"Load failed" error

[Stephen DeLay]

While trying to train or run, it throws a warning saying [!] Load Failed. Looking at the code, this seems to be related to the checkpoint directory, however, I have never actually seen the directory used (it has never had any files in it).

[Terence Broad]

You can only run the model if you have trained and saved one. In this version of the code, the model is only saved once 1000 mini-batches have been processed, this can be changed in line 245 of model.py

if np.mod(counter, 500) == 2 and counter > 450: self.save(config.checkpoint_dir, counter)
I have now edited the readme to clarify this issue.