合成器训练停止后，如何用于其他的训练

之前都是使用作者或者别人打包好的训练集继续训练，但我用自己训练的数据集训练后，继续训练会出现抱错。请问是什么原因。报错如下：
Traceback (most recent call last):
File "D:\mockingbird2\MockingBird-main\synthesizer_train.py", line 37, in
train(**vars(args))
File "D:\mockingbird2\MockingBird-main\synthesizer\train.py", line 215, in train
optimizer.step()
File "C:\Program Files\Python39\lib\site-packages\torch\optim\optimizer.py", line 109, in wrapper
return func(*args, **kwargs)
File "C:\Program Files\Python39\lib\site-packages\torch\autograd\grad_mode.py", line 27, in decorate_context
return func(*args, **kwargs)
File "C:\Program Files\Python39\lib\site-packages\torch\optim\adam.py", line 157, in step
adam(params_with_grad,
File "C:\Program Files\Python39\lib\site-packages\torch\optim\adam.py", line 213, in adam
func(params,
File "C:\Program Files\Python39\lib\site-packages\torch\optim\adam.py", line 255, in _single_tensor_adam
assert not step_t.is_cuda, "If capturable=False, state_steps should not be CUDA tensors."
AssertionError: If capturable=False, state_steps should not be CUDA tensors.

Start with synthesizer_train.py and synthesizer/train.py at line 215, then reproduce the failure while continuing training with the custom dataset. Trace how the Adam optimizer state is created or restored before optimizer.step() and compare it with the PyTorch assertion about CUDA state_steps. Done means resumed training proceeds without this assertion.

Written by the indexing model from the issue text.