[Tensor] fix init context (#931)

* change torch.Parameter to ColoParameter

* fix post assignment for init context

* polish

* polish
pull/935/head
Ziyue Jiang 3 years ago committed by GitHub
parent dfc88b85ea
commit d73c2b1d79
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

@ -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
for name, param in name_list:
delattr(module, name)
# 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)
setattr(module, name, colo_param)
split = name.rfind('.')
if split >= 0: # param in submodule
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))
replaced_tensors = dict() # record mapping between (torch.Tensor, ColoTensor) to distinguish the same reference
for module_name, param_name in name_list:
submodule = module.get_submodule(module_name)
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)
# add mapping record
replaced_tensors[param] = colo_param
delattr(submodule, param_name)
setattr(submodule, param_name, colo_param)
ColoModulize(module)

@ -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)
_run_pretrain_load()
_test_pretrain_load(4)

Loading…
Cancel
Save