knockknock

Why we included this project

If you run long deep learning experiments, you know how easy it is to wait hours on a training run only to find it silently crashed at 2 AM. Knock Knock lets you wrap your main training function in a decorator and get a notification the moment the job finishes or fails, so you can step away and come back when there's something to act on. It covers a dozen channels, from email and Slack to Telegram, SMS, Discord, and WeChat Work, and includes the function's return value in the message. The library stays deliberately small, with no orchestration or monitoring stack required, which makes it a convenient drop-in for individual researchers and small teams who want a heads-up without adopting heavier tooling. It also has a command-line form, so you can alert on any shell command, not just Python training scripts.