quant params from static inputs (#573)

Summary:

Since we allow tensor constants to be quantized inputs, we need to adjust the from_inputs api to search if this input is static or not. If it is static, then we take the first q node in get_attr --> q --> dq. If it is not static, then we just take the dq node to create the QuantParams object.

In the past, we can take in static quant inputs only on weights and biases.

Differential Revision: D49850149

Author: mcr229

backends/xnnpack/operators/quant_params.py

@@ -183,6 +183,11 @@ def from_inputs(cls, tensor_node: torch.fx.Node) -> Optional[QuantParams]:
         if is_dequant(tensor_node) and TagImplicitQDqPass.is_tagged_as_implicit_q_dq(
             tensor_node
         ):
+            dq_input = cast(torch.fx.Node, tensor_node.args[0])
+            if is_quant(dq_input):
+                q_input = cast(torch.fx.Node, dq_input.args[0])
+                if q_input.op == "placeholder":
+                    return cls.from_q_dq_node(dq_input)
             return cls.from_q_dq_node(tensor_node)
 
         return None