Intel® Distribution of OpenVINO™ Toolkit
Community support and discussions about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all things computer vision-related on Intel® platforms.

MYRIAD inference error

sumbals
Beginner
295 Views

When executing the following code (for CPU), everything seems working.

xml_path="Models/test/model/saved_model.xml"
bin_path="Models/test/model/saved_model.bin"

ie = IECore()
net = ie.read_network(model=xml_path, weights=bin_path)
exec_net = ie.load_network(network=net, device_name="CPU")  #device_name="MYRIAD" for NCS2

But when executing for MYRIAD (NCS2), I receive the following error

RuntimeError: [ GENERAL_ERROR ] 
/home/jenkins/agent/workspace/private-ci/ie/build-linux-ubuntu20/b/repos/openvino/src/plugins/intel_myriad/common/src/ngraph/transformations/dynamic_to_static_shape_loop.cpp:46 Encountered unknown input type of a loop v5::Loop StatefulPartitionedCall/model/bidirectional_2/forward_lstm_2/StatefulPartitionedCall/while (StatefulPartitionedCall/model/bidirectional_2/forward_lstm_2/StatefulPartitionedCall/strided_slice[0]:i32{}, StatefulPartitionedCall/model/bidirectional_2/forward_lstm_2/StatefulPartitionedCall/while/ExecutionConditionValue[0]:boolean{}, StatefulPartitionedCall/model/bidirectional_2/forward_lstm_2/StatefulPartitionedCall/time[0]:i32{}, StatefulPartitionedCall/model/bidirectional_2/forward_lstm_2/zeros/Broadcast[0]:f32{1,32}, StatefulPartitionedCall/model/bidirectional_2/forward_lstm_2/zeros_1/Broadcast[0]:f32{1,32}, StatefulPartitionedCall/model/bidirectional_2/forward_lstm_2/StatefulPartitionedCall/strided_slice[0]:i32{}, StatefulPartitionedCall/model/bidirectional_2/forward_lstm_2/StatefulPartitionedCall/transpose[0]:f32{160,1,16}) -> (f32{160,1,32}) at index 0

 Why am I receiving this error?

Has this something to do with (un)supporting layer types? If so, is there a list of compatible layer types for the NCS2?

0 Kudos
2 Replies
Peh_Intel
Moderator
191 Views

Hi sumbals,


Thanks for reaching out to us.


For your information, this Loop-5 operation is newly added support by CPU plugin starting from OpenVINO™ Toolkit 2021.2. This information is available in the Release Note under the sub-section “Inference Engine” of the section “New and Changed in the Release 2”.


Whereas, this Loop-5 operation is newly added support by GPU plugin starting from OpenVINO™ Toolkit 2022.1 (Release Note).


However, it is still not yet supported by MYRIAD plugin.


Besides, you may refer to the supported layers by each plugin here. But some of the newly added support layers might not included in the list. Hence, I would recommend Release Note is always the best documentation to refer for the latest updates.



Regards,

Peh


Peh_Intel
Moderator
155 Views

Hi sumbals,


This thread will no longer be monitored since we have provided answer. If you need any additional information from Intel, please submit a new question. 



Regards,

Peh


Reply