Warning: MXNet Parser has not supported operator SoftmaxActivation with name cls_prob.

[IamJasonYe]

Platform (like ubuntu 16.04/win10): Ubuntu 16.04

Python version: 2.7.12

Source framework with version (like Tensorflow 1.4.1 with GPU): mxnet 1.1.0 with cu80 GPU

Destination framework with version (like CNTK 2.3 with GPU): IR

Pre-trained model path (webpath or webdisk path): N/A

Running scripts: python -m mmdnn.conversion._script.convertToIR -f mxnet -n mynet-symbol.json -w mynet.params -d ir/mynet --inputShape 3 12 12

/home/jason/.virtualenvs/caffe-mxnet-cuda8/local/lib/python2.7/site-packages/h5py/init.py:36: FutureWarning: Conversion of the second argument of issubdtype from float to np.floating is deprecated. In future, it will be treated as np.float64 == np.dtype(float).type. from ._conv import register_converters as _register_converters /home/jason/.virtualenvs/caffe-mxnet-cuda8/local/lib/python2.7/site-packages/mxnet/module/base_module.py:53: UserWarning: You created Module with Module(..., label_names=['softmax_label']) but input with name 'softmax_label' is not found in symbol.list_arguments(). Did you mean one of: data warnings.warn(msg) Warning: MXNet Parser has not supported operator null with name data. Warning: convert the null operator with name [data] into input layer. Warning: MXNet Parser has not supported operator SoftmaxActivation with name cls_prob. IR network structure is saved as [../ir/pnet.json]. IR network structure is saved as [../ir/pnet.pb]. IR weights are saved as [../ir/pnet.npy].

And here is my symbol.json of mxnet model

{ "nodes": [ { "op": "null", "name": "data", "inputs": [] }, { "op": "null", "name": "conv1_weight", "attrs": { "kernel": "(3, 3)", "num_filter": "10" }, "inputs": [] }, { "op": "null", "name": "conv1_bias", "attrs": { "kernel": "(3, 3)", "num_filter": "10" }, "inputs": [] }, { "op": "Convolution", "name": "conv1", "attrs": { "kernel": "(3, 3)", "num_filter": "10" }, "inputs": [[0, 0, 0], [1, 0, 0], [2, 0, 0]] }, { "op": "null", "name": "prelu1_gamma", "attrs": { "init": "[\"Constant\", {\"value\": 0.25}]", "act_type": "prelu" }, "inputs": [] }, { "op": "LeakyReLU", "name": "prelu1", "attrs": {"act_type": "prelu"}, "inputs": [[3, 0, 0], [4, 0, 0]] }, { "op": "Pooling", "name": "pool1", "attrs": { "kernel": "(2, 2)", "pool_type": "max", "pooling_convention": "full", "stride": "(2, 2)" }, "inputs": [[5, 0, 0]] }, { "op": "null", "name": "conv2_weight", "attrs": { "kernel": "(3, 3)", "num_filter": "16" }, "inputs": [] }, { "op": "null", "name": "conv2_bias", "attrs": { "kernel": "(3, 3)", "num_filter": "16" }, "inputs": [] }, { "op": "Convolution", "name": "conv2", "attrs": { "kernel": "(3, 3)", "num_filter": "16" }, "inputs": [[6, 0, 0], [7, 0, 0], [8, 0, 0]] }, { "op": "null", "name": "prelu2_gamma", "attrs": { "init": "[\"Constant\", {\"value\": 0.25}]", "act_type": "prelu" }, "inputs": [] }, { "op": "LeakyReLU", "name": "prelu2", "attrs": {"act_type": "prelu"}, "inputs": [[9, 0, 0], [10, 0, 0]] }, { "op": "null", "name": "conv3_weight", "attrs": { "kernel": "(3, 3)", "num_filter": "32" }, "inputs": [] }, { "op": "null", "name": "conv3_bias", "attrs": { "kernel": "(3, 3)", "num_filter": "32" }, "inputs": [] }, { "op": "Convolution", "name": "conv3", "attrs": { "kernel": "(3, 3)", "num_filter": "32" }, "inputs": [[11, 0, 0], [12, 0, 0], [13, 0, 0]] }, { "op": "null", "name": "prelu3_gamma", "attrs": { "init": "[\"Constant\", {\"value\": 0.25}]", "act_type": "prelu" }, "inputs": [] }, { "op": "LeakyReLU", "name": "prelu3", "attrs": {"act_type": "prelu"}, "inputs": [[14, 0, 0], [15, 0, 0]] }, { "op": "null", "name": "conv4_1_weight", "attrs": { "kernel": "(1, 1)", "num_filter": "2" }, "inputs": [] }, { "op": "null", "name": "conv4_1_bias", "attrs": { "kernel": "(1, 1)", "num_filter": "2" }, "inputs": [] }, { "op": "Convolution", "name": "conv4_1", "attrs": { "kernel": "(1, 1)", "num_filter": "2" }, "inputs": [[16, 0, 0], [17, 0, 0], [18, 0, 0]] }, { "op": "SoftmaxActivation", "name": "cls_prob", "attrs": {"mode": "channel"}, "inputs": [[19, 0, 0]] }, { "op": "null", "name": "conv4_2_weight", "attrs": { "kernel": "(1, 1)", "num_filter": "4" }, "inputs": [] }, { "op": "null", "name": "conv4_2_bias", "attrs": { "kernel": "(1, 1)", "num_filter": "4" }, "inputs": [] }, { "op": "Convolution", "name": "conv4_2", "attrs": { "kernel": "(1, 1)", "num_filter": "4" }, "inputs": [[16, 0, 0], [21, 0, 0], [22, 0, 0]] } ], "arg_nodes": [ 0, 1, 2, 4, 7, 8, 10, 12, 13, 15, 17, 18, 21, 22 ], "node_row_ptr": [ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24 ], "heads": [[20, 0, 0], [23, 0, 0]], "attrs": {"mxnet_version": ["int", 10100]} }%

[kitstar]

Hi @IamJasonYe , it is just a warning which can be ignored. Thanks.

