Smarter checkpoints
Going back to earlier last checkpoint if persormance was better than at end of training
Scalers now work on on 2d and 3d arrays
full_load and full_save added which save and restore the rnn with all parameters
get_rnn_folder added which is a helper function for full_load and provides the folder name for an architecture with the correct notation
Also dataset added
1 parent 83f29d0 commit 697d8a2fdc7820123e311430130c0b87c74e8743
@Sascha Liechti Sascha Liechti authored on 27 Apr 2018
Showing 6 changed files
View
1_to_1_multi_layer.ipynb
View
matched_8hittracks.pkl 0 → 100644
Not supported
View
trained_models/rnn_model_lstm_leaky_relu_5l_[50,40,30,20,10]c/checkpoint 100644 → 0
View
trained_models/rnn_model_lstm_leaky_relu_5l_[50,40,30,20,10]c/rnn_basic.data-00000-of-00001 100644 → 0
Not supported
View
trained_models/rnn_model_lstm_leaky_relu_5l_[50,40,30,20,10]c/rnn_basic.index 100644 → 0
Not supported
View
trained_models/rnn_model_lstm_leaky_relu_5l_[50,40,30,20,10]c/rnn_basic.meta 100644 → 0
Not supported