如何在TensorFlow2.0中解决RuntimeError？

Question

我想在Anaconda的GPU上运行TF2.0中的TF1.X程序，所以我对代码做了一些修改，但我会在下面的最后一行得到一个错误。

train_dataset = tf.data.Dataset.from_tensor_slices((X_train, y_train))
train_dataset = train_dataset.apply(tf.data.experimental.shuffle_and_repeat(10000, seed=0))
train_dataset = train_dataset.batch(self.batchsize)
train_dataset = train_dataset.apply(tf.data.experimental.prefetch_to_device(device))

错误消息是

Traceback (most recent call last):
  File "train.py", line 77, in <module>
    reward = manager.get_rewards(ModelGenerator, state_space.parse_state_space_list(action))
  File "C:\Users\user\tf\tf2.0_gpu_test\manager.py", line 83, in get_rewards
    train_dataset = train_dataset.apply(tf.data.experimental.prefetch_to_device(device))
  File "C:\Users\user\AppData\Local\Continuum\anaconda3\envs\tf_env\lib\site-packages\tensorflow_core\python\data\ops\dataset_ops.py", line 1369, in apply
    dataset = transformation_func(self)
  File "C:\Users\user\AppData\Local\Continuum\anaconda3\envs\tf_env\lib\site-packages\tensorflow_core\python\data\experimental\ops\prefetching_ops.py", line 54, in _apply_fn
    copy_to_device(target_device=device)).prefetch(buffer_size)
  File "C:\Users\user\AppData\Local\Continuum\anaconda3\envs\tf_env\lib\site-packages\tensorflow_core\python\data\ops\dataset_ops.py", line 1369, in apply
    dataset = transformation_func(self)
  File "C:\Users\user\AppData\Local\Continuum\anaconda3\envs\tf_env\lib\site-packages\tensorflow_core\python\data\experimental\ops\prefetching_ops.py", line 78, in _apply_fn
    source_device=source_device).with_options(options)
  File "C:\Users\user\AppData\Local\Continuum\anaconda3\envs\tf_env\lib\site-packages\tensorflow_core\python\data\experimental\ops\prefetching_ops.py", line 102, in __init__
    self._source_device = ops.convert_to_tensor(source_device)
  File "C:\Users\user\AppData\Local\Continuum\anaconda3\envs\tf_env\lib\site-packages\tensorflow_core\python\framework\ops.py", line 1184, in convert_to_tensor
    return convert_to_tensor_v2(value, dtype, preferred_dtype, name)
  File "C:\Users\user\AppData\Local\Continuum\anaconda3\envs\tf_env\lib\site-packages\tensorflow_core\python\framework\ops.py", line 1242, in convert_to_tensor_v2
    as_ref=False)
  File "C:\Users\user\AppData\Local\Continuum\anaconda3\envs\tf_env\lib\site-packages\tensorflow_core\python\framework\ops.py", line 1296, in internal_convert_to_tensor
    ret = conversion_func(value, dtype=dtype, name=name, as_ref=as_ref)
  File "C:\Users\user\AppData\Local\Continuum\anaconda3\envs\tf_env\lib\site-packages\tensorflow_core\python\framework\constant_op.py", line 286, in _constant_tensor_conversion_function
    return constant(v, dtype=dtype, name=name)
  File "C:\Users\user\AppData\Local\Continuum\anaconda3\envs\tf_env\lib\site-packages\tensorflow_core\python\framework\constant_op.py", line 227, in constant
    allow_broadcast=True)
  File "C:\Users\user\AppData\Local\Continuum\anaconda3\envs\tf_env\lib\site-packages\tensorflow_core\python\framework\constant_op.py", line 235, in _constant_impl
    t = convert_to_eager_tensor(value, ctx, dtype)
  File "C:\Users\user\AppData\Local\Continuum\anaconda3\envs\tf_env\lib\site-packages\tensorflow_core\python\framework\constant_op.py", line 96, in convert_to_eager_tensor
    return ops.EagerTensor(value, ctx.device_name, dtype)
RuntimeError: Can't copy Tensor with type string to device /job:localhost/replica:0/task:0/device:GPU:0.

如果有人能帮我那就太好了。

Answer 1

我能够在tensorflow-gpu==2.0.0版本中重现这个问题，如下所示-

代码-

import tensorflow_hub as hub
import tensorflow as tf 
print(tf.__version__)

embed = hub.load("https://tfhub.dev/google/universal-sentence-encoder-large/5")

with tf.device('GPU:0'):

    embeddings = embed(["The quick brown fox jumps over the lazy dog."])

print(embeddings)

输出-

2.0.0
---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
<ipython-input-4-e3758851cab1> in <module>()
      7 with tf.device('GPU:0'):
      8 
----> 9     embeddings = embed(["The quick brown fox jumps over the lazy dog."])
     10 
     11 print(embeddings)

11 frames
/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/framework/constant_op.py in convert_to_eager_tensor(value, ctx, dtype)
     94       dtype = dtypes.as_dtype(dtype).as_datatype_enum
     95   ctx.ensure_initialized()
---> 96   return ops.EagerTensor(value, ctx.device_name, dtype)
     97 
     98 

RuntimeError: Can't copy Tensor with type string to device /job:localhost/replica:0/task:0/device:GPU:0.

然而，这个问题似乎在tensorflow版本的2.2.0中得到了解决。

代码-

import tensorflow_hub as hub
import tensorflow as tf 
print(tf.__version__)

embed = hub.load("https://tfhub.dev/google/universal-sentence-encoder-large/5")

with tf.device('GPU:0'):

    embeddings = embed(["The quick brown fox jumps over the lazy dog."])

print("I Ran Successfully")

输出-

2.2.0
I Ran Successfully

你能把你的tensorflow版本升级到2.2.0吗？

希望这能回答你的问题。快乐学习。