Revert "[Doc] RayServe Single-Host TPU v6e Example with vLLM (ray-pro…

…ject#47814)"

This reverts commit e4a448f.

doc/source/cluster/kubernetes/examples.md

@@ -18,7 +18,6 @@ examples/rayjob-kueue-gang-scheduling
 examples/distributed-checkpointing-with-gcsfuse
 examples/modin-example
 examples/vllm-rayservice
-examples/tpu-single-host-rayservice
 ```
 
 
@@ -37,4 +36,3 @@ This section presents example Ray workloads to try out on your Kubernetes cluste
 - {ref}`kuberay-distributed-checkpointing-gcsefuse`
 - {ref}`kuberay-modin-example`
 - {ref}`kuberay-vllm-rayservice-example`
-- {ref}`kuberay-tpu-single-host-vllm`

doc/source/cluster/kubernetes/examples/vllm-rayservice.md

@@ -4,8 +4,6 @@
 
 This guide demonstrates how to [Serve a Large Language Model with vLLM](https://docs.ray.io/en/latest/serve/tutorials/vllm-example.html) on Kubernetes using KubeRay. The example in this guide deploys the `meta-llama/Meta-Llama-3-8B-Instruct` model from Hugging Face on Google Kubernetes Engine (GKE).
 
-* This example uses vLLM with KubeRay. However, Ray provides a native [Serve LLM API](https://docs.ray.io/en/latest/serve/llm/overview.html) to enable users to deploy multiple LLM models together with the Ray Serve API. This API is the recommended path for running LLMs through Ray Serve.
-
 ## Prerequisites
 
 This example downloads model weights from Hugging Face. You need to complete the following