Closed matsen closed 4 years ago
Be able to use float32 or float64 in TF and NP.
Have a pair of module-level global variables that describe the floating point dtype for TF and NP.
Then have every function that uses those dtypes just use the value encoded in the module.
Closed by #14
Goal
Be able to use float32 or float64 in TF and NP.
Proposal
Have a pair of module-level global variables that describe the floating point dtype for TF and NP.
Then have every function that uses those dtypes just use the value encoded in the module.