评价此页
★ ★ ★ ★ ★

常见图中断#

创建时间：2025 年 7 月 28 日 | 最后更新时间：2025 年 7 月 28 日

以下是一些常见的图中断及其解决方法。

错误代码#

您的代码可能包含错误（意味着即使没有 torch.compile 也会导致错误）。在下面的示例中，由于多余的参数，torch.sin 调用中存在拼写错误。始终禁用 torch.compile 以检查代码是否能正确运行。

@torch.compile
def fn(x):
    y = torch.sin(x, x)
    return y

try:
    fn(torch.ones(3, 3))
except Exception as e:
    pass

Graph break in user code at /tmp/ipykernel_581/343837593.py:3
Graph Break Reason: TypeError when making fake tensor call
  Explanation: 


  Developer debug context: TypeError <built-in method sin of type object at 0x7f08e7582260>: sin() takes 1 positional argument but 2 were given

 For more details about this graph break, please visit: https://meta-pytorch.github.io/compile-graph-break-site/gb/gb0112.html
User code traceback:
  File "/opt/conda/envs/py_3.10/lib/python3.10/runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/opt/conda/envs/py_3.10/lib/python3.10/runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel_launcher.py", line 18, in <module>
    app.launch_new_instance()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/traitlets/config/application.py", line 1075, in launch_instance
    app.start()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelapp.py", line 739, in start
    self.io_loop.start()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/tornado/platform/asyncio.py", line 211, in start
    self.asyncio_loop.run_forever()
  File "/opt/conda/envs/py_3.10/lib/python3.10/asyncio/base_events.py", line 603, in run_forever
    self._run_once()
  File "/opt/conda/envs/py_3.10/lib/python3.10/asyncio/base_events.py", line 1909, in _run_once
    handle._run()
  File "/opt/conda/envs/py_3.10/lib/python3.10/asyncio/events.py", line 80, in _run
    self._context.run(self._callback, *self._args)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 519, in dispatch_queue
    await self.process_one()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 508, in process_one
    await dispatch(*args)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 400, in dispatch_shell
    await result
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/ipkernel.py", line 368, in execute_request
    await super().execute_request(stream, ident, parent)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 767, in execute_request
    reply_content = await reply_content
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/ipkernel.py", line 455, in do_execute
    res = shell.run_cell(
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/zmqshell.py", line 577, in run_cell
    return super().run_cell(*args, **kwargs)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3006, in run_cell
    result = self._run_cell(
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3061, in _run_cell
    result = runner(coro)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/async_helpers.py", line 129, in _pseudo_sync_runner
    coro.send(None)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3266, in run_cell_async
    has_raised = await self.run_ast_nodes(code_ast.body, cell_name,
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3445, in run_ast_nodes
    if await self.run_code(code, result, async_=asy):
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3505, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "/tmp/ipykernel_581/343837593.py", line 7, in <module>
    fn(torch.ones(3, 3))
  File "/tmp/ipykernel_581/343837593.py", line 3, in fn
    y = torch.sin(x, x)

Dynamo 会尽力提示图中断是否由您的代码引起。但有时仍然很难从日志中判断图中断是由代码错误、更复杂的图中断还是 torch.compile bug 引起的。为了区分，我们建议尝试在不使用 torch.compile 的情况下运行您的代码，看看是否仍然会收到图中断报告的错误。

依赖数据的操作#

torch.compile 会在依赖数据的操作上发生图中断，例如依赖数据的控制流（if 语句、带有张量的循环）以及直接访问张量数据（.item, .data_ptr）。

@torch.compile
def fn(x):
    y = x.sum()
    if y > 0:
        return x + y.item()
    return x - y.item()

print(fn(torch.ones(3, 3)))

tensor([[10., 10., 10.],
        [10., 10., 10.],
        [10., 10., 10.]])

Graph break in user code at /tmp/ipykernel_581/3495555842.py:4
Graph Break Reason: Data-dependent branching
  Explanation: Detected data-dependent branching (e.g. `if my_tensor.sum() > 0:`). Dynamo does not support tracing dynamic control flow.
  Hint: This graph break is fundamental - it is unlikely that Dynamo will ever be able to trace through your code. Consider finding a workaround.
  Hint: Use `torch.cond` to express dynamic control flow.

