feat(mge/quantization): add `mapping` parameter for custom modules

GitOrigin-RevId: a4de4261d0

python_module/megengine/module/module.py

@@ -60,7 +60,7 @@ class Module(metaclass=ABCMeta):
     def __init__(self):
         # runtime attributes
         self.training = True
-        self.quantize_diabled = False
+        self.quantize_disabled = False
 
         # hooks
         self._forward_pre_hooks = OrderedDict()
@@ -328,12 +328,12 @@ class Module(metaclass=ABCMeta):
 
     def disable_quantize(self, value=True):
         r"""
-        Set ``module``'s ``quantize_diabled`` attribute and return ``module``.
+        Set ``module``'s ``quantize_disabled`` attribute and return ``module``.
         Could be used as a decorator.
         """
 
         def fn(module: Module) -> None:
-            module.quantize_diabled = value
+            module.quantize_disabled = value
 
         self.apply(fn)
 

python_module/megengine/quantization/quantize.py

@@ -5,7 +5,7 @@
 # Unless required by applicable law or agreed to in writing,
 # software distributed under the License is distributed on an
 # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-from copy import deepcopy
+from copy import copy, deepcopy
 from typing import Callable, Dict, Tuple
 
 from .. import module as Float
@@ -49,19 +49,24 @@ def _get_convert_dict() -> Tuple[
 _float2qat_dict, _qat2quantized_dict = _get_convert_dict()
 
 
-def quantize(module: Module, inplace: bool = True):
+def quantize(module: Module, inplace: bool = True, mapping: dict = None):
     r"""
     Recursively convert :class:`~.QATModule` to :class:`~.QuantizedModule`
     through :meth:`~.Module.apply`.
 
     :param module: root module to do convert recursively.
     :param inplace: whether to convert submodules in-place.
+    :param mapping: a dict indicating how to convert custom modules from QATModule to
+        QuantizedModule. Will be combined with internal default convert mapping dict.
     """
 
     if not inplace:
         module = deepcopy(module)
 
-    qat_modules = tuple(_qat2quantized_dict.keys())
+    convert_dict = copy(_qat2quantized_dict)
+    if mapping is not None:
+        convert_dict.update(mapping)
+    qat_modules = tuple(convert_dict.keys())
 
     def is_qat(mod: Module):
         return isinstance(mod, qat_modules)
@@ -70,7 +75,7 @@ def quantize(module: Module, inplace: bool = True):
     for key, submodule, parent in list(
         module._flatten(with_key=True, with_parent=True, predicate=is_qat)
     ):
-        new_mod = _qat2quantized_dict[type(submodule)].from_qat_module(submodule)
+        new_mod = convert_dict[type(submodule)].from_qat_module(submodule)
         if isinstance(parent, Float.Sequential):
             # cannnot use setattr to be compatible with Sequential's ``__setitem__``
             parent[int(key.split(".")[-1])] = new_mod
@@ -81,7 +86,10 @@ def quantize(module: Module, inplace: bool = True):
 
 
 def quantize_qat(
-    module: Module, inplace: bool = True, qconfig: QConfig = ema_fakequant_qconfig,
+    module: Module,
+    inplace: bool = True,
+    qconfig: QConfig = ema_fakequant_qconfig,
+    mapping: dict = None,
 ):
     r"""
     Recursively convert float :class:`~.Module` to :class:`~.QATModule`
@@ -91,12 +99,17 @@ def quantize_qat(
     :param inplace: whether to convert submodules in-place.
     :param qconfig: an instance of :class:`~.QConfig` to be set as submodules' qconfig.
         default is ``ema_fakequant_qconfig``.
+    :param mapping: a dict indicating how to convert custom modules from Module to QATModule.
+        Will be combined with internal default convert mapping dict.
     """
 
     if not inplace:
         module = deepcopy(module)
 
-    quantable_modules = tuple(_float2qat_dict.keys())
+    convert_dict = copy(_float2qat_dict)
+    if mapping is not None:
+        convert_dict.update(mapping)
+    quantable_modules = tuple(convert_dict.keys())
 
     def is_quantable(mod: Module):
         return isinstance(mod, quantable_modules)
@@ -106,10 +119,10 @@ def quantize_qat(
         module._flatten(with_key=True, with_parent=True, predicate=is_quantable)
     ):
         # only convert top quantable module.
-        if is_quantable(parent) or submodule.quantize_diabled:
+        if is_quantable(parent) or submodule.quantize_disabled:
             continue
 
-        new_mod = _float2qat_dict[type(submodule)].from_float_module(submodule)
+        new_mod = convert_dict[type(submodule)].from_float_module(submodule)
         if isinstance(parent, Float.Sequential):
             # cannnot use setattr to be compatible with Sequential's ``__setitem__``
             parent[int(key.split(".")[-1])] = new_mod

python_module/test/unit/quantization/quantize.py

@@ -52,3 +52,29 @@ def test_disable_quantize():
     qat_net = quantize_qat(net, inplace=False)
     assert isinstance(qat_net.conv, Float.ConvBnRelu2d)
     assert isinstance(qat_net.conv.conv, Float.Conv2d)
+
+
+def test_convert_with_custom_mapping():
+    class FloatExample(Float.Module):
+        def forward(self, x):
+            return x
+
+    class QATExample(QAT.QATModule):
+        def forward(self, x):
+            return x
+
+        @classmethod
+        def from_float_module(cls, float_module):
+            return cls()
+
+    class Net(Float.Module):
+        def __init__(self):
+            super().__init__()
+            self.example = FloatExample()
+
+        def forward(self, x):
+            return self.example(x)
+
+    net = Net()
+    qat_net = quantize_qat(net, inplace=False, mapping={FloatExample: QATExample})
+    assert isinstance(qat_net.example, QATExample)