Hi,
after the model training I’ve got saved training_args.bin. However, loading them with torch.load(...) as suggested in How to load training_args results in an error:
Am I doing something wrong or is it a bug?
Thanks!
A mistake on my side, though may be relevant to somebody. I overwrote the class TrainingArguments to add my own one, and the file with the new class (training_arguments.py in my case) must be present in the same directory from which you are uploading the arguments. Closing this.
1 Like