TypeError: object of type 'NoneType' has no len()

[neso613]

• Tensorflow Version: 2.4.1
• Python version: 3.7
• ONNX version: 1.8.1
  I am stuck while converting TF model to ONNX with this error

TypeError: object of type 'NoneType' has no len()

python3 -m tf2onnx.convert --saved-model my_saved_models/ --signature_def 'serving_default' --tag 'serve' --opset 10 --output my.onnx
tried different opset also

[TomWildenhain-Microsoft]

Can you upload the model you are trying to convert?

[neso613]

Can you upload the model you are trying to convert?

Please accept my apologies. I can not share the model. However,
I am sharing some details of model-

Thanks!

[TomWildenhain-Microsoft]

Hmm, the signature names isn't going to help much. Can you provide the full stack trace/full console output?

[neso613]

Hmm, the signature names isn't going to help much. Can you provide the full stack trace/full console output?
signature_def['__saved_model_init_op']:
The given SavedModel SignatureDef contains the following input(s):
The given SavedModel SignatureDef contains the following output(s):
outputs['__saved_model_init_op'] tensor_info:
dtype: DT_INVALID
shape: unknown_rank
name: NoOp
Method name is:

signature_def['serving_default']:
The given SavedModel SignatureDef contains the following input(s):
inputs['input_2'] tensor_info:
dtype: DT_FLOAT
shape: (-1, 256, 121, 1)
name: serving_default_input_2:0
The given SavedModel SignatureDef contains the following output(s):
outputs['conv1d_151'] tensor_info:
dtype: DT_FLOAT
shape: (-1, 256, 121)
name: StatefulPartitionedCall:0
Method name is: tensorflow/serving/predict

Defined Functions:
Function Name: 'call'
Option #1
Callable with:
Argument #1
inputs: TensorSpec(shape=(None, 256, 121, 1), dtype=tf.float32, name='inputs')
Argument #2
DType: bool
Value: False
Argument #3
DType: NoneType
Value: None
Option #2
Callable with:
Argument #1
input_2: TensorSpec(shape=(None, 256, 121, 1), dtype=tf.float32, name='input_2')
Argument #2
DType: bool
Value: False
Argument #3
DType: NoneType
Value: None
Option #3
Callable with:
Argument #1
input_2: TensorSpec(shape=(None, 256, 121, 1), dtype=tf.float32, name='input_2')
Argument #2
DType: bool
Value: True
Argument #3
DType: NoneType
Value: None
Option #4
Callable with:
Argument #1
inputs: TensorSpec(shape=(None, 256, 121, 1), dtype=tf.float32, name='inputs')
Argument #2
DType: bool
Value: True
Argument #3
DType: NoneType
Value: None

Function Name: '_default_save_signature'
Option #1
Callable with:
Argument #1
input_2: TensorSpec(shape=(None, 256, 121, 1), dtype=tf.float32, name='input_2')

Function Name: 'call_and_return_all_conditional_losses'
Option #1
Callable with:
Argument #1
input_2: TensorSpec(shape=(None, 256, 121, 1), dtype=tf.float32, name='input_2')
Argument #2
DType: bool
Value: False
Argument #3
DType: NoneType
Value: None
Option #2
Callable with:
Argument #1
input_2: TensorSpec(shape=(None, 256, 121, 1), dtype=tf.float32, name='input_2')
Argument #2
DType: bool
Value: True
Argument #3
DType: NoneType
Value: None
Option #3
Callable with:
Argument #1
inputs: TensorSpec(shape=(None, 256, 121, 1), dtype=tf.float32, name='inputs')
Argument #2
DType: bool
Value: True
Argument #3
DType: NoneType
Value: None
Option #4
Callable with:
Argument #1
inputs: TensorSpec(shape=(None, 256, 121, 1), dtype=tf.float32, name='inputs')
Argument #2
DType: bool
Value: False
Argument #3
DType: NoneType
Value: None

[TomWildenhain-Microsoft]

It looks like tf2onnx is calling 'len' on some object that is None. Does the stack trace tell you what line of code this error occurred on? I think you gave me the console output but not the stack trace.

