quant_params should come from implicit nodes (#301)

Summary:
Pull Request resolved: #301

QuantParams are generated for nodes to imply that these operations are quantized operations. However, if the QuantNodes are explicit, i.e. being used to actually convert data type, then we should not be generating quant_params.

This is to help with mixed-datatype modles

Reviewed By: salilsdesai

Differential Revision: D48805092

fbshipit-source-id: 7a7423e93cdb2629842701ad8271a18464e60c33

Author: mcr229

backends/xnnpack/operators/quant_params.py

@@ -9,6 +9,7 @@
 from typing import cast, Optional, Union
 
 import torch
+from executorch.backends.xnnpack.passes.tag_implicit_q_dq_pass import TagImplicitQDqPass
 from executorch.backends.xnnpack.utils.quant_utils import is_dequant, is_quant
 from executorch.backends.xnnpack.utils.utils import check_or_raise
 from executorch.exir.dialects._ops import ops as exir_ops
@@ -179,7 +180,9 @@ def from_weights(cls, tensor_node: torch.fx.Node) -> Optional[QuantParams]:
     @classmethod
     def from_inputs(cls, tensor_node: torch.fx.Node) -> Optional[QuantParams]:
         # tensor_node is quantized if it is produced by a dequant node
-        if is_dequant(tensor_node):
+        if is_dequant(tensor_node) and TagImplicitQDqPass.is_tagged_as_implicit_q_dq(
+            tensor_node
+        ):
             return cls.from_q_dq_node(tensor_node)
 
         return None
@@ -190,7 +193,7 @@ def from_outputs(cls, tensor_node: torch.fx.Node) -> Optional[QuantParams]:
         if len(tensor_node.users) == 1:
             q = list(tensor_node.users.keys())[0]
             # Check if user is a q node
-            if is_quant(q):
+            if is_quant(q) and TagImplicitQDqPass.is_tagged_as_implicit_q_dq(q):
                 return cls.from_q_dq_node(q)
 
         return None