How can I find the requirements for older TensorFlow versions?

Question

Because right now, I am using the Wayback Machine, which is absurd. There must be a better way.

For a given TensorFlow version, e.g. 2.1, how / where can I find the related software requirements?

More specifically, which NVIDIA GPU Drivers, Cuda Toolkit and cuDNN are required.

Answer 1

One way is, you can setup a virtual environment and install tensorflow 2.1

$ pip install tensorflow==2.1.3

Then just calling the library in command line will show its dependencies

$ tensorflow

Check this answer for details.

---

If you don't want to install the library,

1. You can go to the TensorFlow releases.
2. Select the specific package
3. And in the source code for the older version (will have to download its zip) in setup.py under REQUIRED_PACKAGES you can find the list.

Example here (check the path shown below to find the file)- https://github.com/tensorflow/tensorflow/blob/master/tensorflow/tools/pip_package/setup.py

REQUIRED_PACKAGES = [
    'absl-py ~= 0.10',
    'astunparse ~= 1.6.3',
    'flatbuffers ~= 1.12.0',
    'google_pasta ~= 0.2',
    'h5py ~= 3.1.0',
    'keras_preprocessing ~= 1.1.2',
    'numpy ~= 1.19.2',
    'opt_einsum ~= 3.3.0',
    'protobuf >= 3.9.2',
    'six ~= 1.15.0',
    'termcolor ~= 1.1.0',
    'typing_extensions ~= 3.7.4',
    'wheel ~= 0.35',
    'wrapt ~= 1.12.1',
    # These packages need to be pinned exactly as newer versions are
    # incompatible with the rest of the ecosystem
    'gast == 0.4.0',
    # TensorFlow ecosystem packages that TF exposes API for
    # These need to be in sync with the existing TF version
    # They are updated during the release process
    # When updating these, please also update the nightly versions below
    'tensorboard ~= 2.4',
    'tensorflow_estimator ~= 2.3.0',
]