Error running test.py and train.py

[matheusfalcaopinto]

I'm running in google colab:

!python test.py --data data/coco.yaml --img 1280 --batch 32 --conf 0.001 --iou 0.65 --device 0 --cfg cfg/yolor_p6.cfg --weights yolor_p6.pt --name yolor_p6_val

or

!python detect.py --source inference/images/horses.jpg --cfg cfg/yolor_p6.cfg --weights yolor_p6.pt --conf 0.25 --img-size 1280 --device 0

or

python train.py --batch-size 8 --img 1280 1280 --data coco.yaml --cfg cfg/yolor_p6.cfg --weights '' --device 0 --name yolor_p6 --hyp hyp.scratch.1280.yaml --epochs 300

and I get:

_pickle.UnpicklingError: invalid load key, '<'.

Any suggestions? Thanks!

[Faisalsouz]

Having same problem @matheusfalcaopinto have you found solution or anybody else can suggest where exactly problem lies.Thanks

[matheusfalcaopinto]

Nothing for now :/ Can't train or test, just run standard pre-treined weights.

[JoseIgnDiazVilla]

I'm getting the same error. Tho I suspect it's from a version mismatch between torch and some other package.

[matheusfalcaopinto]

I'm getting the same error. Tho I suspect it's from a version mismatch between torch and some other package.

Yes, but here the more I tried to change the version of packages, more errors I get haha

[Faisalsouz]

I found a solution: the reason of the error is that the file pretrain.sh bash is not downloading the weights properly. As script is trying to crease cookies and then supply it API key. The easy way is to just go here "https://drive.google.com/uc?export=download&id=1Tdn3yqpZ79X7R1Ql0zNlNScB1Dv9Fp76" and download weight manually.

[ZBC043]

I found a solution: the reason of the error is that the file pretrain.sh bash is not downloading the weights properly. As script is trying to crease cookies and then supply it API key. The easy way is to just go here "https://drive.google.com/uc?export=download&id=1Tdn3yqpZ79X7R1Ql0zNlNScB1Dv9Fp76" and download weight manually.

Hi @Faisalsouz , it seems like I cannot open this link, can you give me some help?