  Developer debug context: attempted to jump with TensorVariable()

User code traceback:
  File "/opt/conda/envs/py_3.10/lib/python3.10/runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/opt/conda/envs/py_3.10/lib/python3.10/runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel_launcher.py", line 18, in <module>
    app.launch_new_instance()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/traitlets/config/application.py", line 1075, in launch_instance
    app.start()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelapp.py", line 739, in start
    self.io_loop.start()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/tornado/platform/asyncio.py", line 211, in start
    self.asyncio_loop.run_forever()
  File "/opt/conda/envs/py_3.10/lib/python3.10/asyncio/base_events.py", line 603, in run_forever
    self._run_once()
  File "/opt/conda/envs/py_3.10/lib/python3.10/asyncio/base_events.py", line 1909, in _run_once
    handle._run()
  File "/opt/conda/envs/py_3.10/lib/python3.10/asyncio/events.py", line 80, in _run
    self._context.run(self._callback, *self._args)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 519, in dispatch_queue
    await self.process_one()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 508, in process_one
    await dispatch(*args)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 400, in dispatch_shell
    await result
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/ipkernel.py", line 368, in execute_request
    await super().execute_request(stream, ident, parent)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 767, in execute_request
    reply_content = await reply_content
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/ipkernel.py", line 455, in do_execute
    res = shell.run_cell(
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/zmqshell.py", line 577, in run_cell
    return super().run_cell(*args, **kwargs)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3006, in run_cell
    result = self._run_cell(
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3061, in _run_cell
    result = runner(coro)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/async_helpers.py", line 129, in _pseudo_sync_runner
    coro.send(None)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3266, in run_cell_async
    has_raised = await self.run_ast_nodes(code_ast.body, cell_name,
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3445, in run_ast_nodes
    if await self.run_code(code, result, async_=asy):
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3505, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "/tmp/ipykernel_581/3495555842.py", line 8, in <module>
    print(fn(torch.ones(3, 3)))
  File "/tmp/ipykernel_581/3495555842.py", line 4, in fn
    if y > 0:

Graph break from `Tensor.item()`, consider setting:
    torch._dynamo.config.capture_scalar_outputs = True
or:
    env TORCHDYNAMO_CAPTURE_SCALAR_OUTPUTS=1
to include these operations in the captured graph.

Graph break: from user code at:
  File "/tmp/ipykernel_581/3495555842.py", line 5, in torch_dynamo_resume_in_fn_at_4
    return x + y.item()


Graph break in user code at /tmp/ipykernel_581/3495555842.py:5
Graph Break Reason: Unsupported Tensor.item() call with capture_scalar_outputs=False
  Explanation: Dynamo does not support tracing `Tensor.item()` with config.capture_scalar_outputs=False.
  Hint: Set `torch._dynamo.config.capture_scalar_outputs = True` or `export TORCHDYNAMO_CAPTURE_SCALAR_OUTPUTS=1` to include these operations in the captured graph.

  Developer debug context: call_method TensorVariable() item () {}

 For more details about this graph break, please visit: https://meta-pytorch.github.io/compile-graph-break-site/gb/gb0124.html
User code traceback:
  File "/opt/conda/envs/py_3.10/lib/python3.10/runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/opt/conda/envs/py_3.10/lib/python3.10/runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel_launcher.py", line 18, in <module>
    app.launch_new_instance()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/traitlets/config/application.py", line 1075, in launch_instance
    app.start()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelapp.py", line 739, in start
    self.io_loop.start()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/tornado/platform/asyncio.py", line 211, in start
    self.asyncio_loop.run_forever()
  File "/opt/conda/envs/py_3.10/lib/python3.10/asyncio/base_events.py", line 603, in run_forever
    self._run_once()
  File "/opt/conda/envs/py_3.10/lib/python3.10/asyncio/base_events.py", line 1909, in _run_once
    handle._run()
  File "/opt/conda/envs/py_3.10/lib/python3.10/asyncio/events.py", line 80, in _run
    self._context.run(self._callback, *self._args)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 519, in dispatch_queue
    await self.process_one()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 508, in process_one
    await dispatch(*args)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 400, in dispatch_shell
    await result
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/ipkernel.py", line 368, in execute_request
    await super().execute_request(stream, ident, parent)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 767, in execute_request
    reply_content = await reply_content
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/ipkernel.py", line 455, in do_execute
    res = shell.run_cell(
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/zmqshell.py", line 577, in run_cell
    return super().run_cell(*args, **kwargs)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3006, in run_cell
    result = self._run_cell(
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3061, in _run_cell
    result = runner(coro)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/async_helpers.py", line 129, in _pseudo_sync_runner
    coro.send(None)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3266, in run_cell_async
    has_raised = await self.run_ast_nodes(code_ast.body, cell_name,
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3445, in run_ast_nodes
    if await self.run_code(code, result, async_=asy):
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3505, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "/tmp/ipykernel_581/3495555842.py", line 8, in <module>
    print(fn(torch.ones(3, 3)))
  File "/tmp/ipykernel_581/3495555842.py", line 5, in fn
    return x + y.item()

