About Model conversion

[leungyukshing]

Hi, I want to converse an .onnx model to a dabnn model. But after i ran the given command, there is no any output. Could you please tell me some way to debug or fix it?

[daquexian]

Could you please try add --verbose and try once more? Thanks

[Cord3th]

Hi, I also try to convert .onnx model to .dab and get this error: ./onnx2bnn.AppImage resnet18.onnx resnet18.dab [libprotobuf WARNING google/protobuf/io/coded_stream.cc:537] Reading dangerously large protocol message. If the message turns out to be larger than 67108864 bytes, parsing will be halted for security reasons. To increase the limit (or to disable these warnings), see CodedInputStream::SetTotalBytesLimit() in google/protobuf/io/coded_stream.h. [libprotobuf WARNING google/protobuf/io/coded_stream.cc:78] The total number of bytes read was 44790507 terminate called after throwing an instance of 'std::out_of_range' what(): _Map_base::at Aborted (core dumped)

[daquexian]

Could you please send your onnx model to my mailbox: [email protected] ?

Thanks!

On Fri, Nov 8, 2019, 9:46 PM Cord3th [email protected] wrote:

Hi, I also try to convert .onnx model to .dab and get this error: ./onnx2bnn.AppImage resnet18.onnx resnet18.dab [libprotobuf WARNING google/protobuf/io/coded_stream.cc:537] Reading dangerously large protocol message. If the message turns out to be larger than 67108864 bytes, parsing will be halted for security reasons. To increase the limit (or to disable these warnings), see CodedInputStream::SetTotalBytesLimit() in google/protobuf/io/coded_stream.h. [libprotobuf WARNING google/protobuf/io/coded_stream.cc:78] The total number of bytes read was 44790507 terminate called after throwing an instance of 'std::out_of_range' what(): _Map_base::at Aborted (core dumped)

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/JDAI-CV/dabnn/issues/67?email_source=notifications&email_token=ACYRZH74MMCE4BFRM4SGZODQSVUSBA5CNFSM4JKDZGRKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEDSBKBA#issuecomment-551818500, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACYRZH7JJNIY7RKNKYMZZATQSVUSBANCNFSM4JKDZGRA .

[daquexian]

Thanks for your email! I have checked the model and found there are several issues keeping the model from being converted to dabnn format.

1. The batch size of your ONNX model is dynamic. Please set a static batch size and re-export the ONNX model so that the model will be supported by dabnn

2. https://github.com/onnx/onnx/issues/2417 . It is a bug of ONNX. The onnx2daq relies on ONNX optimizer, so it crashes. This problem will be fixed after simplifying your model by https://github.com/daquexian/onnx-simplifier

3. Your model has "Mul" op that dabnn doesn't support currently. I will update dabnn to support it in the next few days.

On Fri, Nov 8, 2019, 9:46 PM Cord3th [email protected] wrote:

Hi, I also try to convert .onnx model to .dab and get this error: ./onnx2bnn.AppImage resnet18.onnx resnet18.dab [libprotobuf WARNING google/protobuf/io/coded_stream.cc:537] Reading dangerously large protocol message. If the message turns out to be larger than 67108864 bytes, parsing will be halted for security reasons. To increase the limit (or to disable these warnings), see CodedInputStream::SetTotalBytesLimit() in google/protobuf/io/coded_stream.h. [libprotobuf WARNING google/protobuf/io/coded_stream.cc:78] The total number of bytes read was 44790507 terminate called after throwing an instance of 'std::out_of_range' what(): _Map_base::at Aborted (core dumped)

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/JDAI-CV/dabnn/issues/67?email_source=notifications&email_token=ACYRZH74MMCE4BFRM4SGZODQSVUSBA5CNFSM4JKDZGRKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEDSBKBA#issuecomment-551818500, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACYRZH7JJNIY7RKNKYMZZATQSVUSBANCNFSM4JKDZGRA .

[sicheng7]

Hi, I also want to converse an .onnx model to a dabnn model. I tried the command with --verbose, there is still no any output. Could you please tell me some way to debug or fix it? Thank you!

[daquexian]

@sicheng7 could you please join the qq group 1021964010?

[sicheng7]

Yes, I have joined. I will contact you immediately.