[neso613]

2021-05-17 14:52:23,241 - INFO - Using tensorflow=2.5.0, onnx=1.9.0, tf2onnx=1.8.5/50049d
2021-05-17 14:52:23,242 - INFO - Using opset <onnx, 11>
2021-05-17 14:52:25,201 - INFO - Computed 0 values for constant folding
2021-05-17 14:52:32,675 - ERROR - Failed to convert node 'StatefulPartitionedCall/model_2/model_1/conv2d_4/BiasAdd_1' (fct=<bound method ConvOp.version_11 of <class 'tf2onnx.onnx_opset.nn.ConvOp'>>)
'OP=Conv\nName=StatefulPartitionedCall/model_2/model_1/conv2d_4/BiasAdd_1\nInputs:\n\tStatefulPartitionedCall/model_2/lambda_1/Slice:0=Slice, None, 1\n\tStatefulPartitionedCall/model_2/model_1/conv2d_4/Conv2D_1/ReadVariableOp:0=Const, [5, 5, 1, 16], 1\n\tStatefulPartitionedCall/model_2/model_1/conv2d_4/BiasAdd_1/ReadVariableOp:0=Const, [16], 1\nOutpus:\n\tStatefulPartitionedCall/model_2/model_1/conv2d_4/BiasAdd_1:0=[-1, 256, 121, 16], 1'
Traceback (most recent call last):
File "/home/geek/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/tfonnx.py", line 294, in tensorflow_onnx_mapping
func(g, node, **kwargs, initialized_tables=initialized_tables, dequantize=dequantize)
File "/home/geek/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/onnx_opset/nn.py", line 386, in version_11
cls.any_version(11, ctx, node, **kwargs)
File "/home/geek/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/onnx_opset/nn.py", line 356, in any_version
if len(input_shape) == spatial + 1:
TypeError: object of type 'NoneType' has no len()
2021-05-17 14:52:32,924 - ERROR - Failed to convert node 'StatefulPartitionedCall/model_2/model_1/conv2d_4/BiasAdd' (fct=<bound method ConvOp.version_11 of <class 'tf2onnx.onnx_opset.nn.ConvOp'>>)
'OP=Conv\nName=StatefulPartitionedCall/model_2/model_1/conv2d_4/BiasAdd\nInputs:\n\tStatefulPartitionedCall/model_2/lambda/Slice:0=Slice, None, 1\n\tStatefulPartitionedCall/model_2/model_1/conv2d_4/Conv2D/ReadVariableOp:0=Const, [5, 5, 1, 16], 1\n\tStatefulPartitionedCall/model_2/model_1/conv2d_4/BiasAdd/ReadVariableOp:0=Const, [16], 1\nOutpus:\n\tStatefulPartitionedCall/model_2/model_1/conv2d_4/BiasAdd:0=[-1, 256, 121, 16], 1'
Traceback (most recent call last):
File "/home/geek/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/tfonnx.py", line 294, in tensorflow_onnx_mapping
func(g, node, **kwargs, initialized_tables=initialized_tables, dequantize=dequantize)
File "/home/geek/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/onnx_opset/nn.py", line 386, in version_11
cls.any_version(11, ctx, node, **kwargs)
File "/home/geek/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/onnx_opset/nn.py", line 356, in any_version
if len(input_shape) == spatial + 1:
TypeError: object of type 'NoneType' has no len()
Traceback (most recent call last):
File "/home/geek/anaconda3/envs/tf2onnx/lib/python3.7/runpy.py", line 193, in _run_module_as_main
"main", mod_spec)
File "/home/geek/anaconda3/envs/tf2onnx/lib/python3.7/runpy.py", line 85, in _run_code
exec(code, run_globals)
File "/home/geek/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/convert.py", line 488, in
main()
File "/home/geek/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/convert.py", line 256, in main
output_path=args.output)
File "/home/geek/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/convert.py", line 153, in _convert_common
g = process_tf_graph(tf_graph, const_node_values=const_node_values, **kwargs)
File "/home/geek/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/tfonnx.py", line 534, in process_tf_graph
output_names, initialized_tables, outputs_to_values, outputs_to_dtypes, op_cnt, attr_cnt)
File "/home/geek/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/tfonnx.py", line 639, in process_parsed_graph
raise exceptions[0]
File "/home/geek/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/tfonnx.py", line 294, in tensorflow_onnx_mapping
func(g, node, **kwargs, initialized_tables=initialized_tables, dequantize=dequantize)
File "/home/geek/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/onnx_opset/nn.py", line 386, in version_11
cls.any_version(11, ctx, node, **kwargs)
File "/home/geek/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/onnx_opset/nn.py", line 356, in any_version
if len(input_shape) == spatial + 1:
TypeError: object of type 'NoneType' has no len()

