notiboard 0.1.0

Monitor AI training on your phone with a drop-in SummaryWriter replacement.

notiboard

notiboard is both the PyPI distribution name and the Python import package.

Monitor your AI training on your phone with a drop-in replacement for TensorBoard's SummaryWriter.

Installation

pip install notiboard

Quick Start

from notiboard import NotiWriter

writer = NotiWriter(
    log_dir="runs/experiment1",
    noti_api_key="nk_your_key_here",   # from the Noti app: Settings > API Keys
    noti_project="my-project",
    noti_run_name="experiment-001",    # optional, auto-generated if not set
)

for step in range(1000):
    loss = train_step()
    writer.add_scalar("loss/train", loss, step)
    writer.set_progress(step, 1000, f"Step {step}/1000")

writer.send_notification("Training complete", "Final loss: 0.001", "Done.")
writer.close()

All standard TensorBoard SummaryWriter methods are preserved. If PyTorch is installed, NotiWriter uses torch.utils.tensorboard. Otherwise it falls back to tensorboardX, which is included as a package dependency.

Configuration

Argument	Env Variable	Default	Description
noti_api_key	NOTI_API_KEY	—	API key from the Noti app (required)
noti_server	NOTI_SERVER	https://notiapi.tech-webs.com	Noti server URL
noti_project	—	"default"	Project name
noti_run_name	—	Auto-generated	Run name
noti_enabled	NOTI_ENABLED	1	Set 0 to disable Noti sync
noti_complete_on_close	—	True	Mark the run as completed when close() is called

Release Files

This package is configured for standard PyPI release tooling:

python -m build
python -m twine check dist/*
python -m twine upload dist/*

There are helper scripts under sdk/scripts/ and a release checklist in sdk/RELEASING.md.

License

MIT