Error in pipeline when we try to read checkpoint models

[wendiya]

Recently I tried to train the model in the repo using python train_network.py -d $dataset_specs_json_file

and when I run it, I keep getting issues regarding models in ckp/airways folder. Pipeline expects files inside this folder to be actually files, but I am given folders and pipeline raises issue.

Could you please help me to resolve this issue?

[wendiya]

@Dhanasrepo could you please help me with this issue if you know what is the problem?
Would very much appreaciate it! Thank you for helping with my previous issue.

[Dhanasrepo]

Thats because the model weights are uploaded as binary files. You can convert them into a .pt file or a similar format by unpickling the binaries using PyTorch. like torch.frombuffer() to load the binaries.