Description
🐛 Describe the bug
torch.onnx.export fails for torch.nn.InstanceNorm1d (and 2d for that sake) when initialized with track_running_stats=True and using dynamic batch size.
Minimal reproducing code:
import torch
B, C, N = 3, 7, 11
model = torch.nn.InstanceNorm1d(C, track_running_stats=True)
model.eval()
inputs = torch.zeros(B, C, N)
torch.onnx.export(model, inputs, '/tmp/model.onnx',
input_names=['inputs'],
output_names=['outputs'],
dynamic_axes={
'inputs': {0: 'batch_size'},
'outputs': {0: 'batch_size'}
})Traceback (This is the relevant line I believe: SymbolicValueError: Unsupported: ONNX export of instance_norm training for unknown batch size.):
---------------------------------------------------------------------------
SymbolicValueError Traceback (most recent call last)
Cell In[57], line 8
6 model.eval()
7 inputs = torch.zeros(B, C, N)
----> 8 torch.onnx.export(model, inputs, '/tmp/model.onnx',
9 input_names=['inputs'],
10 output_names=['outputs'],
11 dynamic_axes={
12 'inputs': {0: 'batch_size'},
13 'outputs': {0: 'batch_size'}
14 })
File ~/.pyenv/versions/3.10.6/envs/default/lib/python3.10/site-packages/torch/onnx/utils.py:506, in export(model, args, f, export_params, verbose, training, input_names, output_names, operator_export_type, opset_version, do_constant_folding, dynamic_axes, keep_initializers_as_inputs, custom_opsets, export_modules_as_functions)
188 @_beartype.beartype
189 def export(
190 model: Union[torch.nn.Module, torch.jit.ScriptModule, torch.jit.ScriptFunction],
(...)
206 export_modules_as_functions: Union[bool, Collection[Type[torch.nn.Module]]] = False,
207 ) -> None:
208 r"""Exports a model into ONNX format.
209
210 If ``model`` is not a :class:`torch.jit.ScriptModule` nor a
(...)
503 All errors are subclasses of :class:`errors.OnnxExporterError`.
504 """
--> 506 _export(
507 model,
508 args,
509 f,
510 export_params,
511 verbose,
512 training,
513 input_names,
514 output_names,
515 operator_export_type=operator_export_type,
516 opset_version=opset_version,
517 do_constant_folding=do_constant_folding,
518 dynamic_axes=dynamic_axes,
519 keep_initializers_as_inputs=keep_initializers_as_inputs,
520 custom_opsets=custom_opsets,
521 export_modules_as_functions=export_modules_as_functions,
522 )
File ~/.pyenv/versions/3.10.6/envs/default/lib/python3.10/site-packages/torch/onnx/utils.py:1548, in _export(model, args, f, export_params, verbose, training, input_names, output_names, operator_export_type, export_type, opset_version, do_constant_folding, dynamic_axes, keep_initializers_as_inputs, fixed_batch_size, custom_opsets, add_node_names, onnx_shape_inference, export_modules_as_functions)
1545 dynamic_axes = {}
1546 _validate_dynamic_axes(dynamic_axes, model, input_names, output_names)
-> 1548 graph, params_dict, torch_out = _model_to_graph(
1549 model,
1550 args,
1551 verbose,
1552 input_names,
1553 output_names,
1554 operator_export_type,
1555 val_do_constant_folding,
1556 fixed_batch_size=fixed_batch_size,
1557 training=training,
1558 dynamic_axes=dynamic_axes,
1559 )
1561 # TODO: Don't allocate a in-memory string for the protobuf
1562 defer_weight_export = (
1563 export_type is not _exporter_states.ExportTypes.PROTOBUF_FILE
1564 )
File ~/.pyenv/versions/3.10.6/envs/default/lib/python3.10/site-packages/torch/onnx/utils.py:1117, in _model_to_graph(model, args, verbose, input_names, output_names, operator_export_type, do_constant_folding, _disable_torch_constant_prop, fixed_batch_size, training, dynamic_axes)
1114 params_dict = _get_named_param_dict(graph, params)
1116 try:
-> 1117 graph = _optimize_graph(
1118 graph,
1119 operator_export_type,
1120 _disable_torch_constant_prop=_disable_torch_constant_prop,
1121 fixed_batch_size=fixed_batch_size,
1122 params_dict=params_dict,
1123 dynamic_axes=dynamic_axes,
1124 input_names=input_names,
1125 module=module,
1126 )
1127 except Exception as e:
1128 torch.onnx.log("Torch IR graph at exception: ", graph)
File ~/.pyenv/versions/3.10.6/envs/default/lib/python3.10/site-packages/torch/onnx/utils.py:665, in _optimize_graph(graph, operator_export_type, _disable_torch_constant_prop, fixed_batch_size, params_dict, dynamic_axes, input_names, module)
662 _C._jit_pass_onnx_set_dynamic_input_shape(graph, dynamic_axes, input_names)
663 _C._jit_pass_onnx_lint(graph)
--> 665 graph = _C._jit_pass_onnx(graph, operator_export_type)
666 _C._jit_pass_onnx_lint(graph)
667 _C._jit_pass_lint(graph)
File ~/.pyenv/versions/3.10.6/envs/default/lib/python3.10/site-packages/torch/onnx/utils.py:1891, in _run_symbolic_function(graph, block, node, inputs, env, operator_export_type)
1886 if symbolic_fn is not None:
1887 # TODO Wrap almost identical attrs assignment or comment the difference.
