[V1][Core] Porting Connector API to V1

[YuhanLiu11]

This PR ports the connector API from V0: It introduces minimal changes to gpu_model_runner.py to send and receive KV caches through the connector.

To see an example of how this connector API works in V1, run the disaggregated prefill script: examples/online_serving/disaggregated_prefill_v1.sh.

Reference implementation: #8498

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

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 either: Add ready label to the PR or enable auto-merge.

🚀

[WoosukKwon]

Hi @YuhanLiu11, thanks for submitting the PR. Unfortunately, we're currently focused on a few other higher-priority items in V1, so we need to hold off on any other changes until those are resolved. This may take 2-4 weeks. I will get back to this PR after that. Thanks for understanding.

[YuhanLiu11]

Hi @YuhanLiu11, thanks for submitting the PR. Unfortunately, we're currently focused on a few other higher-priority items in V1, so we need to hold off on any other changes until those are resolved. This may take 2-4 weeks. I will get back to this PR after that. Thanks for understanding.

Got it! Thanks for the information.

[mergify]

This pull request has merge conflicts that must be resolved before it can be
merged. Please rebase the PR, @YuhanLiu11.

https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork

[orozery]

@YuhanLiu11 Just wondering:
Why is the connector API changed to accept input_ids instead of model_input like in v0?
It seems to me that model_input is more flexible as it may allow for partial kv recv.

[hasB4K]

Why is the connector API changed to accept input_ids instead of model_input like in v0?

I agree with @orozery, giving model_inputs was better. One can for instance use the request_id by using list(model_inputs.request_ids_to_seq_ids.keys()) (it works since the dict keeps the order since Python 3.7).

Providing only input_tokens seems like a downgrade for custom connectors.

[hasB4K]

@YuhanLiu11 I think you can close this PR. #15960 is the new one.

[github-actions]

This pull request has been automatically marked as stale because it has not had any activity within 90 days. It will be automatically closed if no further activity occurs within 30 days. Leave a comment if you feel this pull request should remain open. Thank you!

[github-actions]

This pull request has been automatically closed due to inactivity. Please feel free to reopen if you intend to continue working on it. Thank you!