Torch Script ============ .. contents:: :local: .. automodule:: torch.jit .. currentmodule:: torch.jit Torch Script is a way to create serializable and optimizable models from PyTorch code. Any code written in Torch Script can be saved from your Python process and loaded in a process where there is no Python dependency. We provide tools to incrementally transition a model from being a pure Python program to a Torch Script program that can be run independently from Python, for instance, in a standalone C++ program. This makes it possible to train models in PyTorch using familiar tools and then export the model to a production environment where it is not a good idea to run models as Python programs for performance and multi-threading reasons. Creating Torch Script Code -------------------------- .. autoclass:: ScriptModule :members: .. method:: save(filename) Save an offline version of this module for use in a separate process. The saved module serializes all of the methods and parameters of this module. It can be loaded into the C++ API using ``torch::jit::load(filename)`` or into the Python API with ``torch.jit.load(filename)``. To be able to save a module, it must not make any calls to native python functions. This means that all submodules must be subclasses of ScriptModules as well. .. DANGER:: All modules, no matter their device, are always loaded onto the CPU during loading. This is different from :func:`torch.load`'s semantics and may change in the future. .. autofunction:: load .. autofunction:: trace Mixing Tracing and Scripting ---------------------------- In many cases either tracing or script is an easier approach for converting a model. We allow you to compose tracing and scripting to suit the particular requirements of a part of a model. Scripted functions can call traced ones. This is particularly useful when you need to use control-flow around a simple feed-forward model. For instance the beam search of a sequence to sequence model will typically be written in script but can call an encoder module generated using tracing. Example: :: import torch def foo(x, y): return 2 * x + y traced_foo = torch.jit.trace(foo, (torch.rand(3), torch.rand(3))) @torch.jit.script def bar(x): return traced_foo(x, x) Traced functions can call script functions. This is useful when a small part of a model requires some control-flow even though most of the model is just a feed-forward network. Control-flow inside of a script function called by a traced function is preserved correctly: Example: :: import torch @torch.jit.script def foo(x, y): if x.max() > y.max(): r = x else: r = y return r def bar(x, y, z): return foo(x, y) + z traced_bar = torch.jit.trace(bar, (torch.rand(3), torch.rand(3), torch.rand(3)) This composition also works for modules as well, where it can be used to generate a submodule using tracing that can be called from the methods of a script module: Example: :: import torch import torchvision class MyScriptModule(torch.jit.ScriptModule): def __init__(self): super(MyScriptModule, self).__init__() self.means = torch.nn.Parameter(torch.tensor([103.939, 116.779, 123.68]) .resize_(1, 3, 1, 1)) self.resnet = torch.jit.trace(torchvision.models.resnet18(), torch.rand(1, 3, 224, 224)) @torch.jit.script_method def forward(self, input): return self.resnet(input - self.means) Torch Script Language Reference ------------------------------- Torch Script is a subset of Python that can either be written directly (using the @script annotations) or generated automatically from Python code via tracing. When using tracing, code is automatically converted into this subset of Python by recording only the actual operators on tensors and simply executing and discarding the other surrounding Python code. When writing Torch Script directly using @script annotations, the programmer must only use the subset of Python supported in Torch Script. This section documents what is supported in Torch Script as if it were a language reference for a stand alone language. Any features of Python not mentioned in this reference are not part of Torch Script. As a subset of Python any valid Torch Script function is also a valid Python function. This makes it possible to remove the @script annotations and debug the function using standard Python tools like pdb. The reverse is not true: there are many valid python programs that are not valid Torch Script programs. Instead, Torch Script focuses specifically on the features of Python that are needed to represent neural network models in Torch. .. envvar:: PYTORCH_JIT=1 Setting the environment variable ``PYTORCH_JIT=0`` will disable all script and tracing annotations. If there is hard-to-debug error in one of your ScriptModules, you can use this flag to force everything to run using native Python. This allows the use of tools like ``pdb`` to debug code. Types ~~~~~ The largest difference between Torch Script and the full Python language is that Torch Script only support a small set of types that are needed to express neural net models. In particular Torch Script supports: ``Tensor`` A PyTorch tensor of any dtype, dimension, or backend. ``Tuple[T0, T1, ...]