Utility for generating linear embedding that contains status quo

Summary: A utility for generating an embedding that contains the status quo, so that SQ will be in-design in the embedding. Also adds IntToFloat to the ALEBO transorms since I need that for the current IGTV experiment.

Reviewed By: qingfeng10

Differential Revision: D23275586

fbshipit-source-id: 1180d6c01483037ed5f65cdbcd5cda49dca6f372

ax/modelbridge/strategies/alebo.py

@@ -16,18 +16,23 @@
 from ax.modelbridge.random import RandomModelBridge
 from ax.modelbridge.torch import TorchModelBridge
 from ax.modelbridge.transforms.centered_unit_x import CenteredUnitX
+from ax.modelbridge.transforms.int_to_float import IntToFloat
+from ax.modelbridge.transforms.remove_fixed import RemoveFixed
 from ax.modelbridge.transforms.standardize_y import StandardizeY
 from ax.models.random.alebo_initializer import ALEBOInitializer
 from ax.models.torch.alebo import ALEBO
 
 
+ALEBO_X_trans = [RemoveFixed, IntToFloat, CenteredUnitX]
+
+
 def get_ALEBOInitializer(
     search_space: SearchSpace, B: np.ndarray, **model_kwargs: Any
 ) -> RandomModelBridge:
     return RandomModelBridge(
         search_space=search_space,
         model=ALEBOInitializer(B=B, **model_kwargs),
-        transforms=[CenteredUnitX],
+        transforms=ALEBO_X_trans,  # pyre-ignore
     )
 
 
@@ -45,7 +50,7 @@ def get_ALEBO(
         search_space=search_space,
         data=data,
         model=ALEBO(B=B, **model_kwargs),
-        transforms=[CenteredUnitX, StandardizeY],
+        transforms=ALEBO_X_trans + [StandardizeY],  # pyre-ignore
         torch_dtype=B.dtype,
         torch_device=B.device,
     )