port Already in Use¶

frameworm dashboard --port 8081

API Connection Issues¶

Check that backend is running:

curl http://localhost:8080/api/system/status