`` A tuple containing subtypes ``T0``, ``T1``, etc. (e.g. ``Tuple[Tensor, Tensor]``) ``int`` A scalar integer ``float`` A scalar floating point number ``List[T]`` A list of which all members are type ``T`` Unlike Python, each variable in Torch Script function must have a single static type. This makes it easier to optimize Torch Script functions. Example:: @torch.jit.script def an_error(x): if x: r = torch.rand(1) else: r = 4 return r # Type mismatch: r is set to type Tensor in the true branch # and type int in the false branch By default, all parameters to a Torch Script function are assumed to be Tensor because this is the most common type used in modules. To specify that an argument to a Torch Script function is another type, it is possible to use MyPy-style type annotations using the types listed above: Example:: @torch.jit.script def foo(x, tup): # type: (int, Tuple[Tensor, Tensor]) -> Tensor t0, t1 = tup return t0 + t1 + x print(foo(3, (torch.rand(3), torch.rand(3)))) .. note:: It is also possible to annotate types with Python 3 type annotations. In our examples, we use comment-based annotations to ensure Python 2 compatibility as well. Expressions ~~~~~~~~~~~ The following Python Expressions are supported Literals ``True``, ``False``, ``None``, ``'string literals'``, ``"string literals"``, number literals ``3`` (interpreted as int) ``3.4`` (interpreter as a float) Variables ``a`` .. note:: See `Variable Resolution`_ for how variables are resolved. Tuple Construction ``(3, 4)``, ``(3,)`` List Construction ``[3, 4]``, ``[]``, ``[torch.rand(3), torch.rand(4)]`` .. note:: an empty list is assumed have type ``List[Tensor]``. The types of other list literals are derived from the type of the members. Arithmetic Operators ``a + b`` ``a - b`` ``a * b`` ``a / b`` ``a ^ b`` ``a @ b`` Comparison Operators ``a == b`` ``a != b`` ``a < b`` ``a > b`` ``a <= b`` ``a >= b`` Logical Operators ``a and b`` ``a or b`` ``not b`` Subscripts ``t[0]`` ``t[-1]`` ``t[0:2]`` ``t[1:]`` ``t[:1]`` ``t[:]`` ``t[0, 1]`` ``t[0, 1:2]`` ``t[0, :1]`` ``t[-1, 1:, 0]`` ``t[1:, -1, 0]`` ``t[i:j, i]`` .. note:: Torch Script currently does not support mutating tensors in place, so any tensor indexing can only appear on the right-hand size of an expression. Function calls Calls to built-in functions: ``torch.rand(3, dtype=torch.int)`` Calls to other script functions: :: import torch @torch.jit.script def foo(x): return x + 1 @torch.jit.script def bar(x): return foo(x) Method calls Calls to methods of builtin types like tensor: ``x.mm(y)`` When defining a Script method inside of a ScriptModule, the ``@script_method`` annotation is used. Inside of these methods it is possible to call other methods of this class or access methods on the submodules. Calling a submodule directly (e.g. ``self.resnet(input)``) is equivalent to calling its ``forward`` method (e.g. ``self.resnet.forward(input)``) :: import torch class MyScriptModule(torch.jit.ScriptModule): def __init__(self): super(MyScriptModule, self).__init__() self.means = torch.nn.Parameter(torch.tensor([103.939, 116.779, 123.68]) .resize_(1, 3, 1, 1)) self.resnet = torch.jit.trace(torchvision.models.resnet18(), torch.rand(1, 3, 224, 224)) @torch.jit.script_method def helper(self, input): return self.resnet(input - self.means) @torch.jit.script_method def forward(self, input): return self.helper(input) If expressions ``x if x > y else y`` Casts ``float(ten)``, ``int(3.5)``, ``bool(ten)`` Accessing Module Parameters ``self.my_parameter`` ``self.my_submodule.my_parameter`` Statements ~~~~~~~~~~ Torch Script supports the following types of statements: Simple Assignments :: a = b a += b # short-hand for a = a + b, does not operate in-place on a a -= b Pattern Matching Assignments :: a, b = tuple_or_list a, b, *c = a_tuple Print Statements ``print("the result of an add:", a + b)`` If Statements :: if a < 4: r = -a elif a < 3: r = a + a else: r = 3 * a While Loops :: a = 0 while a < 4: print(a) a += 1 For loops with ``range`` :: x = 0 for i in range(10): x *= i .. note:: Script currently does not support iterating over generic iterable objects like lists or tensors. Script currently does not support start or increment parameters to range. These will be added in a future version. For loops over tuples: :: tup = (3, torch.rand(4)) for x in tup: print(x) .. note:: for loops over tuples will unroll the loop, generating a body for each member of the tuple. The body must type-check correctly for each member. For loops over constant ``torch.nn.ModuleList`` :: class SubModule(torch.jit.ScriptModule): def __init__(self): super(Sub, self).