Support more parallel styles in Transformers backend TP

[hmellor]

Based on:

• https://github.com/huggingface/transformers/blob/7b20915f4e36a801873a26351b8b030c4d8d7b3b/src/transformers/integrations/tensor_parallel.py#L905-L926
• https://huggingface.co/docs/transformers/v4.53.3/en/perf_infer_gpu_multi#partitioning-strategies

This PR:

• Renames the tp_plan fallback to replicate since this is a valid name in tp_plans
• Adds explicit support for ReplicatedLinear if the tp_plan contains replicate
• Adds support for colwise_rep and rowwise_rep by passing the correct arguments to ColumnParallelLinear and RowParallelLinear

---

This will marginally improve the performance of the following models (if they worked prior to this PR) because layers marked with colwise_rep and rowwise_rep will no longer fall back to ReplicatedLinear:

Models which use colwise_rep	Models which use rowwise_rep

[gemini-code-assist]

Code Review

This pull request extends support for more tensor parallelism styles from HuggingFace Transformers, which is a valuable addition. The implementation correctly maps the new styles (colwise_rep, rowwise_rep, replicate) to the corresponding vLLM linear layer configurations. I have one suggestion to improve robustness by ensuring that any unsupported style in the tp_plan raises an error instead of silently defaulting, which will help prevent hard-to-debug configuration issues.

[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.

🚀

[DarkLight1337]

Thanks for extending the support!