1888 attrs = {
1889 k: symbolic_helper._node_get(node, k) for k in node.attributeNames()
1890 }
-> 1891 return symbolic_fn(graph_context, *inputs, **attrs)
1893 attrs = {
1894 k + "_" + node.kindOf(k)[0]: symbolic_helper._node_get(node, k)
1895 for k in node.attributeNames()
1896 }
1897 if namespace == "onnx":
1898 # Clone node to trigger ONNX shape inference
File ~/.pyenv/versions/3.10.6/envs/default/lib/python3.10/site-packages/torch/onnx/symbolic_helper.py:306, in parse_args.<locals>.decorator.<locals>.wrapper(g, *args, **kwargs)
300 if len(kwargs) == 1:
301 assert "_outputs" in kwargs, (
302 f"Symbolic function {fn.__name__}'s '**kwargs' can only contain "
303 f"'_outputs' key at '**kwargs'. "
304 f"{FILE_BUG_MSG}"
305 )
--> 306 return fn(g, *args, **kwargs)
File ~/.pyenv/versions/3.10.6/envs/default/lib/python3.10/site-packages/torch/onnx/symbolic_opset9.py:2895, in instance_norm(g, input, weight, bias, running_mean, running_var, use_input_stats, momentum, eps, cudnn_enabled)
2893 n = input_size[0]
2894 if n is None:
-> 2895 raise errors.SymbolicValueError(
2896 "Unsupported: ONNX export of instance_norm training for unknown "
2897 "batch size.",
2898 input,
2899 )
2900 c = input_size[1]
2901 input_size_reshape[0] = 1
SymbolicValueError: Unsupported: ONNX export of instance_norm training for unknown batch size. [Caused by the value 'input defined in (%input : Float(*, 3, 11, strides=[33, 11, 1], requires_grad=0, device=cpu), %running_mean : Float(3, strides=[1], requires_grad=0, device=cpu), %running_var : Float(3, strides=[1], requires_grad=0, device=cpu), %num_batches_tracked : Long(requires_grad=0, device=cpu) = prim::Param()
)' (type 'Tensor') in the TorchScript graph. The containing node has kind 'prim::Param'.]
Inputs:
Empty
Outputs:
#0: input defined in (%input : Float(*, 3, 11, strides=[33, 11, 1], requires_grad=0, device=cpu), %running_mean : Float(3, strides=[1], requires_grad=0, device=cpu), %running_var : Float(3, strides=[1], requires_grad=0, device=cpu), %num_batches_tracked : Long(requires_grad=0, device=cpu) = prim::Param()
) (type 'Tensor')
#1: running_mean defined in (%input : Float(*, 3, 11, strides=[33, 11, 1], requires_grad=0, device=cpu), %running_mean : Float(3, strides=[1], requires_grad=0, device=cpu), %running_var : Float(3, strides=[1], requires_grad=0, device=cpu), %num_batches_tracked : Long(requires_grad=0, device=cpu) = prim::Param()
) (type 'Tensor')
#2: running_var defined in (%input : Float(*, 3, 11, strides=[33, 11, 1], requires_grad=0, device=cpu), %running_mean : Float(3, strides=[1], requires_grad=0, device=cpu), %running_var : Float(3, strides=[1], requires_grad=0, device=cpu), %num_batches_tracked : Long(requires_grad=0, device=cpu) = prim::Param()
) (type 'Tensor')
#3: num_batches_tracked defined in (%input : Float(*, 3, 11, strides=[33, 11, 1], requires_grad=0, device=cpu), %running_mean : Float(3, strides=[1], requires_grad=0, device=cpu), %running_var : Float(3, strides=[1], requires_grad=0, device=cpu), %num_batches_tracked : Long(requires_grad=0, device=cpu) = prim::Param()
) (type 'Tensor')Note that it works by removing dynamic axes.
