English

简体中文

English

简体中文

Appearance

Bark ​

Bark is a free, open-source message push service application designed specifically for iOS devices. Its core functionality is simple and pure: it allows you to send custom push notifications to your iPhone from any device or platform.

You can think of it as a "privately customized" push service, similar to IFTTT or Telegram Bot, but lighter, more focused on the notifications themselves, and with your data entirely under your control.

If you want to be notified via a Bark message as soon as training is complete or an error occurs, it is highly recommended that you use the Bark notification plugin.

Improving the Plugin

SwanLab plugins are open source. You can view the Github source code. We welcome your suggestions and PRs!

Prerequisites ​

  1. Open the Bark App.
  2. On the "Server" page, copy the URL (the URL format is https://api.day.app/XXXXXXX) and note it down.
  3. On the "Settings" page, copy the Device Token and note it down.

Basic Usage ​

Using the Bark notification plugin is very simple. You just need to initialize a BarkCallback object and fill in the URL and Device Token from the prerequisites:

python
from swanlab.plugin.notification import BarkCallback

bark_callback = BarkCallback(
    key='Your Device Token',
    url='Your URL',
)

Then, pass the bark_callback object into the callbacks parameter of swanlab.init:

python
swanlab.init(callbacks=[bark_callback])

This way, when training finishes or an error occurs (triggering swanlab.finish()), you will receive a Bark message notification.

Custom Alerts ​

You can also use the send_msg method of the BarkCallback object to send custom Bark messages.

This is useful for alerting you when certain metrics reach a specific threshold!

python
if accuracy > 0.95:
    # Send a message in a custom scenario
    bark_callback.send_msg(
        content=f"Current Accuracy: {accuracy}",  # Notification content
    )

Externally Registering a Plugin ​

If you are using the integration of SwanLab with other frameworks and thus find it difficult to locate swanlab.init, you can use the swanlab.register_callbacks method to pass in plugins externally:

python
import swanlab

# Equivalent to swanlab.init(callbacks=[...])
swanlab.register_callbacks([...])

Limitations ​

  • The training completion/exception notification for the Bark notification plugin uses the on_stop lifecycle callback of SwanKitCallback. Therefore, if your process is abruptly killed, or the training machine shuts down unexpectedly, the on_stop callback will not be triggered, resulting in the Bark notification not being sent.