__init__() self.weight = nn.Parameter(torch.randn(2)) @torch.jit.script_method def forward(self, input): return self.weight + input class MyModule(torch.jit.ScriptModule): __constants__ = ['mods'] def __init__(self): super(MyModule, self).__init__() self.mods = torch.nn.ModuleList([SubModule() for i in range(10)]) @torch.jit.script_method def forward(self, v): for module in self.mods: v = m(v) return v .. note:: To use a module list inside a ``@script_method`` it must be marked constant by adding the name of the attribute to the ``__constants__`` list for the type. For loops over a ModuleList will unroll the body of the loop at compile time, with each member of the constant module list. Return ``return a, b`` .. note:: there must be a return statement as the last member of the function and return statements cannot appear anywhere else in the function. This restriction will be removed in the future. Variable Resolution ~~~~~~~~~~~~~~~~~~~ Torch Script supports a subset of Python's variable resolution (i.e. scoping) rules. Local variables behave the same as in Python, except for the restriction that a variable must have the same type along all paths through a function. If a variable has a different type on different sides of an if statement, it is an error to use it after the end of the if statement. Similarly, a variable is not allowed to be used if it is only *defined* along some paths through the function. Example:: @torch.jit.script def foo(x): if x < 0: y = 4 print(y) # Error: undefined value y Non-local variables are resolved to Python values at compile time when the function is defined. These values are then converted into Torch Script values using the rules described in `Use of Python Values`_. Use of Python Values ~~~~~~~~~~~~~~~~~~~~ To make writing Torch Script more convenient, we allow script code to refer to Python values in the surrounding scope. For instance, any time there is a reference to ``torch``, the Torch Script compiler is actually resolving it to the ``torch`` Python module when the function is declared. These Python values are not a first class part of Torch Script. Instead they are desugared at compile-time into the primitive types that Torch Script supports. This section describes the rules that are used when accessing Python values in Torch Script. They depend on the dynamic type of the python valued referenced. Functions Torch Script can call python functions. This functionality is very useful when incrementally converting a model into script. The model can be moved function-by-function to script, leaving calls to Python functions in place. This way you can incrementally check the correctness of the model as you go. Example:: def foo(x): print("I am called with {}".format(x)) import pdb; pdb.set_trace() return x @torch.jit.script def bar(x) return foo(x + 1) .. note:: Attempting to call ``save`` on a ScriptModule that contains calls to Python functions will fail. The intention is that this pathway is used for debugging and the calls removed or turned into script functions before saving. Attribute Lookup On Python Modules Torch Script can lookup attributes on modules. Builtin functions like ``torch.add`` are accessed this way. This allows Torch Script to call functions defined in other modules. Python-defined Constants Torch Script also provides a way to use constants that are defined in Python. These can be used to hard-code hyper-parameters into the function, or to define universal constants. There are two ways of specifying that a Python value should be treated as a constant. 1. Values looked up as attributes of a module are assumed to be constant. Example: ``math.pi`` 2. Attributes of a ScriptModule can be marked constant by listing them as a member of the ``__constants__`` property of the class: Example:: class Foo(torch.jit.ScriptModule): __constants__ = ['a'] def __init__(self): super(Foo, self).