From e074e76b31d4fff13e87a455dbc3acdaa92c537a Mon Sep 17 00:00:00 2001
From: Shi Shuai <126407087+shuaills@users.noreply.github.com>
Date: Fri, 21 Feb 2025 19:25:52 +0000
Subject: [PATCH] docs: Add offline engine launch example and documentation
 (#3771)

---
 docs/backend/offline_engine_api.ipynb    |  3 ++-
 examples/runtime/engine/launch_engine.py | 17 +++++++++++++++++
 2 files changed, 19 insertions(+), 1 deletion(-)
 create mode 100644 examples/runtime/engine/launch_engine.py

diff --git a/docs/backend/offline_engine_api.ipynb b/docs/backend/offline_engine_api.ipynb
index aa2159f3c52..64a0fe3f99d 100644
--- a/docs/backend/offline_engine_api.ipynb
+++ b/docs/backend/offline_engine_api.ipynb
@@ -18,6 +18,8 @@
     "- Non-streaming asynchronous generation\n",
     "- Streaming asynchronous generation\n",
     "\n",
+    "****To launch the offline engine, `__main__` condition is necessary in your own script because we use \"spawn\" to create subprocesses, for more details please refer to [launch_engine](https://github.com/sgl-project/sglang/blob/main/examples/runtime/engine/launch_engine.py).****\n",
+    "\n",
     "Additionally, you can easily build a custom server on top of the SGLang offline engine. A detailed example working in a python script can be found in [custom_server](https://github.com/sgl-project/sglang/blob/main/examples/runtime/engine/custom_server.py)."
    ]
   },
@@ -45,7 +47,6 @@
     "if is_in_ci():\n",
     "    import patch\n",
     "\n",
-    "\n",
     "llm = sgl.Engine(model_path=\"meta-llama/Meta-Llama-3.1-8B-Instruct\")"
    ]
   },
diff --git a/examples/runtime/engine/launch_engine.py b/examples/runtime/engine/launch_engine.py
new file mode 100644
index 00000000000..b97e0bcd579
--- /dev/null
+++ b/examples/runtime/engine/launch_engine.py
@@ -0,0 +1,17 @@
+"""
+This example demonstrates how to launch the offline engine.
+"""
+
+import sglang as sgl
+
+
+def main():
+    llm = sgl.Engine(model_path="meta-llama/Meta-Llama-3.1-8B-Instruct")
+    llm.generate("What is the capital of France?")
+    llm.shutdown()
+
+
+# The __main__ condition is necessary here because we use "spawn" to create subprocesses
+# Spawn starts a fresh program every time, if there is no __main__, it will run into infinite loop to keep spawning processes from sgl.Engine
+if __name__ == "__main__":
+    main()