module 'tensorflow' has no attribute 'absolute_import'

爱被打了一巴掌 2022-05-15 02:13 159阅读 0赞

![image][]  
这个error猝不及防，在查看了tensorflow API的版本后，做了如下更改

# from tensorflow import *
    from tensorflow.python import *

module ‘XXX’ has no attribute ‘XXX’  
解决方案：（1）module未导入成功；（2）版本容错问题。

**error分析原因**：  
本人环境是tensorflow 1.10  
因为有朋友在tensorflow1.3版本上并无此错误，故直接查看了1.3版本与出错的init.py之间的区别，即做出了上述尝试，成功。

from __future__ import absolute_import
    from __future__ import division
    from __future__ import print_function
    
    # pylint: disable=wildcard-import
    from tensorflow.python import *  # 出错参考点
    # pylint: enable=wildcard-import
    
    from tensorflow.python.util.lazy_loader import LazyLoader
    contrib = LazyLoader('contrib', globals(), 'tensorflow.contrib')
    del LazyLoader
    
    del absolute_import
    del division
    del print_function
    
    # These symbols appear because we import the python package which
    # in turn imports from tensorflow.core and tensorflow.python. They
    # must come from this module. So python adds these symbols for the
    # resolution to succeed.
    # pylint: disable=undefined-variable
    del python
    del core

对比出错init

from __future__ import absolute_import
    from tensorflow import *  # 与上面版本的init有出入
    
    from .sg_util import sg_opt
    from .sg_main import *
    from .sg_logging import *
    from .sg_train import *
    from .sg_queue import sg_producer_func
    from .sg_layer import sg_emb
    
    from . import sg_optimize
    from . import sg_data
    from . import sg_initializer
    
    from ._version import __version__
    
    
    __author__ = 'namju.kim@kakaocorp.com'
    
    
    #
    # augmenting tensorflow Variable and Tensor type by injecting custom methods
    #
    
    modules = ['sg_activation', 'sg_metric', 'sg_layer', 'sg_loss', 'sg_transform', 'sg_net']
    for mod in modules:
        sg_inject(__path__[0], mod)

[image]: /images/20220515/ce1b1a4297464c58ad4aa0af6fe15ed6.png