0

My output is 332 dimension (300 glove + 32 my custom vector) values of this vector range from -1 to +1 I got horrible results using sigmoid as it confined the output to 0 to 1. I'm trying Tanh right now. What about Softmax? does it suit my case?

  • 3
    Can you elaborate? What is your end goal? What your model should do? Do you just want to calculate embeddings (for not known in GloVe words or calculate __additional__ 32 features - I didn't understand that part: `300 glove + 32 my custom vector`)? Your question is not clear... – MaxU - stand with Ukraine Jun 29 '18 at 11:47
  • Hey @MaxU , I'm working on a confidential project, so I can't share It here. My email Id is nsvikas1998@gmail.com . Could you please mail me? I saw from your profile you have very good experience in Machine Learning. I really need your help.! –  Jun 29 '18 at 13:24

0 Answers0