Nothing Special   »   [go: up one dir, main page]
Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

torch_save() and torch_load() does not work for large tensors #1210

Open
D-Maar opened this issue Nov 13, 2024 · 1 comment
Open

torch_save() and torch_load() does not work for large tensors #1210

D-Maar opened this issue Nov 13, 2024 · 1 comment

Comments

@D-Maar
Copy link
D-Maar commented Nov 13, 2024 •
edited
Loading

If I save a large tensor (5000x10x300x300) with torch_save() and then try to load it with torch_load() I get the error:

Error in cpp_tensor_load(obj$values, device, base64) : 
  PytorchStreamReader failed reading zip archive: failed finding central directory
Exception raised from valid at ../caffe2/serialize/inline_container.cc:183 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >) + 0x6b (0x7f584f9563cb in /home/.../R/x86_64-pc-linux-gnu-library/4.2/torch/lib/libc10.so)
frame #1: c10::detail::torchCheckFail(char const*, char const*, unsigned int, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&) + 0xce (0x7f584f951d9e in /home/.../R/x86_64-pc-linux-gnu-library/4.2/torch/lib/libc10.so)
frame #2: caffe2::serialize::PyTorchStreamReader::valid(char const*, char const*) + 0x3ca (0x7f575f31631a in /home/.../R/x86_64-pc-linux-gnu-library/4.2/torch/lib/libtorch_cpu.so)
frame #3: caffe2::serialize::PyTorchStreamReader::init() + 0xad (0x7f575f316bad in /home/.../R/x86_64-pc-linux-gnu-library/

If I do the same with a smaller tensor (500x10x300x300) it just works fine.

Example:

#Does not work
dims<-c(5000,10,300,300)
tensor<-torch::torch_tensor(array((1:prod(dims))/length(dims), dims), dtype = torch::torch_float16())
torch::torch_save(tensor, "./test.pt")
tensor<-torch::torch_load("./test.pt")

#Does work 
dims<-c(500,10,300,300)
tensor<-torch::torch_tensor(array((1:prod(dims))/length(dims), dims), dtype = torch::torch_float16())
torch::torch_save(tensor, "./test.pt")
tensor<-torch::torch_load("./test.pt")

Also:

  1. Adding compress = F, did not fix it.
  2. The size of the resulting file seems way to small in the non-working case (445 KB with compression; 400.5 MB without compression).
  3. serializing it with torch:::torch_serialize() and then loading it with torch::torch_load() (no storing between the serialization and the loading; just keeping it in the RAM) does also not work and yields the same error.
@sebffischer
Copy link
Collaborator

For what it's worth: I cannot reproduce this and saving a tensor of this size works for me.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@sebffischer @D-Maar