这些图中断的通用解决方法是避免进行依赖数据的操作。一些具体的解决方法是：

  • 如果您的控制流实际上不依赖于数据值，请考虑修改代码以对常量执行控制流。

# old
x = torch.randn(3, 3)
@torch.compile
def fn(y):
    if x.sum() > 0:
        return y + x
    else:
        return y - x

print(fn(torch.ones(3, 3)))

tensor([[3.6763, 1.8030, 1.3392],
        [1.4292, 0.4356, 2.3070],
        [1.9438, 1.3472, 0.3239]])

Graph break in user code at /tmp/ipykernel_581/2410325100.py:5
Graph Break Reason: Data-dependent branching
  Explanation: Detected data-dependent branching (e.g. `if my_tensor.sum() > 0:`). Dynamo does not support tracing dynamic control flow.
  Hint: This graph break is fundamental - it is unlikely that Dynamo will ever be able to trace through your code. Consider finding a workaround.
  Hint: Use `torch.cond` to express dynamic control flow.

  Developer debug context: attempted to jump with TensorVariable()

User code traceback:
  File "/opt/conda/envs/py_3.10/lib/python3.10/runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/opt/conda/envs/py_3.10/lib/python3.10/runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel_launcher.py", line 18, in <module>
    app.launch_new_instance()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/traitlets/config/application.py", line 1075, in launch_instance
    app.start()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelapp.py", line 739, in start
    self.io_loop.start()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/tornado/platform/asyncio.py", line 211, in start
    self.asyncio_loop.run_forever()
  File "/opt/conda/envs/py_3.10/lib/python3.10/asyncio/base_events.py", line 603, in run_forever
    self._run_once()
  File "/opt/conda/envs/py_3.10/lib/python3.10/asyncio/base_events.py", line 1909, in _run_once
    handle._run()
  File "/opt/conda/envs/py_3.10/lib/python3.10/asyncio/events.py", line 80, in _run
    self._context.run(self._callback, *self._args)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 519, in dispatch_queue
    await self.process_one()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 508, in process_one
    await dispatch(*args)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 400, in dispatch_shell
    await result
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/ipkernel.py", line 368, in execute_request
    await super().execute_request(stream, ident, parent)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 767, in execute_request
    reply_content = await reply_content
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/ipkernel.py", line 455, in do_execute
    res = shell.run_cell(
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/zmqshell.py", line 577, in run_cell
    return super().run_cell(*args, **kwargs)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3006, in run_cell
    result = self._run_cell(
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3061, in _run_cell
    result = runner(coro)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/async_helpers.py", line 129, in _pseudo_sync_runner
    coro.send(None)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3266, in run_cell_async
    has_raised = await self.run_ast_nodes(code_ast.body, cell_name,
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3445, in run_ast_nodes
    if await self.run_code(code, result, async_=asy):
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3505, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "/tmp/ipykernel_581/2410325100.py", line 10, in <module>
    print(fn(torch.ones(3, 3)))
  File "/tmp/ipykernel_581/2410325100.py", line 5, in fn
    if x.sum() > 0:

# new
x = torch.randn(3, 3)
cond = (x.sum() > 0).item()
@torch.compile
def fn(y):
    if cond:
        return y + x
    else:
        return y - x

print(fn(torch.ones(3, 3)))

tensor([[2.8814, 1.2171, 1.6990],
        [1.0319, 2.2353, 0.8251],
        [0.0565, 0.7153, 0.7048]])

  • 使用更高级别的操作，例如 控制流 - Cond，来代替依赖数据的控制流。

# old
@torch.compile
def fn(x):
    if x.sum() > 0:
        return x + 1
    return x - 1

print(fn(torch.ones(3, 3)))

tensor([[2., 2., 2.],
        [2., 2., 2.],
        [2., 2., 2.]])

