kerasplugins 🤓
Installation
pip install kerasplugins
Callbacks
Notification Callback 📮 💬
Get notified as your Keras neural network models trains remotely
Send log messages to your telegram and slack channels. more coming soon
Notify levels available:
- on_train_begin,
- on_train_end,
- on_batch_begin,
- on_batch_end,
- on_epoch_begin,
- on_epoch_end
from kerasplugins import callbacks
#Notify can either be a list, dict or set
notify = {
'on_batch_end', # sends BATCH END: Loss 0.50 Accuracy: 0.75
'on_epoch_end' # sends EPOCH END: Loss 0.43 Accuracy: 0.81
}
# msg is the initial message
msg = "Predicting Bitcoin Price"
telegram = callbacks.TelegramNotify(<token>, <chat_id>, msg=msg, notify=notify)
# channel is "#general" by default
slack = callbacks.SlackNotify(<slack_token>, <channel>, msg=msg, notify=notify)
webhook = callbacks.WebhookNotify('https://example.com/ml')
# Optionals:
#Headers default to "application/json"
headers = {'Content-Type': 'text/plain'}
# Added to every payload sent to the webhook
data = {"auth_token":"cccxxxvvvbbbnnnmmmaaasss"}
webhook = callbacks.WebhookNotify('https://example.com/ml', data=data, headers=headers)
model.fit(X_train, Y_train, validation_data=[X_test, Y_test], batch_size=256, epochs=10,
callbacks=[telegram, slack, webhook])Coming Soon
- Ability to stop training remotely
