Reading a .pt files gives an error

[Het-Shah]

Hi! I have been trying to use ocaml-torch to load a PyTorch model that has already been trained. I have installed PyTorch 1.9.0 which is what is compatible with ocaml-torch. Upon trying to load the file, it throws the following error

Fatal error: exception (Failure
"version_number <= kMaxSupportedFileFormatVersion ASSERT FAILED at /pytorch/caffe2/serialize/inline_container.cc:131, please report a bug to PyTorch. Attempted to read a PyTorch file with version 4, but the maximum supported version for reading is 1. Your PyTorch installation may be too old. (init at /pytorch/caffe2/serialize/inline_container.cc:131)\

Can you suggest what can be done in this case? Thank you!

Edit: Source code
python file -

net = CNN()
net.eval()
example = torch.rand(16384)
traced = torch.jit.trace(net, example)
traced.save("./models/feat.pt")

ocaml file -

open Base
open Torch

let model = Module.load "../models/feat.pt" ;;

[LaurentMazare]

That seems very odd, if you look at the pytorch code triggering this here, this uses kMaxSupportedFileFormatVersion which is set to 6 in pytorch 1.9. This constant was set to 1 only for versions <= 1.3, also the line number in your error message doesn't match 1.9 so maybe you're not building or linking with pytorch 1.9 properly?

[Het-Shah]

The PyTorch version in my conda environment is set to 1.9.0. Is it possible that ocaml is taking some other environment's pytorch version?

>>> import torch 
>>> print(torch.__version__)
1.9.0

(My base conda environment doesn't have pytorch installed.)

[Het-Shah]

Sorry for a lot of doubts, but does opam-torch v0.7 correspond to pytorch 1.9.0 (given in README) or is it v0.13?

[Het-Shah]

Sorry for a lot of doubts, but does opam-torch v0.7 correspond to pytorch 1.9.0 (given in README) or is it v0.13?

Seems like this was the issue. I upgraded to v0.13 and reboot the system (reboot was necessary). And it seems to work. Thanks @LaurentMazare!

[LaurentMazare]

Closing this at it seems fixed now, feel free to re-open if needed.