1.If the input image to be set on the cuda is a bit big , it will report the error ‘segmentation fault(core dump)’.
I try to resize the image to 100x100x3 , it won’t report the error . If I resize the image to 512x512x3 , it will report the error .
2.When import the .pt file generated by torchscript in python, it will report the following error :
terminate called after throwing an instance of ‘c10::Error’
what(): c >= ‘0’ && c <= ‘z’ ASSERT FAILED at /pytorch/torch/csrc/jit/pickler.cpp:487, please report a bug to PyTorch. (readString at /pytorch/torch/csrc/jit/pickler.cpp:487)
frame #0: std::function<std::string ()>::operator()() const + 0x11 (0x7f1be414dbf1 in /media/usr515/ed65d8df-d015-4269-90bc-ac3cccb109a8/dsq/libtorch/libtorch/lib/libc10.so)
frame #1: c10::Error::Error(c10::SourceLocation, std::string const&) + 0x2a (0x7f1be414d52a in /media/usr515/ed65d8df-d015-4269-90bc-ac3cccb109a8/dsq/libtorch/libtorch/lib/libc10.so)
frame #2: torch::jit::Unpickler::readString() + 0x138 (0x7f1be4befaf8 in /media/usr515/ed65d8df-d015-4269-90bc-ac3cccb109a8/dsq/libtorch/libtorch/lib/libtorch.so.1)
frame #3: torch::jit::Unpickler::readInstruction() + 0x5c6 (0x7f1be4bf1366 in /media/usr515/ed65d8df-d015-4269-90bc-ac3cccb109a8/dsq/libtorch/libtorch/lib/libtorch.so.1)
frame #4: torch::jit::Unpickler::run() + 0x55 (0x7f1be4bf2225 in /media/usr515/ed65d8df-d015-4269-90bc-ac3cccb109a8/dsq/libtorch/libtorch/lib/libtorch.so.1)
frame #5: torch::jit::Unpickler::parse_ivalue_list() + 0x1e (0x7f1be4bf259e in /media/usr515/ed65d8df-d015-4269-90bc-ac3cccb109a8/dsq/libtorch/libtorch/lib/libtorch.so.1)
frame #6: + 0x9dcc45 (0x7f1be4d43c45 in /media/usr515/ed65d8df-d015-4269-90bc-ac3cccb109a8/dsq/libtorch/libtorch/lib/libtorch.so.1)
frame #7: torch::jit::load(std::unique_ptr<caffe2::serialize::ReadAdapterInterface, std::default_deletecaffe2::serialize::ReadAdapterInterface >, c10::optionalc10::Device, std::unordered_map<std::string, std::string, std::hashstd::string, std::equal_tostd::string, std::allocator<std::pair<std::string const, std::string> > >&) + 0x10d (0x7f1be4d4549d in /media/usr515/ed65d8df-d015-4269-90bc-ac3cccb109a8/dsq/libtorch/libtorch/lib/libtorch.so.1)
frame #8: torch::jit::load(std::string const&, c10::optionalc10::Device, std::unordered_map<std::string, std::string, std::hashstd::string, std::equal_tostd::string, std::allocator<std::pair<std::string const, std::string> > >&) + 0x68 (0x7f1be4d455c8 in /media/usr515/ed65d8df-d015-4269-90bc-ac3cccb109a8/dsq/libtorch/libtorch/lib/libtorch.so.1)
frame #9: main + 0x6b7 (0x41f4c6 in ./demo)
frame #10: __libc_start_main + 0xf0 (0x7f1b9da8d830 in /lib/x86_64-linux-gnu/libc.so.6)
frame #11: _start + 0x29 (0x41dfe9 in ./demo)
environment:
pytorch 1.1.0
cuda 10.0
libtorch 1.1.0
gpu RTX-TITAN x2
ubuntu 16.04