[Tensor] fix init context (#931)

* change torch.Parameter to ColoParameter

* fix post assignment for init context

* polish

* polish

colossalai/utils/model/colo_init_context.py

@@ -6,6 +6,21 @@ import types
 from torch import nn
 from typing import Iterator, Tuple, Union, Optional
 
+# find named_params includes replica
+def _named_params_with_replica(
+        module: nn.Module,
+        prefix: str = '',
+        recurse: bool = True,
+    ) -> Iterator[Tuple[str, Union[nn.Parameter, ColoTensor]]]:
+    modules = module.named_modules(prefix=prefix) if recurse else [(prefix, module)]
+
+    for mod_prefix, mod in modules:
+        for name, val in mod._parameters.items():
+            if val is None:
+                continue
+            name = mod_prefix + ('.' if mod_prefix else '') + name
+            yield name, val
+
 # Adapted from torch.nn.module.Module.register_param
 def _register_parameter_with_colotensor(self, name: str, param):
     if '_parameters' not in self.__dict__:
@@ -139,21 +154,36 @@ class ColoInitContext(InsertPostInitMethodToModuleSubClasses):
             return
 
         name_list = []
-        for name, param in module.named_parameters(recurse=False):
+        for name, param in _named_params_with_replica(module):
             if isinstance(param, ColoTensor):
                 continue
-            name_list.append((name, param))
 
-        save_torch_payload = True if not self._lazy_memory_allocate else False
+            split = name.rfind('.')
-        for name, param in name_list:
+            if split >= 0: # param in submodule
-            delattr(module, name)
+                module_name = name[:split]
+                param_name = name[split+1:]
+            else:
+                module_name = '' # param in current module
+                param_name = name
+            name_list.append((module_name, param_name))
 
-            # detaching tensor is necessary for optimizers.
+        replaced_tensors = dict() # record mapping between (torch.Tensor, ColoTensor) to distinguish the same reference
-            requires_grad = param.requires_grad
+        for module_name, param_name in name_list:
-            tensor_detached = param.to(self._device).detach()
+            submodule = module.get_submodule(module_name)
-            tensor_detached.requires_grad = requires_grad
+            param = submodule.get_parameter(param_name)
+            if param in replaced_tensors:
+                colo_param = replaced_tensors[param]
+            else:
+                save_torch_payload = True if not self._lazy_memory_allocate else False
+                # detaching tensor is necessary for optimizers.
+                requires_grad = param.requires_grad
+                tensor_detached = param.to(self._device).detach()
+                tensor_detached.requires_grad = requires_grad
 
-            colo_param = ColoParameter.init_from_torch_tensor(tensor=tensor_detached, save_payload=save_torch_payload)
+                colo_param = ColoParameter.init_from_torch_tensor(tensor=tensor_detached, save_payload=save_torch_payload)
-            setattr(module, name, colo_param)
+                # add mapping record
+                replaced_tensors[param] = colo_param
+            delattr(submodule, param_name)
+            setattr(submodule, param_name, colo_param)
 
         ColoModulize(module)

tests/test_tensor/test_model.py

@@ -370,16 +370,22 @@ def _run_pretrain_load():
 
     dict_pretrained = {}
     dict_col = {}
+    c_ref = 0
     for name, param in model_pretrained.named_parameters():
         dict_pretrained[name] = param
+        c_ref += 1
     c1 = 0
     c2 = 0
     for name, param in model.colo_named_parameters():
         if isinstance(param, ColoParameter):
-            c1 = c1 + 1
+            c1 += 1
         else:
-            c2 = c2 + 1
+            c2 +=1
         dict_col[name] = param
+    assert c_ref == c1
+    assert c2 == 0
+    if model_pretrained.cls.predictions.decoder.bias is model_pretrained.cls.predictions.bias:
+        assert model.cls.predictions.decoder.bias is model.cls.predictions.bias
 
     for name, param in dict_pretrained.items():
         check_equal(param, dict_col[name])
@@ -423,5 +429,4 @@ if __name__ == '__main__':
     # test_model_parameters()
     # test_colo_optimizer()
     # test_model()
-    # _test_pretrain_load(4)
+    _test_pretrain_load(4)
-    _run_pretrain_load()