Make Differentiate public to the transform package

- Same reason as materialize

Author: ValerianRey

src/torchjd/autojac/_transform/_differentiate.py

@@ -8,7 +8,7 @@
 from .tensor_dict import _A
 
 
-class _Differentiate(Transform[_A, _A], ABC):
+class Differentiate(Transform[_A, _A], ABC):
     def __init__(
         self,
         outputs: Iterable[Tensor],

src/torchjd/autojac/_transform/grad.py

@@ -3,12 +3,12 @@
 import torch
 from torch import Tensor
 
-from ._differentiate import _Differentiate
+from ._differentiate import Differentiate
 from ._materialize import materialize
 from .tensor_dict import Gradients
 
 
-class Grad(_Differentiate[Gradients]):
+class Grad(Differentiate[Gradients]):
     def __init__(
         self,
         outputs: Iterable[Tensor],

src/torchjd/autojac/_transform/jac.py

@@ -6,12 +6,12 @@
 import torch
 from torch import Size, Tensor
 
-from ._differentiate import _Differentiate
+from ._differentiate import Differentiate
 from ._materialize import materialize
 from .tensor_dict import Jacobians
 
 
-class Jac(_Differentiate[Jacobians]):
+class Jac(Differentiate[Jacobians]):
     def __init__(
         self,
         outputs: Iterable[Tensor],