TypeError: numpy.ndarray or cuda.ndarray are expected.

最新推荐文章于 2025-05-23 12:46:36 发布

原创最新推荐文章于 2025-05-23 12:46:36 发布 · 1.9k 阅读

0 ·

CC 4.0 BY-SA版权

Python 专栏收录该内容

37 篇文章

订阅专栏

本文详细记录了一个在使用Chainer进行模型训练时遇到的CUDA内存问题，具体表现为在训练过程中抛出TypeError异常，指出实际传入的是cupy.cuda.memory.MemoryPointer类型而非期望的numpy.ndarray或cuda.ndarray。通过将cupy类型转换为numpy类型，成功解决了该问题。

遇到一个bug

Will finalize trainer extensions and updater before reraising the exception.
Traceback (most recent call last):
  File "/home/zlee/PycharmProjects/hyperface-youtube/hyperface-master/scripts/train.py", line 176, in <module>
    trainer.run()
  File "/home/zlee/PycharmProjects/hyperface-youtube/venv/local/lib/python2.7/site-packages/chainer/training/trainer.py", line 329, in run
    six.reraise(*sys.exc_info())
  File "/home/zlee/PycharmProjects/hyperface-youtube/venv/local/lib/python2.7/site-packages/chainer/training/trainer.py", line 315, in run
    update()
  File "/home/zlee/PycharmProjects/hyperface-youtube/venv/local/lib/python2.7/site-packages/chainer/training/updaters/standard_updater.py", line 165, in update
    self.update_core()
  File "/home/zlee/PycharmProjects/hyperface-youtube/venv/local/lib/python2.7/site-packages/chainer/training/updaters/standard_updater.py", line 179, in update_core
    optimizer.update(loss_func, **in_arrays)
  File "/home/zlee/PycharmProjects/hyperface-youtube/venv/local/lib/python2.7/site-packages/chainer/optimizer.py", line 680, in update
    loss = lossfun(*args, **kwds)
  File "/home/zlee/PycharmProjects/hyperface-youtube/hyperface-master/scripts/models.py", line 114, in __call__
    h_landmark *= _disconnect(m_landmark)
  File "/home/zlee/PycharmProjects/hyperface-youtube/hyperface-master/scripts/models.py", line 22, in _disconnect
    return chainer.Variable(x.data)
  File "/home/zlee/PycharmProjects/hyperface-youtube/venv/local/lib/python2.7/site-packages/chainer/variable.py", line 505, in __init__
    raise TypeError(msg)
TypeError: numpy.ndarray or cuda.ndarray are expected.
Actual: <type 'cupy.cuda.memory.MemoryPointer'>

在调用函数时出现的问题：

def _disconnect(x):
    print("zlee", type(x), type(x.data))
    return chainer.Variable(x.data)

我打印了一下，是如下类型：

('zlee', <type 'cupy.core.core.ndarray'>, <type 'cupy.cuda.memory.MemoryPointer'>)

改正的方案如下所示：

def _disconnect(x):
    x = cupy.asnumpy(x)
    print("zlee : ", type(x),type(x.data))
    return chainer.Variable(x.data)

另外加入 import cupy