Open DL-Alva opened 5 years ago
I meet the same problem. I want to train this model using my own class, whether should I change the content of "data/attributes.npy"?
using some NLP model(https://github.com/stanfordnlp/GloVe) to generate wordvector.txt, and then transfer the .txt to .npy. but why the loss is keeping Nan. (T T)
Sorry for your time.
When I try to train this model using my dataset, I change the data classes in file "data/voc_classes.txt". Wether should I change the content of "data/attributes.npy" Correspondingly? However, it seems that I can't find the code which generates the file "data/attributes.npy".