colab上的tensorflow文本禁用GPU答案

【问题标题】：tensorflow_text disables GPU on colabcolab上的tensorflow文本禁用GPU
【发布时间】：2021-08-13 20:45:13
【问题描述】：

昨天一切都与变压器一起工作。现在我无法使用 colab 上提供的 GPU。这有效：

%tensorflow_version 2.x

!pip install -q tensorflow_datasets
!pip install -q tensorflow_text
import tensorflow as tf
import tensorflow_datasets as tfds
#import tensorflow_text as text
print(tf.__version__)
device_name = tf.test.gpu_device_name()
if device_name != '/device:GPU:0':
  raise SystemError('GPU device not found')
print('Found GPU at: {}'.format(device_name))

输出：

2.4.1
Found GPU at: /device:GPU:0

如果我取消注释 tensorflow_text 导入，则：


!pip install -q tensorflow_datasets
!pip install -q tensorflow_text
import tensorflow as tf
import tensorflow_datasets as tfds
import tensorflow_text as text
print(tf.__version__)
device_name = tf.test.gpu_device_name()
if device_name != '/device:GPU:0':
  raise SystemError('GPU device not found')
print('Found GPU at: {}'.format(device_name))

输出：

---------------------------------------------------------------------------
NotFoundError                             Traceback (most recent call last)
<ipython-input-8-adc8722d8d1e> in <module>()
      5 import tensorflow as tf
      6 import tensorflow_datasets as tfds
----> 7 import tensorflow_text as text
      8 print(tf.__version__)
      9 device_name = tf.test.gpu_device_name()

3 frames
/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/load_library.py in load_op_library(library_filename)
     55   Raises:
     56     RuntimeError: when unable to load the library or get the python wrappers.
---> 57   """
     58   lib_handle = py_tf.TF_LoadLibrary(library_filename)
     59   try:

NotFoundError: /usr/local/lib/python3.7/dist-packages/tensorflow_text/python/metrics/_text_similarity_metric_ops.so: undefined symbol: _ZN10tensorflow15OpKernelContext15allocate_outputEN4absl14lts_2020_09_2311string_viewERKNS_11TensorShapeEPPNS_6TensorE

如果我终止 GPU 会话并再次连接到托管运行时，我会得到导入的 tensorflow_text：

2.5.0
---------------------------------------------------------------------------
SystemError                               Traceback (most recent call last)
<ipython-input-1-adc8722d8d1e> in <module>()
      9 device_name = tf.test.gpu_device_name()
     10 if device_name != '/device:GPU:0':
---> 11   raise SystemError('GPU device not found')
     12 print('Found GPU at: {}'.format(device_name))

SystemError: GPU device not found

由于某种原因，现在版本已更改为 2.5.0 tensorflow_text 可以导入但未找到 GPU。

【问题讨论】：

标签： tensorflow text google-colaboratory tensorflow2.0

【解决方案1】：

我现在强制执行旧的 tensorflow_text 版本：

!pip install -q tensorflow_text==2.4.1

这似乎有效！

%tensorflow_version 2.x

!pip install -q tensorflow_datasets
!pip install -q tensorflow_text==2.4.1

import tensorflow as tf
import tensorflow_datasets as tfds
import tensorflow_text as text
print(tf.__version__)
device_name = tf.test.gpu_device_name()
if device_name != '/device:GPU:0':
  raise SystemError('GPU device not found {}'.format(device_name))
print('Found GPU at: {}'.format(device_name))

输出：

2.4.1
Found GPU at: /device:GPU:0

感谢支持！

【讨论】：