NuBoard

nuboard is a fast, local, folder-based experiment tracker and viewer for machine learning training. It logs your training runs with three Python calls, keeps every run as a plain folder on disk, and opens thousands of them in a fast viewer that runs on your own machine. There is no account to create, no server to keep running and no database to back up.

With nuboard you can:

• Log with three calls (init, add_scalar, finish). Config, git commit and diff, installed packages, GPUs, command line and rank are recorded for you.
• Keep one folder per run that you can copy, rsync, zip or delete, with stars, notes and tags stored next to the data.
• Group a sweep into mean ± std bands over seeds, smooth, zoom and follow runs live while they train.
• Log a range, not just a number, and see where the time goes with per-step timers.
• Inspect audio with its annotations: overlay a voice activity probability and label tracks on the waveform.
• Catch NaN and Inf values, loss and gradient spikes and throughput drops with health alerts.
• Track versioned datasets and checkpoints with a lineage graph.