Graph break in user code at /tmp/ipykernel_581/520574912.py:4
Graph Break Reason: Data-dependent branching
  Explanation: Detected data-dependent branching (e.g. `if my_tensor.sum() > 0:`). Dynamo does not support tracing dynamic control flow.
  Hint: This graph break is fundamental - it is unlikely that Dynamo will ever be able to trace through your code. Consider finding a workaround.
  Hint: Use `torch.cond` to express dynamic control flow.

  Developer debug context: attempted to jump with TensorVariable()

User code traceback:
  File "/opt/conda/envs/py_3.10/lib/python3.10/runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/opt/conda/envs/py_3.10/lib/python3.10/runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel_launcher.py", line 18, in <module>
    app.launch_new_instance()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/traitlets/config/application.py", line 1075, in launch_instance
    app.start()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelapp.py", line 739, in start
    self.io_loop.start()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/tornado/platform/asyncio.py", line 211, in start
    self.asyncio_loop.run_forever()
  File "/opt/conda/envs/py_3.10/lib/python3.10/asyncio/base_events.py", line 603, in run_forever
    self._run_once()
  File "/opt/conda/envs/py_3.10/lib/python3.10/asyncio/base_events.py", line 1909, in _run_once
    handle._run()
  File "/opt/conda/envs/py_3.10/lib/python3.10/asyncio/events.py", line 80, in _run
    self._context.run(self._callback, *self._args)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 519, in dispatch_queue
    await self.process_one()
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 508, in process_one
    await dispatch(*args)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 400, in dispatch_shell
    await result
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/ipkernel.py", line 368, in execute_request
    await super().execute_request(stream, ident, parent)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/kernelbase.py", line 767, in execute_request
    reply_content = await reply_content
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/ipkernel.py", line 455, in do_execute
    res = shell.run_cell(
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/ipykernel/zmqshell.py", line 577, in run_cell
    return super().run_cell(*args, **kwargs)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3006, in run_cell
    result = self._run_cell(
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3061, in _run_cell
    result = runner(coro)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/async_helpers.py", line 129, in _pseudo_sync_runner
    coro.send(None)
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3266, in run_cell_async
    has_raised = await self.run_ast_nodes(code_ast.body, cell_name,
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3445, in run_ast_nodes
    if await self.run_code(code, result, async_=asy):
  File "/opt/conda/envs/py_3.10/lib/python3.10/site-packages/IPython/core/interactiveshell.py", line 3505, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "/tmp/ipykernel_581/520574912.py", line 8, in <module>
    print(fn(torch.ones(3, 3)))
  File "/tmp/ipykernel_581/520574912.py", line 4, in fn
    if x.sum() > 0:

# new
@torch.compile
def fn(x):
    return torch.cond(
        x.sum() > 0,
        lambda x: x + 1,
        lambda x: x - 1,
        (x,),
    )

print(fn(torch.ones(3, 3)))

tensor([[2., 2., 2.],
        [2., 2., 2.],
        [2., 2., 2.]])

  • 如果您有一个 .item() 调用，可以尝试使用 torch._dynamo.config.capture_scalar_outputs = True 或 TORCHDYNAMO_CAPTURE_SCALAR_OUTPUTS=1。

  • 将函数中有问题的部分包装成自定义运算符。

打印和日志记录#

打印/记录/发出警告将导致图中断。您可以尝试使用 torch._dynamo.config.reorderable_logging_functions 来解决这个问题。此配置用于重新排序日志记录函数，使它们在跟踪函数的末尾被调用，从而避免图中断。但是，如果发生变异等情况，记录的内容可能会有所不同。

torch._dynamo.config.reorderable_logging_functions.add(print)

@torch.compile
def fn(x):
    x += 1
    print("log!")
    return torch.sin(x)

print(fn(torch.ones(3, 3)))

log!
tensor([[0.9093, 0.9093, 0.9093],
        [0.9093, 0.9093, 0.9093],
        [0.9093, 0.9093, 0.9093]])
在此页面上
在 GitHub 上编辑
显示源代码
PyTorch 库