UnpicklingError: invalid load key, '<'.

When running in the Colab Notebook : model = model_type.model(backbone=backbone(pretrained=True), num_classes=len(parser.class_map), **extra_args)

Getting the following error. Hope you may kindly suggest how to fix this!

load checkpoint from local path: checkpoints/vfnet/vfnet_r50_fpn_mstrain_2x_coco_20201027-7cc75bd2.pth
---------------------------------------------------------------------------
UnpicklingError                           Traceback (most recent call last)
[<ipython-input-10-696c7b7efa47>](https://localhost:8080/#) in <module>()
      1 # Instantiate the model
----> 2 model = model_type.model(backbone=backbone(pretrained=True), num_classes=len(parser.class_map), **extra_args)

7 frames
[/usr/local/lib/python3.7/dist-packages/torch/serialization.py](https://localhost:8080/#) in _legacy_load(f, map_location, pickle_module, **pickle_load_args)
    775             "functionality.")
    776 
--> 777     magic_number = pickle_module.load(f, **pickle_load_args)
    778     if magic_number != MAGIC_NUMBER:
    779         raise RuntimeError("Invalid magic number; corrupt file?")

UnpicklingError: invalid load key, '<'.

[dnth]

Hi @jlvahldiek we are in the process of fixing this on the IceVision repo.

If you’re in a hurry, here is a Colab notebook as a workaround to make VFNet work again.

https://colab.research.google.com/gist/dnth/aea556c86ee3ca2e812ceb5eb8968caa/getting_started_object_detection.ipynb

[dnth]

This still seems to be an issue.

I was running getting_started_object_detection.ipynb that I just got from git today.

load checkpoint from local path: checkpoints/vfnet/vfnet_r50_fpn_mstrain_2x_coco_20201027-7cc75bd2.pth
---------------------------------------------------------------------------
UnpicklingError                           Traceback (most recent call last)
Input In [29], in <cell line: 2>()
      1 # Instantiate the model
----> 2 model = model_type.model(backbone=backbone(pretrained=True), num_classes=len(parser.class_map), **extra_args)

File /usr/local/lib/python3.9/dist-packages/icevision/models/mmdet/common/bbox/single_stage/model.py:20, in model(backbone, num_classes, checkpoints_path, force_download, cfg_options)
     12 def model(
     13     backbone: MMDetBackboneConfig,
     14     num_classes: int,
   (...)
     17     cfg_options=None,
     18 ) -> nn.Module:
---> 20     return build_model(
     21         model_type="one_stage_detector_bbox",
     22         backbone=backbone,
     23         num_classes=num_classes,
     24         pretrained=backbone.pretrained,
     25         checkpoints_path=checkpoints_path,
     26         force_download=force_download,
     27         cfg_options=cfg_options,
     28     )

File /usr/local/lib/python3.9/dist-packages/icevision/models/mmdet/common/utils.py:79, in build_model(model_type, backbone, num_classes, pretrained, checkpoints_path, force_download, cfg_options)
     76 _model.init_weights()
     78 if pretrained and (weights_path is not None):
---> 79     load_checkpoint(_model, str(weights_path))
     81 _model.param_groups = MethodType(param_groups, _model)
     82 _model.cfg = cfg  # save the config in the model for convenience

File /usr/local/lib/python3.9/dist-packages/mmcv/runner/checkpoint.py:528, in load_checkpoint(model, filename, map_location, strict, logger, revise_keys)
    503 def load_checkpoint(model,
    504                     filename,
    505                     map_location=None,
    506                     strict=False,
    507                     logger=None,
    508                     revise_keys=[(r'^module\.', '')]):
    509     """Load checkpoint from a file or URI.
    510 
    511     Args:
   (...)
    526         dict or OrderedDict: The loaded checkpoint.
    527     """
--> 528     checkpoint = _load_checkpoint(filename, map_location, logger)
    529     # OrderedDict is a subclass of dict
    530     if not isinstance(checkpoint, dict):

File /usr/local/lib/python3.9/dist-packages/mmcv/runner/checkpoint.py:467, in _load_checkpoint(filename, map_location, logger)
    450 def _load_checkpoint(filename, map_location=None, logger=None):
    451     """Load checkpoint from somewhere (modelzoo, file, url).
    452 
    453     Args:
   (...)
    465            information, which depends on the checkpoint.
    466     """
--> 467     return CheckpointLoader.load_checkpoint(filename, map_location, logger)

File /usr/local/lib/python3.9/dist-packages/mmcv/runner/checkpoint.py:245, in CheckpointLoader.load_checkpoint(cls, filename, map_location, logger)
    242 class_name = checkpoint_loader.__name__
    243 mmcv.print_log(
    244     f'load checkpoint from {class_name[10:]} path: {filename}', logger)
--> 245 return checkpoint_loader(filename, map_location)

File /usr/local/lib/python3.9/dist-packages/mmcv/runner/checkpoint.py:262, in load_from_local(filename, map_location)
    260 if not osp.isfile(filename):
    261     raise IOError(f'{filename} is not a checkpoint file')
--> 262 checkpoint = torch.load(filename, map_location=map_location)
    263 return checkpoint

File /usr/local/lib/python3.9/dist-packages/torch/serialization.py:608, in load(f, map_location, pickle_module, **pickle_load_args)
    606             return torch.jit.load(opened_file)
    607         return _load(opened_zipfile, map_location, pickle_module, **pickle_load_args)
--> 608 return _legacy_load(opened_file, map_location, pickle_module, **pickle_load_args)

File /usr/local/lib/python3.9/dist-packages/torch/serialization.py:777, in _legacy_load(f, map_location, pickle_module, **pickle_load_args)
    771 if not hasattr(f, 'readinto') and (3, 8, 0) <= sys.version_info < (3, 8, 2):
    772     raise RuntimeError(
    773         "torch.load does not work with file-like objects that do not implement readinto on Python 3.8.0 and 3.8.1. "
    774         f"Received object of type \"{type(f)}\". Please update to Python 3.8.2 or newer to restore this "
    775         "functionality.")
--> 777 magic_number = pickle_module.load(f, **pickle_load_args)
    778 if magic_number != MAGIC_NUMBER:
    779     raise RuntimeError("Invalid magic number; corrupt file?")

UnpicklingError: invalid load key, '<'.


cat checkpoints/vfnet/vfnet_r50_fpn_mstrain_2x_coco_20201027-7cc75bd2.pth
<?xml version="1.0" encoding="UTF-8"?>
<Error>
  <Code>AccessDenied</Code>
  <Message>Access denied by bucket policy.</Message>
  <RequestId>62FAB855D4FE7D31395A743C</RequestId>
  <HostId>openmmlab.oss-cn-hangzhou.aliyuncs.com</HostId>
  <Bucket>openmmlab</Bucket>
  <User>nosuchuser</User>
</Error>

Which version of IceVision are you running on? Try installing IceVision from the master branch and see if this still happens.