Traceback (most recent call last):
File "caffe2onnx.py", line 32, in
main()
File "caffe2onnx.py", line 28, in main
convert(args)
File "caffe2onnx.py", line 21, in convert
converter.run()
File "/workspace/caffe2onnx-main/converter/onnx/caffe_parser.py", line 986, in run
checker.check_model(self.model_def)
File "/root/anaconda3/lib/python3.6/site-packages/onnx/checker.py", line 105, in check_model
C.check_model(protobuf_string)
onnx.onnx_cpp2py_export.checker.ValidationError: Mismatched attribute type in 'relu1 : alpha'

==> Context: Bad node spec for node. Name: relu1 OpType: LeakyRelu

Hello author. Please tell me how to convert the LSTM layer carried by BLVC/caffe to onnx? Is there a way to provide it? , the following model. https://github.com/BVLC/caffe/blob/master/src/caffe/layers/lstm_layer.cpp
lstm_caffemodel.zip
Thank you very much for your work and look forward to your reply.
@inisis

I have a caffe distribution that has all of the layers that are in your module.
I have succesfully bulit caffe and pycaffe for that build.
I set the CAFE_ROOT and PYTHONPATH variables to point to my build.

When I run:
python caffe2onnx.py deploy.prototxtmodel_caffemodel onnx_name,
I get:
Traceback (most recent call last):
File "caffe2onnx.py", line 3, in
from converter.onnx.caffe_parser import caffe2onnx_converter
File "/home/mkraus/caffe2onnx/converter/onnx/caffe_parser.py", line 7, in
import caffe
File "/home/mkraus/caffe/python/caffe/init.py", line 1, in
from .pycaffe import Net, SGDSolver, NesterovSolver, AdaGradSolver, RMSPropSolver, AdaDeltaSolver, AdamSolver, NCCL, Timer
File "/home/mkraus/caffe/python/caffe/pycaffe.py", line 13, in
from ._caffe import Net, SGDSolver, NesterovSolver, AdaGradSolver,
ImportError: libprotobuf.so.9: cannot open shared object file: No such file or directory

If I don't set PYTHONPATH, I get a 'can't import caffe' error.

I tried running the model examples, these failed too. Please help