Remove stale code (#3334)

Signed-off-by: Kyunggeun Lee <quic_kyunggeu@quicinc.com>

TrainingExtensions/torch/src/python/aimet_torch/quantsim_config/builder.py

@@ -39,7 +39,6 @@
 import itertools
 from typing import List, Optional, Tuple
 import torch
-from torch import nn
 
 from aimet_common.defs import QuantScheme, QuantizationDataType, MAP_ROUND_MODE_TO_PYMO
 from aimet_common.utils import AimetLogger, log_with_error_and_assert_if_false
@@ -205,41 +204,18 @@ def realize_v2_wrapper(self):
         else:
             quantized_module = _legacy_impl.FakeQuantizationMixin.from_module(self._module_to_wrap)
 
-        def set_recursive(module_list, i, quantizer):
-            """
-            Set quantizer recursively.
-            AIMET V1 handles nested input/output tensors with single input quantizers,
-            whereas V2 quantized module allows having nested input/output quantizers.
-            (For reference, see the class definition of FakeQuantizedLSTM in fake_quant.py)
-
-            To implement V1 behavior, we set the nested input/output quantizers to
-            share the same single quantizer, for example as below:
-              - self.input_quantizers = [q1, q2, q3]
-              - quant_module.input_quantizers = [None, [None, None], None]
-                (before set_recursive)
-              - quant_module.input_quantizers = [q1, [q2, q2], q3]
-                (after set_recursive)
-            """
-            if module_list[i] is None:
-                module_list[i] = quantizer
-            elif isinstance(module_list[i], nn.ModuleList):
-                for j in range(len(module_list[i])):
-                    set_recursive(module_list[i], j, quantizer)
-            else:
-                raise RuntimeError
-
         # For unused modules, quantsim assumes # inputs = # outputs = 1
         # If this is incorrect, propagate the configuration of the last input/output quantizers to the remaining
         # quantizer positions
         for i, _ in list(enumerate(quantized_module.input_quantizers)):
             q_idx = min(i, len(self.input_quantizers) - 1)
             quantizer = self.input_quantizers[q_idx].realize()
-            set_recursive(quantized_module.input_quantizers, i, quantizer)
+            quantized_module.input_quantizers[i] = quantizer
 
         for i, _ in list(enumerate(quantized_module.output_quantizers)):
             q_idx = min(i, len(self.input_quantizers) - 1)
             quantizer = self.output_quantizers[q_idx].realize()
-            set_recursive(quantized_module.output_quantizers, i, quantizer)
+            quantized_module.output_quantizers[i] = quantizer
 
         for param_name, quant_builder in self.param_quantizers.items():
             quantized_module.param_quantizers[param_name] = quant_builder.realize()