inference another models like yolor_W6,E6,D6 error

[ehdrndd]

Detecting with yolor_p6 is fine.

But, another models like 'yolor_W6,E6,D6'

Traceback (most recent call last): File "<string>", line 1, in <module> File "C:\Users\KANG\anaconda3\envs\OD\lib\site-packages\torch\serialization.py", line 594, in load return _load(opened_zipfile, map_location, pickle_module, **pickle_load_args) File "C:\Users\KANG\anaconda3\envs\OD\lib\site-packages\torch\serialization.py", line 853, in _load result = unpickler.load() ModuleNotFoundError: No module named 'models.yolo'

I guess W6,E6,D6 is saved by torch 1.4. but now i using torch 1.7

Is there any solution not downgrading torch?

[WongKinYiu]

all of p6, w6, e6, and d6 are saved by either pytorch 1.8.0a0+52ea372 or 1.9.0a0+df837d0.

but models.yolo means ./models/yolo.py, it can be find in the repository.

[ehdrndd]

Thanks. but, i still don't know solution..

In debug mode.. print(torch.load('yolor_w6.pt', map_location=device)['model'])

Traceback (most recent call last): File "<string>", line 1, in <module> File "C:\Users\KANG\anaconda3\envs\OD\lib\site-packages\torch\serialization.py", line 594, in load return _load(opened_zipfile, map_location, pickle_module, **pickle_load_args) File "C:\Users\KANG\anaconda3\envs\OD\lib\site-packages\torch\serialization.py", line 853, in _load result = unpickler.load() ModuleNotFoundError: No module named 'models.yolo'

print(torch.load('yolor_p6.pt', map_location=device)['model'])

OrderedDict([('module_list.1.Conv2d.weight', tensor([[[[ 7.08389e...='cuda:0')), ('module_list.1.BatchN...m2d.weight', tensor([2.45703, 1.7...='cuda:0')), ('module_list.1.BatchNorm2d.bias', tensor([ 3.51172, 3...='cuda:0')), ('module_list.1.BatchN...nning_mean', tensor([ 3.52287e-03...='cuda:0')), ('module_list.1.BatchN...unning_var', tensor([3.51715e-03,...='cuda:0')), ('module_list.1.BatchN...es_tracked', tensor(0, device='cuda:0')), ('module_list.2.Conv2d.weight', tensor([[[[ 7.91168e...='cuda:0')), ('module_list.2.BatchN...m2d.weight', tensor([1.27051, 1.7...='cuda:0')), ('module_list.2.BatchNorm2d.bias', tensor([ 1.19336, -2...='cuda:0')), ('module_list.2.BatchN...nning_mean', tensor([-1.83984, -0...='cuda:0')), ('module_list.2.BatchN...unning_var', tensor([ 2.45703, 0...='cuda:0')), ('module_list.2.BatchN...es_tracked', tensor(0, device='cuda:0')), ('module_list.3.Conv2d.weight', tensor([[[[ 0.00136]...='cuda:0')), ('module_list.3.BatchN...m2d.weight', tensor([1.99707, 2.0...='cuda:0')), ...])

And models/yolo.py.
I can't find in your repo..

[WongKinYiu]

https://github.com/WongKinYiu/yolor/blob/paper/models/yolo.py

maybe you are using code of main branch and weights of paper branch.

[ehdrndd]

Oh, you're right. thanks.

I moved paper branch. it works!

[ShoaibSajid]

https://github.com/WongKinYiu/yolor/blob/paper/models/yolo.py

maybe you are using code of main branch and weights of paper branch.

So we cannot use weights of paper branch in main branch and vice versa, even for our custom trained models ?

[RizwanMunawar]

https://github.com/WongKinYiu/yolor/blob/paper/models/yolo.py maybe you are using code of main branch and weights of paper branch.

So we cannot use weights of paper branch in main branch and vice versa, even for our custom trained models ?

Thanks, it worked. Yolor-p6 main branch weights: yolor-p6.pt Yolor-p6 paper branch weights: yolor_p6.pt