BNNR

Dashboard Guide (Live + Replay + Mobile)

What you will find here

How to use dashboard as a daily workflow tool:

  • live monitoring during training,
  • replay after training,
  • mobile access via QR,
  • secure controls,
  • export for offline sharing.

1) Live mode during training

Start training with dashboard:

python3 -m bnnr train \
  --config examples/configs/classification/cifar10_example.yaml \
  --dataset cifar10 \
  --max-train-samples 128 \
  --max-val-samples 64 \
  --preset light \
  --with-dashboard \
  --dashboard-port 8080

Terminal prints:

  • Local URL
  • Network URL
  • QR code

Open Local URL on desktop first. Then use QR for phone.

2) Replay mode (no training)

Replay events from existing run(s):

python3 -m bnnr dashboard serve --run-dir reports --port 8080

Use replay when:

  • you want to inspect old runs,
  • you do post-mortem/debug,
  • you demo results without retraining.

3) Mobile access with QR code

Requirements:

  1. Phone and machine on same network.
  2. Dashboard server running.
  3. Firewall allows chosen port (default 8080).

Steps:

  1. Start dashboard (train --with-dashboard or dashboard serve).
  2. Scan terminal QR code.
  3. Open Network URL on phone.

If QR opens but page does not load, test Network URL manually in phone browser.

4) Secure pause/resume controls

Protect control endpoints with token:

python3 -m bnnr dashboard serve --run-dir reports --port 8080 --token "change-me"

Or set env var:

export BNNR_DASHBOARD_TOKEN="change-me"
python3 -m bnnr dashboard serve --run-dir reports --port 8080

For train path use:

python3 -m bnnr train ... --dashboard-token "change-me"

5) Pause/resume workflow

  1. Start long run with dashboard.
  2. Use Pause control in UI (or control endpoint if you automate).
  3. Verify status changes to paused.
  4. Resume and confirm progress continues.

6) Export static report (shareable)

python3 -m bnnr dashboard export \
  --run-dir reports/run_YYYYMMDD_HHMMSS \
  --out exported_dashboard

Artifacts include:

  • index.html
  • data/events.jsonl
  • data/state.json
  • copied run artifacts/

Open exported_dashboard/index.html locally without backend.

  1. Run training with dashboard and token.
  2. Watch live branch decisions and KPI trends.
  3. Stop server after validation (Ctrl+C).
  4. Export static snapshot for review/share.
  5. Archive report.json, events.jsonl, export dir.

8) Common issues

See Troubleshooting for:

  • dashboard dependencies missing,
  • zero runs visible,
  • mobile/QR/network access issues,
  • export rendering issues.