Encounting NotImplementedError when finetuning a Single-Task .pth model

[njzjz]

i take a pretrained-single-task model for finetuning I just modified the training data part of the model's input.json. The finetune command: dp --pt train input.json -t dpa2.pth --use-pretrain-script.

The error is as follow:

Traceback (most recent call last):
File "/home/yanli/deepmd-kit-v3/bin/dp", line 10, in
sys.exit(main())
^^^^^^
File "/home/yanli/deepmd-kit-v3/lib/python3.11/site-packages/deepmd/main.py", line 923, in main
deepmd_main(args)
File "/home/yanli/deepmd-kit-v3/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/init.py", line 346, in wrapper
return f(*args, **kwargs)
^^^^^^^^^^^^^^^^^^
File "/home/yanli/deepmd-kit-v3/lib/python3.11/site-packages/deepmd/pt/entrypoints/main.py", line 562, in main
train(FLAGS)
File "/home/yanli/deepmd-kit-v3/lib/python3.11/site-packages/deepmd/pt/entrypoints/main.py", line 267, in train
config["model"], finetune_links = get_finetune_rules(
^^^^^^^^^^^^^^^^^^^
File "/home/yanli/deepmd-kit-v3/lib/python3.11/site-packages/deepmd/pt/utils/finetune.py", line 140, in get_finetune_rules
if "model" in state_dict:
^^^^^^^^^^^^^^^^^^^^^
File "/home/yanli/deepmd-kit-v3/lib/python3.11/site-packages/torch/jit/_script.py", line 868, in contains
return self.forward_magic_method("contains", key)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yanli/deepmd-kit-v3/lib/python3.11/site-packages/torch/jit/_script.py", line 855, in forward_magic_method
raise NotImplementedError()
NotImplementedError

The input.script is here:
input.json

Is this because the pth model is used instead of the pt model?

Originally posted by @darkgezi in #4255

[njzjz]

Thought: we do need a universal load model API. There are many torch.load and torch.jit.load usages in the code, which are in a mass.