ImportError: /tmp/torch_extensions/inplace_abn/inplace_abn.so: undefined symbol: _ZN3c104impl25tls_local_tensor_type_setEv about self-correction-human-parsing HOT 5 CLOSED

Apologies. Using torch==1.2.0

from self-correction-human-parsing.

@rkhilnani9 I have got the same issue, but when I switched to use torch 1.5.1, torchvision 0.6.1. The issue no longer exists.

from self-correction-human-parsing.

@VyBui Thanks! That error is no longer there and I am almost able to train the model.

Faced another error in train.py though. Would you happen to know anything about this?

Traceback (most recent call last):
File "train.py", line 231, in
main()
File "train.py", line 94, in main
AugmentCE2P = networks.init_model(args.arch, num_classes=args.num_classes, pretrained=args.imagenet_pretrain) File "/rapids/notebooks/atcdata/rkhilnan/Self-Correction-Human-Parsing/networks/init.py", line 13, in init_model return __factory[name](*args, **kwargs)
File "/rapids/notebooks/atcdata/rkhilnan/Self-Correction-Human-Parsing/networks/AugmentCE2P.py", line 336, in resnet101
initialize_pretrained_model(model, settings, pretrained)
File "/rapids/notebooks/atcdata/rkhilnan/Self-Correction-Human-Parsing/networks/AugmentCE2P.py", line 330, in initialize_pretrained_model
model.load_state_dict(new_params)
File "/conda/lib/python3.7/site-packages/torch/nn/modules/module.py", line 847, in load_state_dict self.class.name, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for ResNet:
Unexpected key(s) in state_dict: "state_dict".

from self-correction-human-parsing.

So I want to try out the exp-schp-pascal-person.pth model, and I have downloaded the weights and saved it in pretrained_model folder inside my repo. Do I also have to download resnet-101.pth?

from self-correction-human-parsing.

@rkhilnani9 I have not seen that error yet.

Yes, you have to download resnet101-imagenet.pth and place it in the pretrained_model

from self-correction-human-parsing.