tensorflow-keras-pytorch-cuda-cudnn版本[不定期更新]

最新推荐文章于 2024-09-06 22:15:28 发布

hellopbc

最新推荐文章于 2024-09-06 22:15:28 发布

阅读量290

点赞数

分类专栏： ML and DL 文章标签： tensorflow cuda pytorch gpu

本文链接：https://blog.csdn.net/qq_37774098/article/details/111225326

版权

ML and DL 专栏收录该内容

31 篇文章 1 订阅

订阅专栏

"本文详细介绍了TensorFlow 1.3.0与Keras 2.1.2、1.12与2.2.4的版本搭配，以及CUDA、CUDNN的关系。遇到'No module named 'tensorflow.python.eager'"错误时，提供了解决方案。涵盖官方推荐和环境配置要点。"

摘要由CSDN通过智能技术生成

文章目录

tensorflow-keras-pytorch-cuda-cudnn版本[不定期更新]
note
error
- ModuleNotFoundError: No module named 'tensorflow.python.eager'

tensorflow-keras-pytorch-cuda-cudnn版本[不定期更新]

官网

cuda（cudaToolkit）与NVIDIA显卡驱动的版本关系

tensorflow最新版本的安装要求

tensorflow-gpu与cuda、cudnn的版本关系

tensorflow和python的版本关系

cudnn与cuda的版本关系

tensorflow和keras版本关系

tensorflow-keras-cuda-cudnn

1. tensorflow=1.3.0 keras=2.1.2

tensorflow 1.3.0 0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free
tensorflow-base 1.3.0 py36h5293eaa_1 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free
tensorflow-tensorboard 1.5.1 py36hf484d3e_1 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main

keras 2.1.2 py36_0 https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main

conda install keras
conda install keras=2.1.2

2. tensorflow=1.12 keras=2.2.4 cuda=9.2 cudnn=7.6.5

numpy为1.16

tensorflow-gpu]-cuda-cudnn

pytorch-cuda-cudnn

note

tensorflow1.12；keras2.2.4；不需用.compat.v1，到了tensorflow1.14；keras2.2.5需要

config = tf.compat.v1.ConfigProto()
config.gpu_options.allow_growth = True
# config.gpu_options.allow_growth = False
sess = tf.compat.v1.Session(config=config)
KTF.set_session(sess)
# Helper functions
# K.set_image_dim_ordering('tf')   # 1.12-2.2.4可用
# data_format：字符串。'channels_first'或'channels_last'
K.set_image_data_format('channels_last')

error

ModuleNotFoundError: No module named ‘tensorflow.python.eager’

**前提：**conda默认安装keras，即conda install keras

环境：

  backports          anaconda/pkgs/main/noarch::backports-1.0-pyhd3eb1b0_2
  backports.weakref  anaconda/pkgs/free/linux-64::backports.weakref-1.0rc1-py36_0
  bleach             anaconda/pkgs/free/linux-64::bleach-1.5.0-py36_0
  h5py               anaconda/pkgs/main/linux-64::h5py-2.10.0-py36hd6299e0_1
  hdf5               anaconda/pkgs/main/linux-64::hdf5-1.10.6-hb1b8bf9_0
  html5lib           anaconda/pkgs/free/linux-64::html5lib-0.9999999-py36_0
  importlib-metadata anaconda/pkgs/main/noarch::importlib-metadata-2.0.0-py_1
  keras              anaconda/pkgs/main/linux-64::keras-2.3.1-0				#########################################
  keras-applications anaconda/pkgs/main/noarch::keras-applications-1.0.8-py_1
  keras-base         anaconda/pkgs/main/linux-64::keras-base-2.3.1-py36_0
  keras-preprocessi~ anaconda/pkgs/main/noarch::keras-preprocessing-1.1.0-py_1
  libprotobuf        anaconda/pkgs/main/linux-64::libprotobuf-3.13.0.1-hd408876_0
  markdown           anaconda/pkgs/main/linux-64::markdown-3.3.3-py36h06a4308_0
  protobuf           anaconda/pkgs/main/linux-64::protobuf-3.13.0.1-py36he6710b0_1
  pyyaml             anaconda/pkgs/main/linux-64::pyyaml-5.3.1-py36h7b6447c_1
  tensorflow         anaconda/pkgs/free/linux-64::tensorflow-1.3.0-0		###########################################
  tensorflow-base    anaconda/pkgs/free/linux-64::tensorflow-base-1.3.0-py36h5293eaa_1
  tensorflow-tensor~ anaconda/pkgs/main/linux-64::tensorflow-tensorboard-1.5.1-py36hf484d3e_1
  werkzeug           anaconda/pkgs/main/noarch::werkzeug-1.0.1-py_0
  yaml               anaconda/pkgs/main/linux-64::yaml-0.2.5-h7b6447c_0
  zipp               anaconda/pkgs/main/noarch::zipp-3.4.0-pyhd3eb1b0_0

报错：

ModuleNotFoundError: No module named 'tensorflow.python.eager

解决：

tensorflow提示：No module named ''tensorflow.python.eager".

conda install keras=2.1.2