feat: allow to use hf-models

[cgebbe]

What does this PR do?

Problem: This doesn't currently work, but should in my opinion?

        import transformers
        model = transformers.AutoBackbone.from_pretrained(
            "hf-hub:bioptimus/H-optimus-0",
            use_timm_backbone=True,
            use_pretrained_backbone=True,
        )

Solution approach:

Just remove the two lines which explicitly check whether the string is in timm.list_models(). Alternatively, we could introduce more e.g. backbone_kwargs arguments.

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline,
  Pull Request section?
• Was this discussed/approved via a Github issue or the forum? Please add a link
  to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.
• Did you write any new necessary tests?

Who can review?

Models:

• vision models: @amyeroberts, @qubvel

[LysandreJik]

In addition to pinging @qubvel, pinging @rwightman for knowledge

[rwightman]

@LysandreJik @cgebbe I assume author of PR has verified that this fix allows the desired use case to work so I'm not looking at that. A quick look I believe it should allow the use case for hf-hub timm models which definitely should be supported.

I also believe the removal of that check is appropriate in any case. If the requested model isn't in timm it will error anyways. e.g. timm.create_model('crumpet') --> RuntimeError: Unknown model (crumpet)

If ValueError is preferred for uniform exception handling I can change RuntimeError to slightly more specific ValueError.

[qubvel]

Hi @Debbie, thanks for the contribution! Agree with the messages above, it makes sense to remove this check and enable loading timm-compatible weight from the hub 👍