OS: MacOS Mojave 10.14.4 Python 3.7.6
conda 4.7.12 conda env create --name envname --file=environments.yml Python 3.7.6
python make_berty_up.py
We store SMSSpam in the date folder "./data/SMSSpam"
The training looks as follows
Train for 46 steps Epoch 00001: LearningRateScheduler reducing learning rate to 0.005. Epoch 1/2 2020-01-26 23:53:55.499368: I tensorflow/core/profiler/lib/profiler_session.cc:225] Profiler session started. 46/46 [==============================] - 357s 8s/step - loss: 0.3938 - acc: 0.8696 Epoch 00002: LearningRateScheduler reducing learning rate to 0.005. Epoch 2/2 46/46 [==============================] - 321s 7s/step - loss: 0.2327 - acc: 0.9443 Evaluating the model, we get train_acc: 0.96875 test_acc: 0.96875