PaliGemma 2 support

[janimo]

Minor changes to support PaliGemma 2.

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.
Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can do one of these:

• Add ready label to the PR
• Enable auto-merge.

🚀

[Isotr0py]

Can you update the docs/source/models/supported_models.rst to state the support of paligemma2 explicitly as well?

[janimo]

thanks for the review, I have force-pushed a new commit including the requested changes.

[Isotr0py]

Thanks for fixing this!

[exceedzhang]

The large language model 'google/paligemma2-10b-pt-896' that you are currently using has experienced inference failure.

[janimo]

@exceedzhang I have not tested that particular model size, nor the serving interface, just the 3b models via the direct llm.Engine() API.
Do the smaller PaliGemma2 models and PaliGemma 1 models work for you in the same setup?

[exceedzhang]

The same mistake！

[janimo]

@exceedzhang If it happens with paligemma 1 models too, please file an issue for it with the exact command to reproduce it.