Serialization 如何调试保存模型类型错误：can'；t pickle SwigPyObject对象？

我试图保存一个模型，它是从

nn.Module

继承的类的对象。它训练得很好，没有问题，但当我尝试运行代码时：

    torch.save(
        obj=model,
        f=os.path.join(tensorboard_writer.get_logdir(), 'model.ckpt'))

我收到错误：

TypeError:无法pickle SwigPyObject对象

我不知道什么是

SwigPyObject

对象。如何调试此错误以保存模型

完全回溯：

Traceback (most recent call last):
  File "/usr/local/lib/python3.6/dist-packages/torch/serialization.py", line 149, in _with_file_like
    return body(f)
  File "/usr/local/lib/python3.6/dist-packages/torch/serialization.py", line 224, in <lambda>
    return _with_file_like(f, "wb", lambda f: _save(obj, f, pickle_module, pickle_protocol))
  File "/usr/local/lib/python3.6/dist-packages/torch/serialization.py", line 296, in _save
    pickler.dump(obj)
TypeError: can't pickle SwigPyObject objects

然后，我构建并保存模型：

    tensorboard_writer = SummaryWriter()

    model = RecurrentModel(
        core_str='rnn',
        core_kwargs={},
        tensorboard_writer=tensorboard_writer)

    torch.save(
        obj=model,
        f=os.path.join(tensorboard_writer.get_logdir(), 'model.ckpt')
    )

---

PyTorch论坛上有人澄清了这一点。SummaryWriter有一个文件句柄，通常无法序列化。问题在于：

self.tensorboard\u writer=tensorboard\u writer

    tensorboard_writer = SummaryWriter()

    model = RecurrentModel(
        core_str='rnn',
        core_kwargs={},
        tensorboard_writer=tensorboard_writer)

    torch.save(
        obj=model,
        f=os.path.join(tensorboard_writer.get_logdir(), 'model.ckpt')
    )