BVLC model inference failed

[chop2]

I have my own caffemodel base on BVLC/caffe，but when run in nvcaffe, some error occur as follow. Is nvcaffe's caffemodel file format diffence from bvlccaffe's ?

I0330 00:28:04.993422 24940 net.cpp:437] Top memory (TEST) required for data: 5831872 diff: 5831872
I0330 00:28:04.993432 24940 net.cpp:440] Bottom memory (TEST) required for data: 5830720 diff: 5830720
I0330 00:28:04.993438 24940 net.cpp:443] Shared (in-place) memory (TEST) by data: 2512896 diff: 2512896
I0330 00:28:04.993445 24940 net.cpp:446] Parameters memory (TEST) required for data: 5746304 diff: 5746304
I0330 00:28:04.993453 24940 net.cpp:449] Parameters shared memory (TEST) by data: 0 diff: 0
I0330 00:28:04.993460 24940 net.cpp:455] Network initialization done.
F0330 00:28:04.993594 24940 upgrade_proto.cpp:98] Check failed: ReadProtoFromBinaryFile(param_file, param) Failed to parse NetParameter file: tmp.caffemodel
*** Check failure stack trace: ***
@ 0x7f2024a611fd google::LogMessage::Fail()
@ 0x7f2024a6307f google::LogMessage::SendToLog()
@ 0x7f2024a60d93 google::LogMessage::Flush()
@ 0x7f2024a6399e google::LogMessageFatal::~LogMessageFatal()
@ 0x7f202e60ba3e caffe::ReadNetParamsFromBinaryFileOrDie()
@ 0x7f202e57b627 caffe::Net::CopyTrainedLayersFromBinaryProto()
@ 0x7f202e57b696 caffe::Net::CopyTrainedLayersFrom()
@ 0x40f2c6 test()
@ 0x40b9bd main
@ 0x7f201b781b15 __libc_start_main
@ 0x40c5c1 (unknown)
已放弃

[chop2]

this problem has been solved -.-||
the reason is caffemodel contain training layer which nvcaffe not support.