TF/Keras ML code for https://theartofpour.com
Thank you to everyone who attended my talk at this year's HomebrewCon.
I'll be working over the next few days on getting any remaining code/data deployed, as well as setting licenses and documentation.
Case in point, this README.
Cheers!
The trained models can be found as .h5 files in the taop-api /models
-
python 1-sql_to_pickle.py
to query and convert data to pickle format -
python 2-pickle_to_one_hot.py
to convert data to one-hot format -
python 3-train_model.py
to train the TF model -
python test.py
|python test_individual.py
|python test_specific.py
for specific testing runs
- Dynamic input dimensions
- Compile TF to match server hw