A python package for real time tracking of keras model training.
- Tensorflow or Keras
- pika (RabbitMQ Python client)
- requests (Request HTTP Python library)
- Active Internet Connection
Meant to be used in combination with an Android Client: https://github.com/thevarunsharma/DL-Tracker-App
Note: Some additional prior setup is needed to be done before using this package which includes setting up a webhook for database updation and an AMQP cloud server.
- First, make sure that the above requirements are satisfied.
- Then, add this package in your current working directory
"""
model:
A tensorflow or keras model instance created using keras.models.Model or keras.models.Sequential
modelKey:
unique key identifying the model, to be obtained from the App
password:
user profile password
"""
# import Tracker callback
from dl_tracker import Tracker
# create a Tracker instance, pass modelKey and password as arguments
tracker = Tracker(modelKey, password)
# pass tracker among callbacks while calling .fit ot .fit_generator method on model
model.fit(..., callbacks=[tracker, ...])
# or
model.fit_generator(..., callbacks=[tracker, ...])