Training errors (Tensor vs Tuple)

[JianbangZ]

I encounter this error while attempting to train, anyone got ideas?
looks like the torch.nn.transofrmer forward func is retrunning Tensor, yet the code is expecting Tuple

File "bin/trainer.py", line 1046, in
main()
File "bin/trainer.py", line 1039, in main
run(rank=0, world_size=1, args=args)
File "bin/trainer.py", line 902, in run
scan_pessimistic_batches_for_oom(
File "bin/trainer.py", line 1004, in scan_pessimistic_batches_for_oom
_, loss, _ = compute_loss(
File "bin/trainer.py", line 451, in compute_loss
predicts, loss, metrics = model(
File "/home/zhangjb2/ml/audio/virtual_valle/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
File "/home/zhangjb2/ml/audio/vall-e/valle/models/valle.py", line 575, in forward
xy_dec, _ = self.ar_decoder(
File "/home/zhangjb2/ml/audio/virtual_valle/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
File "/home/zhangjb2/ml/audio/virtual_valle/lib/python3.8/site-packages/torch/nn/modules/transformer.py", line 219, in forward
target_type=src.dtype
AttributeError: 'tuple' object has no attribute 'dtype'