[TomWildenhain-Microsoft]

Let's try #1524 and see if it fixes it.

[TomWildenhain-Microsoft]

The change has merged. @neso613 try the latest tf2onnx from GitHub and let me know if it works.

[neso613]

The change has merged. @neso613 try the latest tf2onnx from GitHub and let me know if it works.

still have same issue. I have installed latest versions.
2021-05-21 12:20:41,129 - INFO - Using tensorflow=2.5.0, onnx=1.9.0, tf2onnx=1.8.5/50049d

[TomWildenhain-Microsoft]

The issue is caused because the input rank to the conv is unknown. This should be very rare. Do you have any idea why the input rank would be unknown for this model? You can try passing in a shape to the inputs --inputs someinp:0[1,2,3,4]

I would be very surprised if the line number is the same. Do you have a new stack trace?

[neso613]

The issue is caused because the input rank to the conv is unknown. This should be very rare. Do you have any idea why the input rank would be unknown for this model? You can try passing in a shape to the inputs --inputs someinp:0[1,2,3,4]

I would be very surprised if the line number is the same. Do you have a new stack trace?
Getting this error when --inputs someinp:0[1,2,3,4] used .

WARNING:tensorflow:From /home/nehasoni/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/tf_loader.py:603: extract_sub_graph (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use tf.compat.v1.graph_util.extract_sub_graph
2021-05-24 15:08:18,173 - WARNING - From /home/nehasoni/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/tf_loader.py:603: extract_sub_graph (from tensorflow.python.framework.graph_util_impl) is deprecated and will be removed in a future version.
Instructions for updating:
Use tf.compat.v1.graph_util.extract_sub_graph
Traceback (most recent call last):
File "/home/nehasoni/anaconda3/envs/tf2onnx/lib/python3.7/runpy.py", line 193, in _run_module_as_main
"main", mod_spec)
File "/home/nehasoni/anaconda3/envs/tf2onnx/lib/python3.7/runpy.py", line 85, in _run_code
exec(code, run_globals)
File "/home/nehasoni/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/convert.py", line 488, in
main()
File "/home/nehasoni/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/convert.py", line 213, in main
args.large_model, return_initialized_tables=True, return_tensors_to_rename=True)
File "/home/nehasoni/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/tf_loader.py", line 512, in from_saved_model
tag, signatures, concrete_function, large_model)
File "/home/nehasoni/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/tf_loader.py", line 469, in _from_saved_model_v2
frozen_graph = from_function(concrete_func, inputs, outputs, large_model)
File "/home/nehasoni/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/tf_loader.py", line 179, in from_function
graph_def = tf_optimize(input_names, output_names, graph_def)
File "/home/nehasoni/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tf2onnx/tf_loader.py", line 603, in tf_optimize
graph_def = extract_sub_graph(graph_def, needed_names)
File "/home/nehasoni/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tensorflow/python/util/deprecation.py", line 337, in new_func
return func(*args, **kwargs)
File "/home/nehasoni/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tensorflow/python/framework/graph_util_impl.py", line 209, in extract_sub_graph
_assert_nodes_are_present(name_to_node, dest_nodes)
File "/home/nehasoni/anaconda3/envs/tf2onnx/lib/python3.7/site-packages/tensorflow/python/framework/graph_util_impl.py", line 164, in _assert_nodes_are_present
assert d in name_to_node, "%s is not in graph" % d
AssertionError: someinp is not in graph

[TomWildenhain-Microsoft]

someinp:0[1,2,3,4] was an example. You will need to replace it with the actual model input name and shape.

[guschmue]

I assume this is resolved.