Open FengZhiheng opened 5 years ago
Sure! The code's pretty old and not very clean, I apologize.
The script that trains the model doesn't actually save it (that part is at the bottom and is commented out). Since I haven't kept this up to date for more recent versions of keras
I can't guarantee it'll work but if you just uncomment the bottom of addMNISTrnn.py
it should save the model, letting you open it in evaluateRNN.py
Note: the way I saved the model was a hack at the time because of the way saving/loading worked. Ever since TimeDistributed
got full support, it should work to just call model.save(...)
and then model = keras.models.load_model(...)
No such file or directory: '../models/basicRNN.json'. I'm very appreciate it if you can help me out.