Hi Gaetan,

   All you need to do is declare the outputs as "float32" this should
   give you single precision outputs.

Hope this helps,

  Nick

Reply via email to