报错：RuntimeError: one of the variables needed for gradient

问题遇到的现象和发生背景

复现CycN-Net时出现报错：RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation

用代码块功能插入代码，请勿粘贴截图

在loss.backward(retain_graph=True)处报错

Traceback (most recent call last):
  File "/root/autodl-tmp/N-Net_and_CycNet-master/CycN-Net/main_train_CycNnet.py", line 158, in 
    loss.backward(retain_graph=True)
  File "/root/miniconda3/envs/my-env/lib/python3.6/site-packages/torch/_tensor.py", line 307, in backward
    torch.autograd.backward(self, gradient, retain_graph, create_graph, inputs=inputs)
  File "/root/miniconda3/envs/my-env/lib/python3.6/site-packages/torch/autograd/__init__.py", line 156, in backward
    allow_unreachable=True, accumulate_grad=True)  # allow_unreachable flag
运行结果及报错内容

第一行加入torch.set_grad_enabled(True)后，报错为

[W python_anomaly_mode.cpp:104] Warning: Error detected in PreluBackward0. Traceback of forward call that caused the error:
  File "/root/autodl-tmp/N-Net_and_CycNet-master/CycN-Net/main_train_CycNnet.py", line 146, in 
    prediction = model( img_seq_1, img_seq_2, img_seq_3, prior )
  File "/root/miniconda3/envs/my-env/lib/python3.6/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
    return forward_call(*input, **kwargs)
  File "/root/autodl-tmp/N-Net_and_CycNet-master/CycN-Net/model_CycNnet.py", line 302, in forward
    out = self.finalconv(Tconv_7) # 508@16 --> 512@8 --> 512@1
  File "/root/miniconda3/envs/my-env/lib/python3.6/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
    return forward_call(*input, **kwargs)
  File "/root/miniconda3/envs/my-env/lib/python3.6/site-packages/torch/nn/modules/container.py", line 141, in forward
    input = module(input)
  File "/root/miniconda3/envs/my-env/lib/python3.6/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
    return forward_call(*input, **kwargs)
  File "/root/miniconda3/envs/my-env/lib/python3.6/site-packages/torch/nn/modules/activation.py", line 1076, in forward
    return F.prelu(input, self.weight)
  File "/root/miniconda3/envs/my-env/lib/python3.6/site-packages/torch/nn/functional.py", line 1500, in prelu
    return torch.prelu(input, weight)
 (function _print_stack)
我的解答思路和尝试过的方法

搜索了PreluBackward0，但是没找到