Fix vllm api_server v1/models error (#12867)

intel · Feb 21, 2025 · 3ea5389 · 3ea5389
1 parent 8077850
commit 3ea5389
Showing 1 changed file with 7 additions and 3 deletions.
diff --git a/python/llm/src/ipex_llm/vllm/xpu/entrypoints/openai/api_server.py b/python/llm/src/ipex_llm/vllm/xpu/entrypoints/openai/api_server.py
@@ -278,6 +278,10 @@ def base(request: Request) -> OpenAIServing:
     return tokenization(request)
 
 
+def models(request: Request) -> OpenAIServingModels:
+    return request.app.state.openai_serving_models
+
+
 def chat(request: Request) -> Optional[OpenAIServingChat]:
     return request.app.state.openai_serving_chat
 
@@ -345,10 +349,10 @@ async def detokenize(request: DetokenizeRequest, raw_request: Request):
 
 @router.get("/v1/models")
 async def show_available_models(raw_request: Request):
-    handler = base(raw_request)
+    handler = models(raw_request)
 
-    models = await handler.show_available_models()
-    return JSONResponse(content=models.model_dump())
+    models_ = await handler.show_available_models()
+    return JSONResponse(content=models_.model_dump())
 
 
 @router.get("/version")