GB0264

Graph-Break Type

Short name describing what triggered the graph break

Attempted to use torch.nn.Parameter() constructor with Dynamo

Context

Values or code snippet captured at the break point

Explanation

Explanation of why the graph break was triggered

Dynamo does not support this

Hints

Hints on how to resolve the graph break

• Try to construct torch.nn.Parameter() outside the compiled region.
• If this is not possible, turn graph_break_on_nn_param_ctor off
• It may be possible to write Dynamo tracing rules for this code. Please report an issue to PyTorch if you encounter this graph break often and it is causing performance issues.

Additional Information

Example code that causes this graph break:

import torch

@torch.compile(fullgraph=True)
def fn(x):
    param = torch.nn.Parameter(torch.randn_like(x))
    return param, param + x

fn(torch.randn(3, 3))

Try to construct nn.Parameter() outside the compiled region.

@torch.compile(fullgraph=True)
def fn(x, param):
    return param + x

fn(torch.randn(3, 3), torch.nn.Parameter(torch.randn(3, 3)))

If this is not possible, turn torch._dynamo.config.graph_break_on_nn_param_ctor off (NOT RECOMMENDED):

torch._dynamo.config.graph_break_on_nn_param_ctor = False

@torch.compile(fullgraph=True)
def fn(x):
    param = torch.nn.Parameter(torch.randn_like(x))
    return param, param + x

fn(torch.randn(3, 3))

Click here to add Additional Info

Back to Registry