[Docs] Add example for Renderer.render_cmpl() usage (#54265)
Signed-off-by: null-Exception1 <rising1legend@gmail.com>
Shaurya Pratap Srivastava committed
Sep 4, 2026 at 12:05 UTC
e862c2f45bc81c72bb84f3807d88d60cfb93d694
1 file changed
+11
vllm/v1/engine/async_llm.py
+11
@@ -652,6 +652,17 @@ class AsyncLLM(EngineClient):
652
653
The caller of generate() iterates the returned AsyncGenerator,
654
returning the RequestOutput back to the caller.
655
+
656
+ Note:
657
+ Passing a raw prompt string directly to this method is deprecated.
658
+ Advanced power-users can manually bypass the raw-prompt fallback
659
+ path using the Engine's underlying Renderer pipeline:
660
+
661
+ >>> from vllm.inputs import parse_model_prompt
662
+ >>> parsed = parse_model_prompt(self.model_config, "Prompt text")
663
+ >>> params = self.renderer.default_cmpl_tok_params
664
+ >>> (engine_input,) = self.renderer.render_cmpl([parsed], params)
665
+ >>> gen = self.generate(engine_input, sampling_params, request_id)
666
"""
667
668
q: RequestOutputCollector | None = None