Keras Callback

In order to receive all the updates of your model training let you use the KerasTelegramCallback. You can choose if to receive the current epoch progress bar and metrics plots!

Example

Screenshot

Code

import keras
from keras.models import Sequential
from keras.layers import Dense
from keras.optimizers import RMSprop
import numpy as np

from bob_telegram_tools.keras import KerasTelegramCallback
from bob_telegram_tools.bot import TelegramBot

X = np.random.rand(1000, 100)
y = (np.random.rand(1000, 3) > 0.5).astype('float32')

model = Sequential()
model.add(Dense(512, activation='relu', input_shape=(100,)))
model.add(Dense(512, activation='relu'))
model.add(Dense(3, activation='softmax'))

model.compile(loss='categorical_crossentropy',
              optimizer=RMSprop(),
              metrics=['accuracy'])

n_epochs = 3

token = '<your_token>'
user_id = int('<your_chat_id>')
bot = TelegramBot(token, user_id)

tl = KerasTelegramCallback(bot, epoch_bar=True, to_plot=[
    {
        'metrics': ['loss', 'val_loss']
    },
    {
        'metrics': ['acc', 'val_acc'],
        'title':'Accuracy plot',
        'ylabel':'acc',
        'ylim':(0, 1),
        'xlim':(1, n_epochs)
    }
])

history = model.fit(X, y,
                    batch_size=10,
                    epochs=n_epochs,
                    validation_split=0.15,
                    callbacks=[tl])