Enabled GPU support for the embedding lookup (#13)
The test split (how often the test data is divided to reduce memory pressure) is now a flag; Also replaced np.split with np.array_split since the first one throws an exception when equal splitting is not possible
Gradient summaries are written and their histograms can be viewed using tensor board
Preprocessing:
renamed preprocessing_step1.py to pattern_matching.py
renamed mappingsv2.py to word_mappings.py
I included the call to word_mappings.py in build_vocab.sh since for me it belongs together logically
Emoticons are now in a list and not hidden in the code
Added a script for running the preprocessing
Remote:
I moved the Euler test script and the tensor flow test script to a separate folder "remote"
I did multiple things in this PR:
General:
Preprocessing:
preprocessing_step1.py
topattern_matching.py
mappingsv2.py
toword_mappings.py
word_mappings.py
inbuild_vocab.sh
since for me it belongs together logicallyRemote: