Source code for torch.export
import copy
import dataclasses
import io
import pathlib
import typing
from enum import auto, Enum
from typing import Any, Callable, Dict, List, Optional, Tuple, Union
import sympy
import torch
import torch.fx._pytree as fx_pytree
import torch.utils._pytree as pytree
from torch.fx._compatibility import compatibility
from torch.fx.experimental.symbolic_shapes import StrictMinMaxConstraint
from torch.fx.passes.infra.pass_base import PassResult
from torch.fx.passes.infra.pass_manager import PassManager
__all__ = [
"ArgumentKind",
"ArgumentSpec",
"Constraint",
"ExportBackwardSignature",
"ExportGraphSignature",
"ExportedProgram",
"ModuleCallEntry",
"ModuleCallSignature",
"constrain_as_size",
"constrain_as_value",
"dynamic_dim",
"export",
"load",
"save",
]
PassType = Callable[[torch.fx.GraphModule], Optional[PassResult]]
[docs]@dataclasses.dataclass
class ExportBackwardSignature:
gradients_to_parameters: Dict[str, str]
gradients_to_user_inputs: Dict[str, str]
loss_output: str
[docs]@dataclasses.dataclass
class ExportGraphSignature:
"""
:class:`ExportGraphSignature` models the input/output signature of Export Graph,
which is a fx.Graph with stronger invariants gurantees.
Export Graph is functional and does not access "states" like parameters
or buffers within the graph via ``getattr`` nodes. Instead, :func:`export`
gurantees that parameters and buffers are lifted out of the graph as inputs.
Similarly, any mutations to buffers are not included in the graph either,
instead the updated values of mutated buffers are modeled as additional outputs
of Export Graph.
The ordering of all inputs and outputs are::
Inputs = [*parameters_buffers, *flattened_user_inputs]
Outputs = [*mutated_inputs, *flattened_user_outputs]
e.g. If following module is exported::
class CustomModule(nn.Module):
def __init__(self):
super(CustomModule, self).__init__()
# Define a parameter
self.my_parameter = nn.Parameter(torch.tensor(2.0))
# Define two buffers
self.register_buffer('my_buffer1', torch.tensor(3.0))
self.register_buffer('my_buffer2', torch.tensor(4.0))
def forward(self, x1, x2):
# Use the parameter, buffers, and both inputs in the forward method
output = (x1 + self.my_parameter) * self.my_buffer1 + x2 * self.my_buffer2
# Mutate one of the buffers (e.g., increment it by 1)
self.my_buffer2.add_(1.0) # In-place addition
return output
Resulting Graph would be::
graph():
%arg0_1 := placeholder[target=arg0_1]
%arg1_1 := placeholder[target=arg1_1]
%arg2_1 := placeholder[target=arg2_1]
%arg3_1 := placeholder[target=arg3_1]
%arg4_1 := placeholder[target=arg4_1]
%add_tensor := call_function[target=torch.ops.aten.add.Tensor](args = (%arg3_1, %arg0_1), kwargs = {})
%mul_tensor := call_function[target=torch.ops.aten.mul.Tensor](args = (%add_tensor, %arg1_1), kwargs = {})
%mul_tensor_1 := call_function[target=torch.ops.aten.mul.Tensor](args = (%arg4_1, %arg2_1), kwargs = {})
%add_tensor_1 := call_function[target=torch.ops.aten.add.Tensor](args = (%mul_tensor, %mul_tensor_1), kwargs = {})
%add_tensor_2 := call_function[target=torch.ops.aten.add.Tensor](args = (%arg2_1, 1.0), kwargs = {})
return (add_tensor_2, add_tensor_1)
Resulting ExportGraphSignature would be::
ExportGraphSignature(
# Indicates that there is one parameter named `my_parameter`
parameters=['L__self___my_parameter'],
# Indicates that there are two buffers, `my_buffer1` and `my_buffer2`
buffers=['L__self___my_buffer1', 'L__self___my_buffer2'],
# Indicates that the nodes `arg3_1` and `arg4_1` in produced graph map to
# original user inputs, ie. x1 and x2
user_inputs=['arg3_1', 'arg4_1'],
# Indicates that the node `add_tensor_1` maps to output of original program
user_outputs=['add_tensor_1'],
# Indicates that there is one parameter (self.my_parameter) captured,
# its name is now mangled to be `L__self___my_parameter`, which is now
# represented by node `arg0_1` in the graph.
inputs_to_parameters={'arg0_1': 'L__self___my_parameter'},
# Indicates that there are two buffers (self.my_buffer1, self.my_buffer2) captured,
# their name are now mangled to be `L__self___my_my_buffer1` and `L__self___my_buffer2`.
# They are now represented by nodes `arg1_1` and `arg2_1` in the graph.
inputs_to_buffers={'arg1_1': 'L__self___my_buffer1', 'arg2_1': 'L__self___my_buffer2'},
# Indicates that one buffer named `L__self___my_buffer2` is mutated during execution,
# its new value is output from the graph represented by the node named `add_tensor_2`
buffers_to_mutate={'add_tensor_2': 'L__self___my_buffer2'},
# Backward graph not captured
backward_signature=None,
# Work in progress feature, please ignore now.
assertion_dep_token=None
)
"""
# A list of parameters uniquely identified by mangled fully qualified name
parameters: List[str]
# A list of buffers uniquely identified by mangled fully qualified name
buffers: List[str]
# Graph node names of pytree-flattened inputs of original program
user_inputs: List[str]
# Graph node names of pytree-flattened outputs of original program
user_outputs: List[str]
# A dictionary mapping graph input node names to parameters. If a graph input
# name is found in this dictionary, it is guranteed to be a lifted parameter.
inputs_to_parameters: Dict[str, str]
# A dictionary mapping graph input node names to buffers. If a graph input
# name is found in this dictionary, it is guranteed to be a lifted buffer.
inputs_to_buffers: Dict[str, str]
# A dictionary mapping graph output node names to buffers that are mutated in the
# original program. Buffers that are not mutated will not be found in this dictionary.
buffers_to_mutate: Dict[str, str]
backward_signature: Optional[ExportBackwardSignature]
# Map from assertion dependency token index to assertion dep token output
# name in output. The shape of output after aot_autograd will be like:
# (updated_inputs, user_outputs, dep_token).
assertion_dep_token: Optional[Dict[int, str]] = None
def __post_init__(self) -> None:
assertion_dep_token = self.assertion_dep_token
if assertion_dep_token is None:
return
assert len(assertion_dep_token) == 1
assertion_dep_token_index = list(assertion_dep_token.keys())[0]
assert (
len(self.user_outputs) + len(self.buffers_to_mutate)
== assertion_dep_token_index
)
[docs]@dataclasses.dataclass
class ArgumentSpec:
kind: ArgumentKind
value: Any
def __post_init__(self):
if self.kind in (ArgumentKind.Tensor, ArgumentKind.SymInt):
assert isinstance(self.value, str)
[docs]@dataclasses.dataclass
class ModuleCallSignature:
inputs: List[ArgumentSpec]
outputs: List[ArgumentSpec]
in_spec: pytree.TreeSpec
out_spec: pytree.TreeSpec
[docs]@dataclasses.dataclass
class ModuleCallEntry:
fqn: str
signature: Optional[ModuleCallSignature] = None
[docs]class ExportedProgram:
"""
Package of a program from :func:`export`. It contains
an :class:`torch.fx.Graph` that represents Tensor computation, a state_dict containing
tensor values of all lifted parameters and buffers, and various metadata.
You can call an ExportedProgram like the original callable traced by
:func:`export` with the same calling convention.
To perform transformations on the graph, use ``.module`` property to access
an :class:`torch.fx.GraphModule`. You can then use
`FX transformation <https://pytorch.org/docs/stable/fx.html#writing-transformations>`_
to rewrite the graph. Afterwards, you can simply use :func:`export`
again to construct a correct ExportedProgram.
"""
def __init__(
self,
root: Union[torch.nn.Module, Dict[str, Any]],
graph: torch.fx.Graph,
graph_signature: ExportGraphSignature,
call_spec: Any,
state_dict: Dict[str, Union[torch.Tensor, torch.nn.Parameter]],
range_constraints: Dict[sympy.Symbol, Any],
equality_constraints: List[Tuple[Any, Any]],
module_call_graph: List[ModuleCallEntry],
example_inputs: Optional[Tuple[Tuple[Any, ...], Dict[str, Any]]] = None,
):
from torch._export.exported_program import (
_create_graph_module_for_export,
CallSpec,
)
from torch._export.passes.add_runtime_assertions_for_constraints_pass import (
InputDim,
RangeConstraint,
)
# Remove codegen related things from the graph. It should just be a flat graph.
graph._codegen = torch.fx.graph.CodeGen()
self._graph_module = _create_graph_module_for_export(root, graph)
if isinstance(root, torch.fx.GraphModule):
self._graph_module.meta.update(root.meta)
self._graph_signature: ExportGraphSignature = graph_signature
self._call_spec: CallSpec = call_spec
self._state_dict: Dict[str, Any] = state_dict
self._range_constraints: Dict[sympy.Symbol, RangeConstraint] = range_constraints
self._equality_constraints: List[
Tuple[InputDim, InputDim]
] = equality_constraints
self._module_call_graph: List[ModuleCallEntry] = module_call_graph
self._example_inputs = example_inputs
@property
@compatibility(is_backward_compatible=False)
def graph_module(self):
return self._graph_module
@property
@compatibility(is_backward_compatible=False)
def graph(self):
return self.graph_module.graph
@property
@compatibility(is_backward_compatible=False)
def graph_signature(self):
return self._graph_signature
@property
@compatibility(is_backward_compatible=False)
def state_dict(self):
return self._state_dict
@property
@compatibility(is_backward_compatible=False)
def call_spec(self):
return self._call_spec
@property
@compatibility(is_backward_compatible=False)
def range_constraints(self):
return self._range_constraints
@property
@compatibility(is_backward_compatible=False)
def equality_constraints(self):
return self._equality_constraints
@property
@compatibility(is_backward_compatible=False)
def module_call_graph(self):
return self._module_call_graph
@property
@compatibility(is_backward_compatible=False)
def example_inputs(self):
return self._example_inputs
def __call__(self, *args: Any, **kwargs: Any) -> Any:
import torch._export.error as error
from torch._export import combine_args_kwargs
if self.call_spec.in_spec is not None:
try:
user_args = combine_args_kwargs(args, kwargs)
args = fx_pytree.tree_flatten_spec(user_args, self.call_spec.in_spec) # type: ignore[assignment]
except Exception:
_, received_spec = pytree.tree_flatten(user_args)
raise error.InternalError(
"Trying to flatten user inputs with exported input tree spec: \n"
f"{self.call_spec.in_spec}\n"
"but actually got inputs with tree spec of: \n"
f"{received_spec}"
)
ordered_params = tuple(
self.state_dict[name] for name in self.graph_signature.parameters
)
ordered_buffers = tuple(
self.state_dict[name] for name in self.graph_signature.buffers
)
self._check_input_constraints(*ordered_params, *ordered_buffers, *args)
with torch.no_grad():
# NOTE: calling convention is first params, then buffers, then args as user supplied them.
# See: torch/_functorch/aot_autograd.py#L1034
res = torch.fx.Interpreter(self.graph_module).run(
*ordered_params, *ordered_buffers, *args, enable_io_processing=False
)
if self.call_spec.out_spec is not None:
mutation = self.graph_signature.buffers_to_mutate
num_mutated = len(mutation)
mutated_buffers = res[:num_mutated]
# Exclude dependency token from final result.
assertion_dep_token = self.graph_signature.assertion_dep_token
if assertion_dep_token is not None:
assertion_dep_token_index = list(assertion_dep_token.keys())[0]
res = res[:assertion_dep_token_index]
res = res[num_mutated:]
try:
res = pytree.tree_unflatten(res, self.call_spec.out_spec)
except Exception:
_, received_spec = pytree.tree_flatten(res)
raise error.InternalError(
"Trying to flatten user outputs with exported output tree spec: \n"
f"{self.call_spec.out_spec}\n"
"but actually got outputs with tree spec of: \n"
f"{received_spec}"
)
finally:
ix = 0
for buffer in self.graph_signature.buffers_to_mutate.values():
self.state_dict[buffer] = mutated_buffers[ix]
ix += 1
return res
def __str__(self) -> str:
graph_module = self.graph_module.print_readable(print_output=False).replace(
"\n", "\n "
)
string = (
"ExportedProgram:\n"
f" {graph_module}\n"
f"Graph Signature: {self.graph_signature}\n"
f"Symbol to range: {self.range_constraints}\n"
)
return string
[docs] def module(self) -> torch.nn.Module:
"""
Returns a self contained GraphModule with all the parameters/buffers inlined.
"""
from torch._export.exported_program import unlift_exported_program_lifted_states
return unlift_exported_program_lifted_states(self)
def _transform(self, *passes: PassType) -> "ExportedProgram":
from torch._export.passes.add_runtime_assertions_for_constraints_pass import (
RangeConstraint,
)
pm = PassManager(list(passes))
res = pm(self.graph_module)
transformed_gm = res.graph_module if res is not None else self.graph_module
assert transformed_gm is not None
def _get_updated_range_constraints(
gm: torch.fx.GraphModule,
) -> Dict[sympy.Symbol, RangeConstraint]:
def get_shape_env(gm):
vals = [
node.meta["val"]
for node in gm.graph.nodes
if node.meta.get("val", None) is not None
]
from torch._guards import detect_fake_mode
fake_mode = detect_fake_mode(vals)
if fake_mode is not None:
return fake_mode.shape_env
for v in vals:
if isinstance(v, torch.SymInt):
return v.node.shape_env
shape_env = get_shape_env(gm)
if shape_env is None:
return {}
range_constraints = {
k: RangeConstraint(v.lower, v.upper)
for k, v in shape_env.var_to_range.items()
}
return range_constraints
def get_output_node_names(gm):
output_node = list(gm.graph.nodes)[-1]
assert output_node.op == "output"
return [str(arg) for arg in output_node.args[0]]
def get_input_node_names(gm):
return [node.name for node in gm.graph.nodes if node.op == "placeholder"]
def _generate_new_graph_signature(old_ep, new_gm):
"""
Update graph_signature according to graph after transformation.
Transformations can lead to node name changes, which are used in
graph_signature to identify inputs and outputs. Therefore, after each
transformation, we need to update the graph_signature according to
new node names.
WARNING: This implementation makes a few assumptions
- The transformation doesn't change number of inputs/outputs
- Each input/output still has the same meaning.
- For inputs, that means that the inputs in transformed
graph map to the same lifted parameter/buffer or user
input as the input of the same position in the graph
before transformation.
- Similarly for outputs, each output should correspond to the
same mutated buffer or user output as the output value of
the same position in the graph before transformation.
It is difficult to programatically validate these assumptions, but they
should hold true most of the time as inputs/outputs of the graph rarely
need to be changed.
"""
old_signature = old_ep.graph_signature
old_gm = old_ep.graph_module
old_graph_input_node_names = get_input_node_names(old_gm)
new_graph_input_node_names = get_input_node_names(new_gm)
assert len(old_graph_input_node_names) == len(
new_graph_input_node_names
), f"""
Number of input nodes changed from {len(old_graph_input_node_names)}
to {len(new_graph_input_node_names)} after transformation. This
transformation is currently not supported.
"""
old_graph_output_node_names = get_output_node_names(old_gm)
new_graph_output_node_names = get_output_node_names(new_gm)
assert len(old_graph_output_node_names) == len(
new_graph_output_node_names
), f"""
Number of output values changed from {len(old_graph_output_node_names)}
to {len(new_graph_output_node_names)} after transformation. This
transformation is currently not supported.
"""
node_names_mapping = dict(
zip(
old_graph_input_node_names + old_graph_output_node_names,
new_graph_input_node_names + new_graph_output_node_names,
)
)
new_signature = copy.deepcopy(old_signature)
new_signature.user_inputs = [
node_names_mapping[old_user_input]
for old_user_input in old_signature.user_inputs
]
new_signature.user_outputs = [
node_names_mapping[old_user_output]
for old_user_output in old_signature.user_outputs
]
new_signature.inputs_to_parameters = {
node_names_mapping[old_input_name]: old_signature.inputs_to_parameters[
old_input_name
]
for old_input_name in old_signature.inputs_to_parameters.keys()
}
new_signature.inputs_to_buffers = {
node_names_mapping[old_input_name]: old_signature.inputs_to_buffers[
old_input_name
]
for old_input_name in old_signature.inputs_to_buffers.keys()
}
new_signature.buffers_to_mutate = {
node_names_mapping[old_output_name]: old_signature.buffers_to_mutate[
old_output_name
]
for old_output_name in old_signature.buffers_to_mutate.keys()
}
return new_signature
new_graph_signature = _generate_new_graph_signature(self, transformed_gm)
transformed_ep = ExportedProgram(
transformed_gm,
transformed_gm.graph,
new_graph_signature,
copy.deepcopy(self.call_spec),
self.state_dict,
_get_updated_range_constraints(transformed_gm),
copy.deepcopy(self.equality_constraints),
copy.deepcopy(self._module_call_graph),
self.example_inputs,
)
transformed_ep.graph_module.meta.update(self.graph_module.meta)
transformed_ep.graph_module.meta.update(res.graph_module.meta)
return transformed_ep
def _check_input_constraints(self, *args):
from torch._export.passes.add_runtime_assertions_for_constraints_pass import (
_AddRuntimeAssertionsForConstraintsPass,
)
# TODO(zhxchen17) Don't generate a runtime graph on the fly.
_assertion_graph = torch.fx.GraphModule({}, torch.fx.Graph())
for p in self.graph.nodes:
if p.op != "placeholder":
continue
new_p = _assertion_graph.graph.placeholder(p.name)
new_p.meta = p.meta
_assertion_graph.graph.output(())
_assertion_graph_res = _AddRuntimeAssertionsForConstraintsPass(
self.range_constraints,
self.equality_constraints,
)(_assertion_graph)
assert _assertion_graph_res is not None
_assertion_graph = _assertion_graph_res.graph_module
_assertion_graph(*args)
def _validate(self):
# TODO(zhxchen17) check for get_attr
# TODO(zhxchen17) check for funcitonal ops
for gm in self.graph_module.modules():
if not isinstance(gm, torch.fx.GraphModule):
continue
for node in gm.graph.nodes:
if node.op == "call_function":
assert node.target != torch.ops.higher_order._export_tracepoint
@dataclasses.dataclass
class _ConstraintTarget:
"""
This represents input tensor dimensions. Don't create this
class directly; instead, use :func:`dynamic_dim`.
"""
w_tensor: Any # weakref to torch.Tensor
# TODO: We don't need t_id; we can get it off of w_tensor
t_id: int
dim: int
class _ConstraintFactory(type):
"""
Metaclass that ensures a private constructor for :class:`Constraint`
"""
def __call__(cls, *args, **kwargs):
raise TypeError(
f"{cls.__module__}.{cls.__qualname__} has no public constructor. "
f"Please use torch.export.dynamic_dim() to create one"
)
def _create(cls, w_tensor, t_id, dim, constraint_range, shared=None):
return super().__call__(w_tensor, t_id, dim, constraint_range, shared)
def _create_constraint(w_tensor, t_id, dim, constraint_range, shared=None):
return Constraint._create(w_tensor, t_id, dim, constraint_range, shared)
[docs]@dataclasses.dataclass
class Constraint(_ConstraintTarget, metaclass=_ConstraintFactory):
"""
.. warning::
Do not construct :class:`Constraint` directly, use :func:`dynamic_dim` instead.
This represents constraints on input tensor dimensions, e.g., requiring
them to be fully polymorphic or within some range.
"""
# NOTE(avik): In the future, this could be Union[StrictMinMaxConstraint, <other kinds>]
constraint_range: StrictMinMaxConstraint
# Represent that `constraint_range` is shared with another _ConstraintTarget, which
# typically arises because of a specified equality with another dynamic dimension.
shared: Optional[_ConstraintTarget] = None
def _clone_with_range(self, lower=2, upper=sympy.oo):
from torch.utils._sympy.value_ranges import ValueRanges
constraint_range = StrictMinMaxConstraint(
vr=self.constraint_range.vr & ValueRanges(lower=lower, upper=upper),
warn_only=False,
)
return _create_constraint(
self.w_tensor, self.t_id, self.dim, constraint_range, self.shared
)
def __ge__(self, lower):
return self._clone_with_range(lower=lower)
def __gt__(self, lower):
return self._clone_with_range(lower=lower + 1)
def __le__(self, upper):
return self._clone_with_range(upper=upper)
def __lt__(self, upper):
return self._clone_with_range(upper=upper - 1)
def __bool__(self):
# NOTE(avik): We do not support compound expressions like a <= x <= b.
# This is because Python implicitly desugars them into bool(a <= x) and bool(x <= b),
# and moreover, enforces that any overload of __bool__ must return True or False.
# FWIW, sympy also raises TypeError in this case.
raise TypeError(
"Cannot determine truth value of Constraint. "
"If you are trying to combine Constraint's with logical connectives, "
"you can specify them separately instead."
)
@property
def serializable_spec(self):
# We need a serialization compatible format of the constraint so that it
# can be savedin the graph module w/o breaking the module serialization.
# The saved constraints will be used directly for the post-exporting pass
# that converts constraints to runtime assertion. The saved constraints
# will not be saved in the serialized module.
# TODO: A better way is needed. Currently we use 't_id' to map the constraint,
# which is not reliable
return {
"t_id": self.t_id,
"dim": self.dim,
"min": self.constraint_range.vr.lower,
"max": self.constraint_range.vr.upper,
"shared": (
None
if self.shared is None
else {
"t_id": self.shared.t_id,
"dim": self.shared.dim,
}
),
}
def __eq__(self, other):
if not isinstance(other, Constraint):
raise TypeError(
"A dynamic dim can be specified equal only to another dynamic dim. "
f"Equality with {type(other)} is not supported."
)
constraint_range = StrictMinMaxConstraint(
vr=self.constraint_range.vr & other.constraint_range.vr,
warn_only=False,
)
return _create_constraint(
self.w_tensor,
self.t_id,
self.dim,
constraint_range,
shared=_ConstraintTarget(other.w_tensor, other.t_id, other.dim),
)
[docs]def constrain_as_value(symbol, min: Optional[int] = None, max: Optional[int] = None):
"""
Hint :func:`export` about the constraint of an intermediate scalar value so that subsequent
branching behaviors that check on the range of aforementioned scalar value can be
soundly traced.
.. warning::
(Note that if the intermediate scalar value will be used like a size, including
being passed as size arg to a tensor factory or view, call :func:`constrain_as_size`
instead.)
Args:
symbol: Intermediate scalar value (int-only now) to apply range constraint on.
min (Optional[int]): Minimum possible value of given symbol (inclusive)
max (Optional[int]): Maximum possible value of given symbol (inclusive)
Returns:
None
For example, following program can not be traced soundly::
def fn(x):
v = x.max().item()
if v > 1024:
return x
else:
return x * 2
``v`` is a data-dependent value, which is assumed to have a range of (-inf, inf).
:func:`export()` a hint about which branch to take would not be able to determine
if the traced branching decision is correct or not. Thus :func:`export()`
would give following error::
torch._dynamo.exc.UserError: Consider annotating your code using
torch.export.constrain_as_size() or torch.export().constrain_as_value() APIs.
It appears that you're trying to get a value out of symbolic int/float whose value
is data-dependent (and thus we do not know the true value.) The expression we were
trying to evaluate is f0 > 1024 (unhinted: f0 > 1024).
Assuming the actual range of ``v`` can be between [10, 200], you can add a call to
:func:`constrain_as_value` in the source code like this::
def fn(x):
v = x.max().item()
# Give export() a hint
torch.export.constrain_as_value(v, min=10, max=200)
if v > 1024:
return x
else:
return x * 2
With the additional hint, :func:`export` would be able to trace the program correctly by taking
the ``else`` branch, resulting in following graph::
graph():
%arg0_1 := placeholder[target=arg0_1]
# v = x.max().item()
%max_1 := call_function[target=torch.ops.aten.max.default](args = (%arg0_1,))
%_local_scalar_dense := call_function[target=torch.ops.aten._local_scalar_dense.default](args = (%max_1,))
# Asserting 10 <= v <= 200
%ge := call_function[target=operator.ge](args = (%_local_scalar_dense, 10))
%scalar_tensor := call_function[target=torch.ops.aten.scalar_tensor.default](args = (%ge,))
%_assert_async := call_function[target=torch.ops.aten._assert_async.msg](
args = (%scalar_tensor, _local_scalar_dense is outside of inline constraint [10, 200].))
%le := call_function[target=operator.le](args = (%_local_scalar_dense, 200))
%scalar_tensor_1 := call_function[target=torch.ops.aten.scalar_tensor.default](args = (%le,))
%_assert_async_1 := call_function[target=torch.ops.aten._assert_async.msg](
args = (%scalar_tensor_1, _local_scalar_dense is outside of inline constraint [10, 200].))
%sym_constrain_range := call_function[target=torch.ops.aten.sym_constrain_range.default](
args = (%_local_scalar_dense,), kwargs = {min: 10, max: 200})
# Always taking `else` branch to multiply elements `x` by 2 due to hints above
%mul := call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, 2), kwargs = {})
return (mul,)
"""
from torch._export.constraints import constrain_as_value
return constrain_as_value(symbol, min, max)
[docs]def constrain_as_size(symbol, min: Optional[int] = None, max: Optional[int] = None):
"""
Hint :func:`export` about the constraint of an intermediate scalar value that
represents shape of a tensor so that subsequent tensor constructors can be
traced correctly because many operators need to make assumption about range
of sizes.
Args:
symbol: Intermediate scalar value (int-only now) to apply range constraint on.
min (Optional[int]): Minimum possible value of given symbol (inclusive)
max (Optional[int]): Maximum possible value of given symbol (inclusive)
Returns:
None
For example, following program can not be traced soundly wihout using
:func:`constrain_as_size` to give :func:`export` a hint about shape ranges::
def fn(x):
d = x.max().item()
return torch.ones(v)
:func:`export` would give following error::
torch._dynamo.exc.Unsupported: guard on data-dependent symbolic int/float
Assuming the actual range of ``d`` can be between [3, 10], you can add a call to
:func:`constrain_as_size` in the source code like this::
def fn(x):
d = x.max().item()
torch.export.constrain_as_size(d, min=3, max=10)
return torch.ones(d)
With the additional hint, :func:`export` would be able to trace the program correctly by taking
the ``else`` branch, resulting in following graph::
graph():
%arg0_1 := placeholder[target=arg0_1]
# d = x.max().item()
%max_1 := call_function[target=torch.ops.aten.max.default](args = (%arg0_1,))
%_local_scalar_dense := call_function[target=torch.ops.aten._local_scalar_dense.default](args = (%max_1,))
# Asserting 3 <= d <= 10
%ge := call_function[target=operator.ge](args = (%_local_scalar_dense, 3))
%scalar_tensor := call_function[target=torch.ops.aten.scalar_tensor.default](args = (%ge,))
%_assert_async := call_function[target=torch.ops.aten._assert_async.msg](
args = (%scalar_tensor, _local_scalar_dense is outside of inline constraint [3, 10].))
%le := call_function[target=operator.le](args = (%_local_scalar_dense, 10))
%scalar_tensor_1 := call_function[target=torch.ops.aten.scalar_tensor.default](args = (%le,))
%_assert_async_1 := call_function[target=torch.ops.aten._assert_async.msg](
args = (%scalar_tensor_1, _local_scalar_dense is outside of inline constraint [3, 10].))
%sym_constrain_range_for_size := call_function[target=torch.ops.aten.sym_constrain_range_for_size.default](
args = (%_local_scalar_dense,), kwargs = {min: 3, max: 10})
# Constructing new tensor with d
%full := call_function[target=torch.ops.aten.full.default](
args = ([%_local_scalar_dense], 1),
kwargs = {dtype: torch.float32, layout: torch.strided, device: cpu, pin_memory: False})
......
.. warning::
if your size is intended to be dynamic, do NOT test if sizes are equal to 0 or 1,
these will SILENTLY report false and be bypassed
"""
from torch._export.constraints import constrain_as_size
return constrain_as_size(symbol, min, max)
[docs]def dynamic_dim(t: torch.Tensor, index: int):
"""
:func:`dynamic_dim` constructs a :class:`Constraint` object that describes the dynamism of
a dimension ``index`` of tensor ``t``. :class:`Constraint` objects should be passed to
``constraints`` argument of :func:`export`.
Args:
t (torch.Tensor): Example input tensor that have dynamic dimension size(s)
index (int): Index of dynamic dimension
Returns:
A :class:`Constraint` object that describes shape dynamism. It can be passed to :func:`export` so
that :func:`export` does not assume static size of specified tensor, i.e. keeping it dynamic
as a symbolic size rather than specializing according to size of example tracing input.
Specifically :func:`dynamic_dim` can be used to express following types of dynamism.
- Size of a dimension is dynamic and unbounded::
t0 = torch.rand(2, 3)
t1 = torch.rand(3, 4)
# First dimension of t0 can be dynamic size rather than always being static size 2
constraints = [dynamic_dim(t0, 0)]
ep = export(fn, (t0, t1), constraints=constraints)
- Size of a dimension is dynamic with a lower bound::
t0 = torch.rand(10, 3)
t1 = torch.rand(3, 4)
# First dimension of t0 can be dynamic size with a lower bound of 5 (inclusive)
# Second dimension of t1 can be dynamic size with a lower bound of 2 (exclusive)
constraints = [
dynamic_dim(t0, 0) >= 5,
dynamic_dim(t1, 1) > 2,
]
ep = export(fn, (t0, t1), constraints=constraints)
- Size of a dimension is dynamic with an upper bound::
t0 = torch.rand(10, 3)
t1 = torch.rand(3, 4)
# First dimension of t0 can be dynamic size with a upper bound of 16 (inclusive)
# Second dimension of t1 can be dynamic size with a upper bound of 8 (exclusive)
constraints = [
dynamic_dim(t0, 0) <= 16,
dynamic_dim(t1, 1) < 8,
]
ep = export(fn, (t0, t1), constraints=constraints)
- Size of a dimension is dynamic and it is always equal to size of another dynamic dimension::
t0 = torch.rand(10, 3)
t1 = torch.rand(3, 4)
# Sizes of second dimension of t0 and first dimension are always equal
constraints = [
dynamic_dim(t0, 1) == dynamic_dim(t1, 0),
]
ep = export(fn, (t0, t1), constraints=constraints)
- Mix and match all types above as long as they do not express conflicting requirements
"""
from torch._export import dynamic_dim
return dynamic_dim(t, index)
[docs]def export(
f: Callable,
args: Tuple[Any, ...],
kwargs: Optional[Dict[str, Any]] = None,
*,
constraints: Optional[List[Constraint]] = None,
) -> ExportedProgram:
"""
:func:`export` takes an arbitrary Python callable (an nn.Module, a function or
a method) and produces a traced graph representing only the Tensor
computation of the function in an Ahead-of-Time (AOT) fashion, which can
subsequently be executed with different outputs or serialized. The traced
graph (1) produces a normalized operator set consisting only of functional
`Core ATen Operator Set <https://pytorch.org/docs/stable/ir.html>`_
and user specified custom operators, (2) has eliminated all Python control
flow and data structures (except for certain
conditions), and (3) has the set of shape constraints needed to show that
this normalization and control flow elimination is sound for a future
input.
**Soundness Guarantee**
While tracing, :func:`export()` takes note of shape-related assumptions
made by the user program and the underlying PyTorch operator kernels.
The output :class:`ExportedProgram` is considered valid only when these
assumptions hold true.
There are 2 types of assumptions made during tracing
- Shapes (not values) of input tensors.
- Ranges (lower and upper bound) of values extracted from intermediate tensors via ``.item()`` or direct indexing.
All assumptions must be validated at graph capture time for :func:`export`
to succeed. Specifically:
- Assumptions on static shapes of input tensors are automatically validated without additional effort.
- Assumptions on dynamic shape of input tensors require explicit `Input Constraint`
constructed with :func:`dynamic_dim` APIs
- Assumptions on range of intermediate values require explicit `Inline Constraint`,
constructed use :func:`constrain_as_size` and :func:`constraint_as_value` APIs.
If any assumption can not be validated, a fatal error will be raised. When that happens,
the error message will include suggested code needed to construct necessary
constraints to validate the assumptions, for example :func:`export` would suggest
following code for input constraints::
def specify_constraints(x):
return [
# x:
dynamic_dim(x, 0) <= 5,
]
This example means the program requires the dim 0 of input ``x`` to be less
than or equal to 5 to be valid. You can inspect the constraints needed and
then copy this exact function into your code to generated needed
constraints to be passed into ``constraints`` argument.
Args:
f: The callable to trace.
args: Example positional inputs.
kwargs: Optional example keyword inputs.
constraints: An optional list of constraints on the dynamic arguments
that specify their possible range of shapes. By default, shapes of
input torch.Tensors are assumed to be static. If an input torch.Tensor
is expected to have dynamic shapes, please use :func:`dynamic_dim`
to define :class:`Constraint` objects that specify the dynamics and the possible
range of shapes. See :func:`dynamic_dim` docstring for examples on
how to use it.
Returns:
An :class:`ExportedProgram` containing the traced callable.
**Acceptable input/output types**
Acceptable types of inputs (for ``args`` and ``kwargs``) and outputs include:
- Primitive types, i.e. ``torch.Tensor``, ``int``, ``float``, ``bool`` and ``str``.
- (Nested) Data structures comprising of ``dict``, ``list``, ``tuple``, ``namedtuple`` and
``OrderedDict`` containing all above types.
"""
from torch._export import export
return export(f, args, kwargs, constraints)
[docs]def save(
ep: ExportedProgram,
f: Union[str, pathlib.Path, io.BytesIO],
*,
extra_files: Optional[Dict[str, Any]] = None,
opset_version: Optional[Dict[str, int]] = None,
) -> None:
"""
.. warning::
Under active development, saved files may not be usable in newer versions
of PyTorch.
Saves an :class:`ExportedProgram` to a file-like object. It can then be
loaded using the Python API :func:`torch.export.load <torch.export.load>`.
Args:
ep (ExportedProgram): The exported program to save.
f (Union[str, pathlib.Path, io.BytesIO): A file-like object (has to
implement write and flush) or a string containing a file name.
extra_files (Optional[Dict[str, Any]]): Map from filename to contents
which will be stored as part of f.
opset_version (Optional[Dict[str, int]]): A map of opset names
to the version of this opset
Example::
import torch
import io
class MyModule(torch.nn.Module):
def forward(self, x):
return x + 10
ep = torch.export.export(MyModule(), torch.randn(5))
# Save to file
torch.export.save(ep, 'exported_program.pt2')
# Save to io.BytesIO buffer
buffer = io.BytesIO()
torch.export.save(ep, buffer)
# Save with extra files
extra_files = {'foo.txt': b'bar'}
torch.export.save(ep, 'exported_program.pt2', extra_files=extra_files)
"""
from torch._export import save
save(ep, f, extra_files=extra_files, opset_version=opset_version)
[docs]def load(
f: Union[str, pathlib.Path, io.BytesIO],
*,
extra_files: Optional[Dict[str, Any]] = None,
expected_opset_version: Optional[Dict[str, int]] = None,
) -> ExportedProgram:
"""
.. warning::
Under active development, saved files may not be usable in newer versions
of PyTorch.
Loads an :class:`ExportedProgram` previously saved with
:func:`torch.export.save <torch.export.save>`.
Args:
ep (ExportedProgram): The exported program to save.
f (Union[str, pathlib.Path, io.BytesIO): A file-like object (has to
implement write and flush) or a string containing a file name.
extra_files (Optional[Dict[str, Any]]): The extra filenames given in
this map would be loaded and their content would be stored in the
provided map.
expected_opset_version (Optional[Dict[str, int]]): A map of opset names
to expected opset versions
Returns:
An :class:`ExportedProgram` object
Example::
import torch
import io
# Load ExportedProgram from file
ep = torch.export.load('exported_program.pt2')
# Load ExportedProgram from io.BytesIO object
with open('exported_program.pt2', 'rb') as f:
buffer = io.BytesIO(f.read())
buffer.seek(0)
ep = torch.export.load(buffer)
# Load with extra files.
extra_files = {'foo.txt': ''} # values will be replaced with data
ep = torch.export.load('exported_program.pt2', extra_files=extra_files)
print(extra_files['foo.txt'])
"""
from torch._export import load
return load(
f, extra_files=extra_files, expected_opset_version=expected_opset_version
)
