A realtime remote service to get the keras callbacks to the telegram including the details of metrics
Features:-
- It helps by getting the updates of your model including metrics and loss function graphs which help user the view and get to a statistical conclusion about the model remotely.
- It is a biggest advantage for the users who need not spend hours and hours infront of system for watching the updates of the model.
- Updates you get are from a telegram bot.
Installation:-
You can easily install this telegram using following command.
pip install tensorgram
Dependencies/Requirements:-
- Keras
- Tensorflow
- Requests
- Matplotlib
Works on python>=3.7
How to use:-
- Create a nueral network in keras.The sample code is as follows.
from keras.models import Sequential
from keras.layers import Dense, Dropout, Activation
from keras.optimizers import SGD
import numpy as np
import keras
X = np.array([[0,0],[0,1],[1,0],[1,1]])
y = np.array([[0],[1],[1],[0]])
model = Sequential()
model.add(Dense(8, input_dim=2))
model.add(Activation('tanh'))
model.add(Dense(1))
model.add(Activation('sigmoid'))
sgd = SGD(lr=0.1)
model.compile(loss='binary_crossentropy', optimizer=sgd,metrics=['accuracy'])
- Now go to Telegram app and search for @tensorgram_bot and join the channel by clicking on the chat.
-
Store it safely as it will be required later.
-
Now we need to import the TensorGram from tensorgram library using following code.
from tensorgram import TensorGram
- Now we need to create a object of TensorGram by specifying the following attributes like model name and chat id which you obtained before.
tf=TensorGram("model-name","123456789")
- Now you can start training the model and specify the object in the callbacks.
model.fit(X, y, batch_size=1, epochs=10,callbacks=[tf],verbose=1)
- Now if you open the telegram app you will find the updates as follows.
Bug / Feature Request:-
If you find a bug (gave undesired results), kindly open an issue here by including your search query and the expected result.
If you'd like to request a new function, feel free to do so by opening an issue here. Please include sample queries and their corresponding results.
Licence:-
This code is licensed under the MIT license, see LICENSE.txt.
Contact:-
For any kind of suggesstions/ help in code Please mail me at ksdkamesh99@gmail.com.