[IamJasonYe]

Thx @kitstar .

[Solved]

Actually I found SoftmaxActivation opeartor is not a supportive type in the mxnet parser right now. So I changed it into the SoftmaxOutput and with parameter multi_output": "1".

ref by Joey in Zhihu

[lygstate]

My network have such an error

[01:22:00] C:\Jenkins\workspace\mxnet-tag\mxnet\src\nnvm\legacy_json_util.cc:209: Loading symbol saved by previous version v0.9.5. Attempting to upgrade...
[01:22:01] C:\Jenkins\workspace\mxnet-tag\mxnet\src\nnvm\legacy_json_util.cc:217: Symbol successfully upgraded!
e:\ci-cor-ready\ai\face-io\face-detection\mx2tensorflow-js\tensorflow_env\lib\site-packages\mxnet\module\base_module.py:54: UserWarning: You created Module with Module(..., label_names=['softmax_label']) but input with name 'softmax_label' is not found in symbol.list_arguments(). Did you mean one of:
        data
  warnings.warn(msg)
Warning: MXNet Parser has not supported operator null with name data.
Warning: convert the null operator with name [data] into input layer.
[0, 0, 62, 36]
{'op': 'Activation', 'name': 'hres4_relu', 'attr': {'act_type': 'relu'}, 'inputs': [[188, 0, 0]]}
{'act_type': 'relu'}
[0, 0, 122, 70]
{'op': 'Activation', 'name': 'hres3_relu', 'attr': {'act_type': 'relu'}, 'inputs': [[132, 0, 0]]}
{'act_type': 'relu'}
[0, 0, 121, 68]
{'op': 'Activation', 'name': 'hres3_frelu', 'attr': {'act_type': 'relu'}, 'inputs': [[263, 0, 0]]}
{'act_type': 'relu'}
Warning: SoftmaxOutput attrs are not supported in IR.
infer_shape error. Arguments:
  data: (1, 3, 960, 540)
Traceback (most recent call last):
  File "c:\users\lygstate\appdata\local\programs\python\python36\Lib\runpy.py", line 193, in _run_module_as_main
    "__main__", mod_spec)
  File "c:\users\lygstate\appdata\local\programs\python\python36\Lib\runpy.py", line 85, in _run_code
    exec(code, run_globals)
  File "E:\CI-Cor-Ready\ai\face-io\face-detection\mx2tensorflow-js\tensorflow_env\Scripts\mmconvert.exe\__main__.py", line 9, in <module>
  File "e:\ci-cor-ready\ai\face-io\face-detection\mx2tensorflow-js\tensorflow_env\lib\site-packages\mmdnn\conversion\_script\convert.py", line 102, in _main
    ret = convertToIR._convert(ir_args)
  File "e:\ci-cor-ready\ai\face-io\face-detection\mx2tensorflow-js\tensorflow_env\lib\site-packages\mmdnn\conversion\_script\convertToIR.py", line 115, in _convert
    parser.run(args.dstPath)
  File "e:\ci-cor-ready\ai\face-io\face-detection\mx2tensorflow-js\tensorflow_env\lib\site-packages\mmdnn\conversion\common\DataStructure\parser.py", line 22, in run
    self.gen_IR()
  File "e:\ci-cor-ready\ai\face-io\face-detection\mx2tensorflow-js\tensorflow_env\lib\site-packages\mmdnn\conversion\mxnet\mxnet_parser.py", line 265, in gen_IR
    func(current_node)
  File "e:\ci-cor-ready\ai\face-io\face-detection\mx2tensorflow-js\tensorflow_env\lib\site-packages\mmdnn\conversion\mxnet\mxnet_parser.py", line 629, in rename_SoftmaxOutput
    self.set_output_shape(source_node, IR_node)
  File "e:\ci-cor-ready\ai\face-io\face-detection\mx2tensorflow-js\tensorflow_env\lib\site-packages\mmdnn\conversion\mxnet\mxnet_parser.py", line 284, in set_output_shape
    arg_shape, output_shape, aux_shape = sym.infer_shape(data = self.data_shape)
  File "e:\ci-cor-ready\ai\face-io\face-detection\mx2tensorflow-js\tensorflow_env\lib\site-packages\mxnet\symbol\symbol.py", line 990, in infer_shape
    res = self._infer_shape_impl(False, *args, **kwargs)
  File "e:\ci-cor-ready\ai\face-io\face-detection\mx2tensorflow-js\tensorflow_env\lib\site-packages\mxnet\symbol\symbol.py", line 1120, in _infer_shape_impl
    ctypes.byref(complete)))
  File "e:\ci-cor-ready\ai\face-io\face-detection\mx2tensorflow-js\tensorflow_env\lib\site-packages\mxnet\base.py", line 149, in check_call
    raise MXNetError(py_str(_LIB.MXGetLastError()))
mxnet.base.MXNetError: Error in operator cls_prob_l2_: [01:22:09] C:\Jenkins\workspace\mxnet-tag\mxnet\src\nnvm\legacy_op_util.cc:194: Check failed: prop.inputs.size() == iattr->size() (2 vs. 1) op=SoftmaxOutput, inputs.size=2, iattr.size=1, arg.size=2

What's happening?