mirror of
https://github.com/pytorch/pytorch.git
synced 2025-10-20 21:14:14 +08:00
Updated Autograd Onboarding Lab (markdown)
@ -39,7 +39,7 @@ You can write this in a small standalone script that defines the Function and ru
|
||||
## III) Write native composite function and OpInfo
|
||||
|
||||
Now that you know that the formula is correct, you can implement a brand new native function in “native_functions.yaml” as a CompositeImplicit function.
|
||||
You can then add an OpInfo entry (in common_methods_invocations.py) to be able to test the new operator.
|
||||
You can then add an OpInfo entry (in [common_methods_invocations.py](torch/testing/_internal/common_methods_invocations.py) to be able to test the new operator.
|
||||
|
||||
## IV) Write custom formula in derivatives.yaml
|
||||
|
||||
|
Reference in New Issue
Block a user