You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Just thought that we could also generate the files included inside src/operators/implementation (i.e. operator__onnx__relu__6__T_tensor_float.c) from the Python script. So we can avoid manually creating them. Maybe all types is too much by now, so I would create only the float variants.
Quite useful also if someone wants to implement an operator. If the file and function is already in place (but empty) would be easier for someone without much knowledge of the code to implement and operator. We can abstract them from the complexity of running the Python script and so on.
Just thought that we could also generate the files included inside
src/operators/implementation
(i.e.operator__onnx__relu__6__T_tensor_float.c
) from the Python script. So we can avoid manually creating them. Maybe all types is too much by now, so I would create only thefloat
variants.Quite useful also if someone wants to implement an operator. If the file and function is already in place (but empty) would be easier for someone without much knowledge of the code to implement and operator. We can abstract them from the complexity of running the Python script and so on.
Originally posted by @alrevuelta in #18 (comment)
The text was updated successfully, but these errors were encountered: