Setting Up TensorFlow Using Archives


Step 1:
Here are few links from where wheel (with .WHL extension) files can be downloaded:

    1. https://repo.anaconda.com/pkgs/main/win-64/
    2. https://repo.continuum.io/pkgs/free/win-64/
    3. Unofficial Windows Binaries for Python Extension Packages: https://www.lfd.uci.edu/~gohlke/pythonlibs/

Step 2:
For this post, we will be focusing on TensorFlow (v1.12.0):

Following is the list of URLs to those archives required by TensorFlow:

    1. https://files.pythonhosted.org/packages/8a/48/a76be51647d0eb9f10e2a4511bf3ffb8cc1e6b14e9e4fab46173aa79f981/termcolor-1.1.0.tar.gz
    2. https://files.pythonhosted.org/packages/73/fb/00a976f728d0d1fecfe898238ce23f502a721c0ac0ecfedb80e0d88c64e9/six-1.12.0-py2.py3-none-any.whl
    3. https://files.pythonhosted.org/packages/2f/30/2b3caddd3d7e15e919e8b03059ccf0eb9ccef62a9085a9ad92a5823193b7/protobuf-3.6.1-cp35-cp35m-win_amd64.whl
    4. https://files.pythonhosted.org/packages/0c/63/f505d2d4c21db849cf80bad517f0065a30be6b006b0a5637f1b95584a305/absl-py-0.6.1.tar.gz
    5. https://files.pythonhosted.org/packages/a1/1a/d3491298c548870dd9c31d40f0234ff71a1f337d98581c978338d6b83d00/numpy-1.15.4-cp35-none-win_amd64.whl
    6. https://files.pythonhosted.org/packages/5c/78/ff794fcae2ce8aa6323e789d1f8b3b7765f601e7702726f430e814822b96/gast-0.2.0.tar.gz
    7. https://files.pythonhosted.org/packages/62/ed/88e6bfb6ed363725480847b63d571283afb35031bc76d04d40d2c1960e78/h5py-2.9.0-cp35-cp35m-win_amd64.whl
    8. https://files.pythonhosted.org/packages/3f/c4/2ff40221029f7098d58f8d7fb99b97e8100f3293f9856f0fb5834bef100b/Keras_Applications-1.0.6-py2.py3-none-any.whl
    9. https://files.pythonhosted.org/packages/fc/94/74e0fa783d3fc07e41715973435dd051ca89c550881b3454233c39c73e69/Keras_Preprocessing-1.0.5-py2.py3-none-any.whl
    10. https://files.pythonhosted.org/packages/20/c4/12e3e56473e52375aa29c4764e70d1b8f3efa6682bef8d0aae04fe335243/Werkzeug-0.14.1-py2.py3-none-any.whl
    11. https://files.pythonhosted.org/packages/7a/6b/5600647404ba15545ec37d2f7f58844d690baf2f81f3a60b862e48f29287/Markdown-3.0.1-py2.py3-none-any.whl
    12. https://files.pythonhosted.org/packages/ec/19/e3d7c8d6bbb7c3681b1d77dfc985d45d07baea226fc57448e76b986cb154/grpcio-1.17.1-cp35-cp35m-win_amd64.whl
    13. https://files.pythonhosted.org/packages/bd/7e/528c868bb8a0542c8a5686ff3a08502d2691bd50499c6e55f8989fa8e5a0/tensorboard-1.12.1-py3-none-any.whl
    14. https://files.pythonhosted.org/packages/35/6b/11530768cac581a12952a2aad00e1526b89d242d0b9f59534ef6e6a1752f/astor-0.7.1-py2.py3-none-any.whl
    15. https://files.pythonhosted.org/packages/8d/dd/1c16b9be09299e1557fdb28ed51b43f2b536fca289a48878b3fdd752c3c7/tensorflow-1.12.0-cp35-cp35m-win_amd64.whl

Step 3:
Create the .bat file named "TensorFlow_Installation_Script.bat" with following contents:

pip --no-cache-dir install termcolor-1.1.0.tar.gz
pip --no-cache-dir install six-1.12.0-py2.py3-none-any.whl
pip --no-cache-dir install protobuf-3.6.1-cp35-cp35m-win_amd64.whl
pip --no-cache-dir install absl-py-0.6.1.tar.gz
pip --no-cache-dir install numpy-1.15.4-cp35-none-win_amd64.whl
pip --no-cache-dir install gast-0.2.0.tar.gz
pip --no-cache-dir install h5py-2.9.0-cp35-cp35m-win_amd64.whl
pip --no-cache-dir install Keras_Applications-1.0.6-py2.py3-none-any.whl
pip --no-cache-dir install Keras_Preprocessing-1.0.5-py2.py3-none-any.whl
pip --no-cache-dir install Werkzeug-0.14.1-py2.py3-none-any.whl
pip --no-cache-dir install Markdown-3.0.1-py2.py3-none-any.whl
pip --no-cache-dir install grpcio-1.17.1-cp35-cp35m-win_amd64.whl
pip --no-cache-dir install tensorboard-1.12.1-py3-none-any.whl
pip --no-cache-dir install astor-0.7.1-py2.py3-none-any.whl
pip --no-cache-dir install tensorflow-1.12.0-cp35-cp35m-win_amd64.whl

Step 4:
Installation process is:
    1. Keep the above mentioned archives in a directory. Suppose it is "C:\archives", then do "cd C:\archives"
    2. Copy-paste the .bat file here in this directory.
    3. Launch “TensorFlow_Installation_Script.bat” from here in command prompt. Order in which to install these packages is important.

No comments:

Post a Comment