update proxy docs (#3796)

Author: CUHKSZzxy

docs/en/llm/proxy_server.md

@@ -7,7 +7,7 @@ The request distributor service can parallelize multiple api_server services. Us
 Start the proxy service:
 
 ```shell
-lmdeploy serve proxy --server-name {server_name} --server-port {server_port} --strategy "min_expected_latency"
+lmdeploy serve proxy --server-name {server_name} --server-port {server_port} --routing-strategy "min_expected_latency" --serving-strategy Hybrid
 ```
 
 After startup is successful, the URL of the proxy service will also be printed by the script. Access this URL in your browser to open the Swagger UI.
@@ -88,6 +88,13 @@ response = requests.post(url, headers=headers, data='', params=params)
 print(response.text)
 ```
 
+## Serving Strategy
+
+LMDeploy currently supports two serving strategies:
+
+- Hybrid: Does not distinguish between Prefill and Decoding instances, following the traditional inference deployment mode.
+- DistServe: Separates Prefill and Decoding instances, deploying them on different service nodes to achieve more flexible and efficient resource scheduling and scalability.
+
 ## Dispatch Strategy
 
 The current distribution strategies of the proxy service are as follows:

docs/zh_cn/llm/proxy_server.md

@@ -7,7 +7,7 @@
 启动代理服务：
 
 ```shell
-lmdeploy serve proxy --server-name {server_name} --server-port {server_port} --strategy "min_expected_latency"
+lmdeploy serve proxy --server-name {server_name} --server-port {server_port} --routing-strategy "min_expected_latency" --serving-strategy Hybrid
 ```
 
 启动成功后，代理服务的 URL 也会被脚本打印。浏览器访问这个 URL，可以打开 Swagger UI。
@@ -87,6 +87,13 @@ response = requests.post(url, headers=headers, data='', params=params)
 print(response.text)
 ```
 
+## 服务策略
+
+LMDeploy 当前支持混合部署服务（Hybrid），以及 PD 分离部署服务（DistServe）
+
+- Hybrid: 不区分 Prefill 和 Decoding 实例，即传统的推理部署模式。
+- DistServe: 将 Prefill 和 Decoding 实例分离，部署在不同的服务节点上以实现更灵活高效的资源调度和扩展。
+
 ## 分发策略
 
 代理服务目前的分发策略如下：