I’ll look at using callbacks using LSTM networks…

Environment Setup

Python 3.5 is needed for TensorFlow 1.4+. A link to the requirements.txt can be found here


brew install pyenv
pyenv install 3.5.4
cd
virtualenv -p /Users/mchirico/.pyenv/versions/3.5.4/bin/python3.5 pbug
. pbug/bin/activate

pip install -r requirements.txt



import keras

class My_Callback(keras.callbacks.Callback):
    def on_train_begin(self, logs={}):
        return

    def on_train_end(self, logs={}):
        return

    def on_epoch_begin(self, logs={}):
        return

    def on_epoch_end(self, epoch, logs={}):
        return

    def on_batch_begin(self, batch, logs={}):
        return

    def on_batch_end(self, batch, logs={}):
        self.losses.append(logs.get('loss'))
        return

References

  1. Blog with ideas

  2. Video – pull out links

  3. Deep Learning with Keras