search

pesser / edflow

Reduce boilerplate code for your ML projects. TensorFlow and PyTorch.
https://edflow.readthedocs.io/
MIT License
24 stars 13 forks source link

do we need fastnumbers? #237

Closed pesser closed 4 years ago

pesser commented 4 years ago

are we happy without fastnumbers? #235 @jhaux can you say something? iirc you introduced it in the first place

jhaux commented 4 years ago

I would say, for now we are happy without fastnumbers as this bug has been encountered by quite a few people (@hperrot , @mritv, Richard). The reason for choosing fast_int was that we omit the less efficient try-except statement (see the image). fast_int is roughly 8 times faster but on a timescale of nano seconds. Thus we should try to get fast_int working again, but for now not using it is the fine and allows people to use edflow.

tiimeit