[conda] Older Torchvision versions (<0.6.1) no longer available on PyTorch's conda channel

Not sure if this issue is a duplicate but I tried searching for issues on both the PyTorch and Torchvision repos and couldn't find anything. I'm guessing this is intentional and was done for reasons not apparent to me, but I'm creating this issue in the hope that this was somehow an oversight. It appears that the pytorch channel on conda has deleted pytorch packages that are older than v1.5.1 and torchvision packages that are older than v0.6.1. I'm hoping these older releases can be restored as I personally have several projects (for example, logicnets) based on these older versions, which I know can reproduce results from several papers.

Unfortunately, this means that following basic instructions for "getting started" for older PyTorch versions on the PyTorch website fail.

For example:

conda create -n pt1.4 pytorch==1.4.0 torchvision==0.5.0 cpuonly -c pytorch
Collecting package metadata (current_repodata.json): done
Solving environment: failed with repodata from current_repodata.json, will retry with next repodata source.
Collecting package metadata (repodata.json): done
Solving environment: failed

PackagesNotFoundError: The following packages are not available from current channels:

  - torchvision==0.5.0

Current channels:

  - https://conda.anaconda.org/pytorch/linux-64
  - https://conda.anaconda.org/pytorch/noarch
  - https://repo.anaconda.com/pkgs/main/linux-64
  - https://repo.anaconda.com/pkgs/main/noarch
  - https://repo.anaconda.com/pkgs/r/linux-64
  - https://repo.anaconda.com/pkgs/r/noarch

To search for alternate channels that may provide the conda package you're
looking for, navigate to

    https://anaconda.org

and use the search bar at the top of the page.

I haven't tried installing these older versions from pip, but I suspect that they would work and that it would workaround this issue.

PS: Apologies for creating this issue in both repos, but it appears to affect both projects.

Start by reproducing the conda command shown in the issue and checking the linked PyTorch channel package listings for pytorch 1.4.0 and torchvision 0.5.0. Review the linked previous-versions installation instructions and determine whether the packages can be restored; done means the documented command resolves successfully or the affected instructions are updated.