Maybe a bug about path_lm_ft.py

[xuhuanchen]

When I run this python file on my own datasets.
It reported a bug about this code:
if global_step in ckpt_steps:
ckpt_path = os.path.join(args.output_dir, f'step_{global_step}.bin')
torch.save(model, ckpt_path) ###this one report a bug.

INFO about this bug:
File "path_lm_ft_2.py", line 1320, in
train(args, train_dataset, model, criterion, tokenizer, load_id=args.load_id, load_element_id=args.load_element_id)
File "path_lm_ft_2.py", line 833, in train
torch.save(model, ckpt_path)
File "/users7/hcxu/miniconda3/lib/python3.7/site-packages/torch/serialization.py", line 379, in save
_save(obj, opened_zipfile, pickle_module, pickle_protocol)
File "/users7/hcxu/miniconda3/lib/python3.7/site-packages/torch/serialization.py", line 484, in _save
pickler.dump(obj)
TypeError: can't pickle torch._C.ScriptFunction objects

Now I just used "torch.save(model.state_dict(), ckpt_path)" to avoid this bug, but I don`t know why this bug occurs.