It "only" gives a warning if initializing the layer with track_running_stats=False:
============= Diagnostic Run torch.onnx.export version 2.0.0+cu117 =============
verbose: False, log level: Level.ERROR
======================= 0 NONE 0 NOTE 0 WARNING 0 ERROR ========================
/home/cdeln/.pyenv/versions/3.10.6/envs/default/lib/python3.10/site-packages/torch/onnx/symbolic_helper.py:1466: UserWarning: ONNX export mode is set to TrainingMode.EVAL, but operator 'instance_norm' is set to train=True. Exporting with train=True.
warnings.warn(
Related issues:
- ONNX export with InstanceNormalization module fails on ONNX DML runtime if use_input_stats is not false #77979
- onnx export error with nn.InstanceNorm2d #123158 (Unclear issue, seems to be related to channel size)
- ONNX export of batch_norm for unknown channel size issue. #92977 (Issue with channel size)
- Why nn.Upsample/F.interpolate followed by nn.InstanceNorm2d will report error "Unsupported: ONNX export of instance_norm for unknown channel size." #99140 (Issue with channel size)
Versions
Collecting environment information...
PyTorch version: 2.0.0+cu117
Is debug build: False
CUDA used to build PyTorch: 11.7
ROCM used to build PyTorch: N/A
OS: Ubuntu 22.04.2 LTS (x86_64)
GCC version: (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Clang version: 14.0.0-1ubuntu1.1
CMake version: version 3.26.3
Libc version: glibc-2.35
Python version: 3.10.6 (main, Apr 26 2023, 09:09:46) [GCC 11.3.0] (64-bit runtime)
Python platform: Linux-6.5.0-27-generic-x86_64-with-glibc2.35
Is CUDA available: False
CUDA runtime version: No CUDA
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: No CUDA
Nvidia driver version: No CUDA
cuDNN version: No CUDA
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True
CPU:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 39 bits physical, 48 bits virtual
Byte Order: Little Endian
CPU(s): 12
On-line CPU(s) list: 0-11
Vendor ID: GenuineIntel
Model name: 12th Gen Intel(R) Core(TM) i5-1235U
CPU family: 6
Model: 154
Thread(s) per core: 2
Core(s) per socket: 10
Socket(s): 1
Stepping: 4
CPU max MHz: 4400.0000
CPU min MHz: 400.0000
BogoMIPS: 4992.00
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb ssbd ibrs ibpb stibp ibrs_enhanced fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid rdseed adx smap clflushopt clwb intel_pt sha_ni xsaveopt xsavec xgetbv1 xsaves split_lock_detect avx_vnni dtherm ida arat pln pts hwp hwp_notify hwp_act_window hwp_epp hwp_pkg_req hfi umip pku ospke waitpkg gfni vaes vpclmulqdq rdpid movdiri movdir64b fsrm md_clear serialize arch_lbr ibt flush_l1d arch_capabilities
L1d cache: 352 KiB (10 instances)
L1i cache: 576 KiB (10 instances)
L2 cache: 6.5 MiB (4 instances)
L3 cache: 12 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0-11
Vulnerability Gather data sampling: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Enhanced / Automatic IBRS, IBPB conditional, RSB filling, PBRSB-eIBRS SW sequence
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected
Versions of relevant libraries:
[pip3] fft-conv-pytorch==1.1.3
[pip3] numpy==1.24.3
[pip3] onnx==1.14.0
[pip3] onnxruntime==1.15.1
[pip3] onnxsim==0.4.33
[pip3] pytorch3d==0.7.4
[pip3] pytorchltr==0.2.1
[pip3] torch==2.0.0
[pip3] torchaudio==2.0.2
[pip3] torchmetrics==1.2.0
[pip3] torchvision==0.15.1
[pip3] triton==2.0.0
[conda] Could not collect