__init__(False) self.a = 1 + 4 @torch.jit.ScriptModule def forward(self, input): return self.a + input Supported constant Python Values are * ``int`` * ``bool`` * ``torch.device`` * ``torch.layout`` * ``torch.dtype`` * tuples containing supported types * ``torch.nn.ModuleList`` which can be used in a TorchScript for loop Debugging ~~~~~~~~~ Disable JIT for Debugging If you want to disable all JIT modes (tracing and scripting) so you can debug your program in raw Python, you can use the ``PYTORCH_JIT`` environment variable. ``PYTORCH_JIT`` can be used to globally disable the JIT by setting its value to ``0``. Given an example script:: @torch.jit.script def scripted_fn(x : torch.Tensor): for i in range(12): x = x + x return x def fn(x): x = torch.neg(x) import pdb; pdb.set_trace() return scripted_fn(x) traced_fn = torch.jit.trace(fn, (torch.rand(4, 5),)) traced_fn(torch.rand(3, 4)) Debugging this script with PDB works except for when we invoke the @script function. We can globally disable JIT, so that we can call the @script function as a normal python function and not compile it. If the above script is called ``disable_jit_example.py``, we can invoke it like so:: $ PYTORCH_JIT=0 python disable_jit_example.py and we will be able to step into the @script function as a normal Python function. Interpreting Graphs TorchScript uses a static single assignment (SSA) intermediate representation (IR) to represent computation. The instructions in this format consist of ATen (the C++ backend of PyTorch) operators and other primitive operators, including control flow operators for loops and conditionals. As an example:: @torch.jit.script def foo(len): # type: (int) -> torch.Tensor rv = torch.zeros(3, 4) for i in range(len): if i < 10: rv = rv - 1.0 else: rv = rv + 1.0 return rv print(foo.graph) A ``ScriptModule`` with a single ``forward`` method will have an attribute ``graph``, which you can use to inspect the IR representing the computation. If the ScriptModule has more than one method, you will need to access ``.graph`` on the method itself and not the module. We can inspect the graph of a method named ``bar`` on a ScriptModule by accessing ``.bar.graph``. The example script above produces the graph:: graph(%len : int) { %13 : float = prim::Constant[value=1]() %10 : int = prim::Constant[value=10]() %2 : int = prim::Constant[value=4]() %1 : int = prim::Constant[value=3]() %3 : int[] = prim::ListConstruct(%1, %2) %4 : int = prim::Constant[value=6]() %5 : int = prim::Constant[value=0]() %6 : int[] = prim::Constant[value=[0, -1]]() %rv.1 : Dynamic = aten::zeros(%3, %4, %5, %6) %8 : int = prim::Constant[value=1]() %rv : Dynamic = prim::Loop(%len, %8, %rv.1) block0(%i : int, %12 : Dynamic) { %11 : int = aten::lt(%i, %10) %rv.4 : Dynamic = prim::If(%11) block0() { %14 : int = prim::Constant[value=1]() %rv.2 : Dynamic = aten::sub(%12, %13, %14) -> (%rv.2) } block1() { %16 : int = prim::Constant[value=1]() %rv.3 : Dynamic = aten::add(%12, %13, %16) -> (%rv.3) } %19 : int = prim::Constant[value=1]() -> (%19, %rv.4) } return (%rv); } Take the instruction ``%rv.1 : Dynamic = aten::zeros(%3, %4, %5, %6)`` for example. ``%rv.1 : Dynamic`` means we assign the output to a (unique) value named ``rv.1``, and that value is of ``Dynamic`` type, i.e. we do not know its concrete shape. ``aten::zeros`` is the operator (equivalent to ``torch.zeros``) and the input list ``(%3, %4, %5, %6)`` specifies which values in scope should be passed as inputs. The schema for built-in functions like ``aten::zeros`` can be found at `Builtin Functions`_. Notice that operators can also have associated ``blocks``, namely the ``prim::Loop`` and ``prim::If`` operators. In the graph print-out, these operators are formatted to reflect their equivalent source code forms to facilitate easy debugging. Graphs can be inspected as shown to confirm that the computation described by a ``ScriptModule`` is correct, in both automated and manual fashion, as described below. Tracing Edge Cases There are some edge cases that exist where the trace of a given Python function/module will not be representative of the underlying code. These cases can include: * Tracing of control flow that is dependent on inputs (e.g. tensor shapes) * Tracing of in-place operations of tensor views (e.g. indexing on the left-hand side of an assignment) Note that these cases may in fact be traceable in the future. Automatic Trace Checking One way to automatically catch many errors in traces is by using ``check_inputs`` on the ``torch.jit.trace()`` API. ``check_inputs`` takes a list of tuples of inputs that will be used to re-trace the computation and verify the results. For example:: def loop_in_traced_fn(x): result = x[0] for i in range(x.size(0)): result = result * x[i] return result inputs = (torch.rand(3, 4, 5),) check_inputs = [(torch.rand(4, 5, 6),), (torch.rand(2, 3, 4),)] traced = torch.jit.trace(loop_in_traced_fn, inputs, check_inputs=check_inputs) Gives us the following diagnostic information:: ERROR: Graphs differed across invocations! Graph diff: graph(%0 : Dynamic) { %1 : int = prim::Constant[value=0]() %2 : int = prim::Constant[value=0]() %3 : Dynamic = aten::select(%0, %1, %2) %4 : int = prim::Constant[value=0]() %5 : int = prim::Constant[value=0]() %6 : Dynamic = aten::select(%0, %4, %5) %7 : Dynamic = aten::mul(%3, %6) %8 : int = prim::Constant[value=0]() %9 : int = prim::Constant[value=1]() %10 : Dynamic = aten::select(%0, %8, %9) %11 : Dynamic = aten::mul(%7, %10) %12 : int = prim::Constant[value=0]() %13 : int = prim::Constant[value=2]() %14 : Dynamic = aten::select(%0, %12, %13) %15 : Dynamic = aten::mul(%11, %14) + %16 : int = prim::Constant[value=0]() + %17 : int = prim::Constant[value=3]() + %18 : Dynamic = aten::select(%0, %16, %17) + %19 : Dynamic = aten::mul(%15, %18) - return (%15); ? ^ + return (%19); ? ^ } This message indicates to us that the computation differed between when we first traced it and when we traced it with the ``check_inputs``. Indeed, the loop within the body of ``loop_in_traced_fn`` depends on the shape of the input ``x``, and thus when we try another ``x`` with a different shape, the trace differs. In this case, data-dependent control flow like this can be captured using script instead:: def fn(x): result = x[0] for i in range(x.size(0)): result = result * x[i] return result inputs = (torch.rand(3, 4, 5),) check_inputs = [(torch.rand(4, 5, 6),), (torch.rand(2, 3, 4),)] scripted_fn = torch.jit.script(fn) print(scripted_fn.graph) for input_tuple in [inputs] + check_inputs: torch.testing.assert_allclose(fn(*input_tuple), scripted_fn(*input_tuple)) Which produces:: graph(%x : Dynamic) { %1 : int = prim::Constant[value=0]() %2 : int = prim::Constant[value=0]() %result.1 : Dynamic = aten::select(%x, %2, %1) %4 : int = aten::size(%x, %1) %5 : int = prim::Constant[value=1]() %result : Dynamic = prim::Loop(%4, %5, %result.1) block0(%i : int, %7 : Dynamic) { %9 : int = prim::Constant[value=0]() %10 : Dynamic = aten::select(%x, %9, %i) %result.2 : Dynamic = aten::mul(%7, %10) %12 : int = prim::Constant[value=1]() -> (%12, %result.2) } return (%result); } Tracer Warnings The tracer produces warnings for several problematic patterns in traced computation. As an example, take a trace of a function that contains an in-place assignment on a slice (a view) of a Tensor:: def fill_row_zero(x): x[0] = torch.rand(*x.shape[1:2]) return x traced = torch.jit.trace(fill_row_zero, (torch.rand(3, 4),)) print(traced.graph) Produces several warnings and a graph which simply returns the input:: fill_row_zero.py:4: TracerWarning: There are 2 live references to the data region being modified when tracing in-place operator copy_ (possibly due to an assignment). This might cause the trace to be incorrect, because all other views that also reference this data will not not reflect this change in the trace! On the other hand, if all other views use the same memory chunk, but are disjoint (e.g. are outputs of torch.split), this might still be safe. x[0] = torch.rand(*x.shape[1:2]) fill_row_zero.py:6: TracerWarning: Output nr 1. of the traced function does not match the corresponding output of the Python function. Detailed error: Not within tolerance rtol=1e-05 atol=1e-05 at input[0, 1] (0.09115803241729736 vs. 0.6782537698745728) and 3 other locations (33.00%) traced = torch.jit.trace(fill_row_zero, (torch.rand(3, 4),)) graph(%0 : Float(3, 4)) { return (%0); } We can fix this by modifying the code to not use the in-place update, but rather build up the result tensor out-of-place with `torch.cat`:: def fill_row_zero(x): x = torch.cat((torch.rand(1, *x.shape[1:2]), x[1:2]), dim=0) return x traced = torch.jit.trace(fill_row_zero, (torch.rand(3, 4),)) print(traced.graph) Builtin Functions ~~~~~~~~~~~~~~~~~ Torch Script supports a subset of the builtin tensor and neural network functions that PyTorch provides. Most methods on Tensor as well as functions in the ``torch`` namespace are available. Many functions in ``torch.nn.functional`` are also availiable. We currently do not provide any builtin ScriptModules e.g. a ``Linear`` or ``Conv`` module. This functionality is something that will be developed in the future. For now we suggest using ``torch.jit.trace`` to transform standard ``torch.nn`` modules into ScriptModules on construction. .. automodule:: torch.jit.supported_ops