ModelCore.calc_criterion is not callable when running on multi-GPUs with torch.nn.DataParallel

[WenjieDu]

1. System Info

pypots v0.13

2. Information

• The official example scripts
• My own created scripts

3. Reproduction

Call calc_criterion() when the model is running on multiple GPUs

PyPOTS/pypots/nn/modules/base_model_core.py

Line 48 in 0259a6c

def calc_criterion(self, inputs: dict) -> dict:

4. Expected behavior

It fails because the model is wrapped in DataParallel, which only provides forward() for parallel running.

5. Your contribution

PR is on the way