search

nicolas-ivanov / debug_seq2seq

[unmaintained] Make seq2seq for keras work
233 stars 86 forks source link

I found a small bug #7

Closed vangorik closed 8 years ago

vangorik commented 8 years ago

File: lib/nn_model/train.py Line: 68 def save_model(nn_model): #model_full_path = os.path.join(DATA_PATH, 'nn_models', NN_MODEL_PATH) nn_model.save_weights(NN_MODEL_PATH, overwrite=True)

liveabstract commented 8 years ago

@vangorik : nn_models and nn_model, is that what you are saying?

Also, after fixing the issue, was there any change in the quality of results?

vangorik commented 8 years ago

before fix python reported error because:

If i set in line 5: DATA_PATH = './test'

config.py line: 53 NN_MODEL_PATH = os.path.join(DATA_PATH, 'nn_models', NN_MODEL_NAME)

NN_MODEL_PATH = './test/nn_models/'+NN_MODEL_NAME

and in lib/nn_model/train.py: model_full_path = os.path.join(DATA_PATH, 'nn_models', NN_MODEL_PATH)

model_full_path =

'./test/nn_models/./test/nn_models/'+NN_MODEL_NAME <<----- FAIL and save_weights dont save nn_model nn_model.save_weights(model_full_path, overwrite=True)

Sorry about my poor English ;)

vangorik commented 8 years ago

ok i create pull request and corect others bugs :)

nicolas-ivanov commented 8 years ago

@vangorik thank you, I'm gonna take a look at it now