21 lines
723 B
Python
21 lines
723 B
Python
|
# flake8: noqa: F401
|
||
|
|
r"""
|
||
|
|
This file is in the process of migration to `torch/ao/quantization`, and
|
||
|
|
is kept here for compatibility while the migration process is ongoing.
|
||
|
|
If you are adding a new entry/functionality, please, add it to the
|
||
|
|
appropriate files under `torch/ao/quantization/fx/`, while adding an import statement
|
||
|
|
here.
|
||
|
|
"""
|
||
|
|
from torch.ao.quantization.fx.utils import (
|
||
|
|
all_node_args_have_no_tensors,
|
||
|
|
assert_and_get_unique_device,
|
||
|
|
create_getattr_from_value,
|
||
|
|
get_custom_module_class_keys,
|
||
|
|
get_linear_prepack_op_for_dtype,
|
||
|
|
get_new_attr_name_with_prefix,
|
||
|
|
get_non_observable_arg_indexes_and_types,
|
||
|
|
get_qconv_prepack_op,
|
||
|
|
graph_module_from_producer_nodes,
|
||
|
|
maybe_get_next_module,
|
||
|
|
)
|