error when "torch.load()"

[sveits6]

❔Question

Traceback (most recent call last):
File "D:\Envs\pyt1.4\lib\site-packages\torch\serialization.py", line 527, in load
with _open_zipfile_reader(f) as opened_zipfile:
File "D:\Envs\pyt1.4\lib\site-packages\torch\serialization.py", line 224, in init
super(_open_zipfile_reader, self).init(torch.C.PyTorchFileReader(name_or_buffer))
RuntimeError: version <= kMaxSupportedFileFormatVersion INTERNAL ASSERT FAILED at ..\caffe2\serialize\inline_container.cc:132, please report a bug to PyTorch. Attempted to read a PyTorch file with version 3, but the maximum supported version for reading is 2. Your PyTorch installation may be too old. (init at ..\caffe2\serialize\inline_container.cc:132)
(no backtrace available)

How to solve it?This just happen on the newest weights file of your google drive.The weights file of last version is ok.

Additional context

[glenn-jocher]

@sveits6 I don't recognize that command, and we don't assist in debugging custom code.

[mfruhner]

Hello, @glenn-jocher, I am having the same issue. It happens with the new v2.0, when starting a training in the ultralytics docker container. I pulled the latest container and started a training. It starts to download the pt file (yolov5m.pt) and then errors out:

Downloading https://drive.google.com/uc?export=download&id=1vobuEExpWQVpXExsJ2w-Mbf3HJjWkQJr as yolov5m.pt... Done (5.7s)
Traceback (most recent call last):
  File "train.py", line 469, in <module>
    train(hyp, tb_writer, opt, device)
  File "train.py", line 128, in train
    ckpt = torch.load(weights, map_location=device)  # load checkpoint
  File "/opt/conda/lib/python3.6/site-packages/torch/serialization.py", line 585, in load
    with _open_zipfile_reader(f) as opened_zipfile:
  File "/opt/conda/lib/python3.6/site-packages/torch/serialization.py", line 245, in __init__
    super(_open_zipfile_reader, self).__init__(torch._C.PyTorchFileReader(name_or_buffer))
RuntimeError: version_ <= kMaxSupportedFileFormatVersion INTERNAL ASSERT FAILED at ../caffe2/serialize/inline_container.cc:132, please report a bug to PyTorch. Attempted to read a PyTorch file with version 3, but the maximum supported version for reading is 2. Your PyTorch installation may be too old.

Were those new weights trained with pytorch 1.6? The docker container uses 1.5.x

Greetings and Thanks,
Maik

[mfruhner]

Just checked it, by installing torch==1.6.0 and torchvision==0.7.0.
It now loads correctly and starts to train. So the requirements.txt needs to be updated and the docker container rebuilt accordingly. Cheers.

[glenn-jocher]

@mfruhner thanks! Yes we are in the process of updating requirements.txt to 1.6, and will be pushing native mixed precision training PR as well.

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[bbyuksel]

Just checked it, by installing torch==1.6.0 and torchvision==0.7.0. It now loads correctly and starts to train. So the requirements.txt needs to be updated and the docker container rebuilt accordingly. Cheers.

This solved my same problem