Skip to main content

Receive notifications about your model training anywhere you want!

Project description

TensorFlow Notification Callback

forthebadge made-with-python

PyPI version

A Tensorflow/Keras callback which sends information about your model training, on various messaging platforms.

Installation

Using pip:

pip install tf_notification_callback

Usage

Import the required module and add it to the list callbacks while training your model.

Example:

>>> from tf_notification_callback import TelegramCallback
>>> telegram_callback = TelegramCallback('<BotToken>',
                                     '<ChatID>',
	                             'CNN Model',
	                             ['loss', 'val_loss'],
	                             ['accuracy', 'val_accuracy'],
	                             True)
>>> model.fit(x_train, y_train,
          batch_size=32,
          epochs=10,
          validation_data=(x_test, y_test),
          callbacks=[telegram_callback])

Telegram

  1. Create a telegram bot using BotFather
    • Search for @BotFather on telegram.
    • Send /help to get list of all commands.
    • Send /newbot to create a new bot and complete the setup.
    • Copy the bot token after creating the bot.
  2. Get the chat ID
    • Search for the bot you created and send it any random message.
    • Go to this URL https://api.telegram.org/bot<BOT_TOKEN>/getUpdates (replace <BOT_TOKEN> with your bot token)
    • Copy the chat id of the user you want to send messages to.
  3. Use the TelegramCallback() class.
TelegramCallback(bot_token=None, chat_id=None, modelName='model', loss_metrics=['loss'], acc_metrics=[], getSummary=False):

Arguments:

bot_token : unique token of Telegram bot {str} chat_id : Telegram chat id you want to send message to {str} modelName : name of your model {str} loss_metrics : loss metrics you want in the loss graph {list of strings} acc_metrics : accuracy metrics you want in the accuracy graphs {list of strings} getSummary : Do you want message for each epoch (False) or a single message containing information about all epochs (True). {bool}

Slack

  1. Create a Slack workspace
  2. Create a new channel
  3. Search for the Incoming Webhooks in the Apps and install it.
  4. Copy the Webhook URL
  5. Import the SlackCallback() class. It takes in the following arguments

webhookURL : unique webhook URL of the app {str} channel : channel name or username you want to send message to {str} modelName : name of your model {str} loss_metrics : loss metrics you want in the loss graph {list of strings} acc_metrics : accuracy metrics you want in the accuracy graph {list of strings} getSummary : Do you want message for each epoch (False) or a single message containing information about all epochs (True). {bool}

Sending images in Slack is not supported currently.

ToDo

  • WhatsApp
  • E-Mail
  • Zulip
  • Messages

Motivation

As the Deep Learning models are getting more and more complex and computationally heavy, they take a very long time to train. During my internship, people used to start the model training and left it overnight. They could only check its progress the next day. So I thought it would be great if there was a simple way to get the training info remotely on their devices.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for tf-notification-callback, version 0.2
Filename, size File type Python version Upload date Hashes
Filename, size tf_notification_callback-0.2-py3-none-any.whl (4.2 kB) File type Wheel Python version py3 Upload date Hashes View
Filename, size tf_notification_callback-0.2.tar.gz (3.9 kB) File type Source Python version None Upload date Hashes View

Supported by

Pingdom Pingdom Monitoring Google Google Object Storage and Download Analytics Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page