about zerodivisionerror

[JasonYorkPig]

hello amine! I tried to train 5 emotions with using landmarks. the train environment is like this:

• emotions = 5
• model = B
• optimizer = 'momentum'
• learning_rate = 0.016
• learning_rate_decay = 0.864
• otimizer_param (momentum) = 0.95
• keep_prob = 0.956
• epochs = 100
• use landmarks = True
• use hog + landmarks = False
• use hog sliding window + landmarks = False
• use batchnorm after conv = True
• use batchnorm after fc = False_

and I also changed [output_size=5] in parameters.py. but when I tried to train,

Traceback (most recent call last): File "train.py", line 124, in train() File "train.py", line 60, in train n_epoch=TRAINING.epochs) File "/anaconda3/envs/tf/lib/python3.6/site-packages/tflearn/models/dnn.py", line 216, in fit callbacks=callbacks) File "/anaconda3/envs/tf/lib/python3.6/site-packages/tflearn/helpers/trainer.py", line 339, in fit show_metric) File "/anaconda3/envs/tf/lib/python3.6/site-packages/tflearn/helpers/trainer.py", line 847, in _train e = evaluate_flow(self.session, eval_ops, self.test_dflow) File "/anaconda3/envs/tf/lib/python3.6/site-packages/tflearn/helpers/trainer.py", line 1003, in evaluate_flow res = [r / dataflow.n_samples for r in res] File "/anaconda3/envs/tf/lib/python3.6/site-packages/tflearn/helpers/trainer.py", line 1003, in res = [r / dataflow.n_samples for r in res] ZeroDivisionError: float division by zero

Is there anything else that needs to be fixed? :)

[amineHorseman]

Can you add a print statement before line 55 to check the size of data['X'], data['X2'] and data['Y']?

[JasonYorkPig]

What I found was that when I specify the emotions, skip two or more label numbers(0,2,4) then the ZeroDivisionError was happened.

[LauraKag]

I have the exact same issue...how can I fix it?