From eaffe4486cb1d7edf884e6e254cab33fc397e308 Mon Sep 17 00:00:00 2001
From: Kay Yan <kay.yan@daocloud.io>
Date: Thu, 18 Sep 2025 19:36:47 +0800
Subject: [PATCH] [Docs] Fix pooling-params doc references in
 openai_compatible_server.md (#24939)

---
 docs/api/README.md                       |  1 -
 docs/serving/openai_compatible_server.md | 20 ++++++++++++--------
 vllm/pooling_params.py                   | 20 ++++++++++++++------
 3 files changed, 26 insertions(+), 15 deletions(-)

diff --git a/docs/api/README.md b/docs/api/README.md
index 148211756480c..86e310f567dd3 100644
--- a/docs/api/README.md
+++ b/docs/api/README.md
@@ -46,7 +46,6 @@ Engine classes for offline and online inference.
 Inference parameters for vLLM APIs.
 
 [](){ #sampling-params }
-[](){ #pooling-params }
 
 - [vllm.SamplingParams][]
 - [vllm.PoolingParams][]
diff --git a/docs/serving/openai_compatible_server.md b/docs/serving/openai_compatible_server.md
index bc52d02a50bd2..bac3f6c1fe90c 100644
--- a/docs/serving/openai_compatible_server.md
+++ b/docs/serving/openai_compatible_server.md
@@ -317,10 +317,11 @@ Full example: <gh-file:examples/online_serving/pooling/openai_chat_embedding_cli
 
 #### Extra parameters
 
-The following [pooling parameters][pooling-params] are supported.
+The following [pooling parameters][vllm.PoolingParams] are supported.
 
 ```python
---8<-- "vllm/entrypoints/openai/protocol.py:embedding-pooling-params"
+--8<-- "vllm/pooling_params.py:common-pooling-params"
+--8<-- "vllm/pooling_params.py:embedding-pooling-params"
 ```
 
 The following extra parameters are supported by default:
@@ -527,10 +528,11 @@ curl -v "http://127.0.0.1:8000/classify" \
 
 #### Extra parameters
 
-The following [pooling parameters][pooling-params] are supported.
+The following [pooling parameters][vllm.PoolingParams] are supported.
 
 ```python
---8<-- "vllm/entrypoints/openai/protocol.py:classification-pooling-params"
+--8<-- "vllm/pooling_params.py:common-pooling-params"
+--8<-- "vllm/pooling_params.py:classification-pooling-params"
 ```
 
 The following extra parameters are supported:
@@ -733,10 +735,11 @@ Full example: <gh-file:examples/online_serving/openai_cross_encoder_score_for_mu
 
 #### Extra parameters
 
-The following [pooling parameters][pooling-params] are supported.
+The following [pooling parameters][vllm.PoolingParams] are supported.
 
 ```python
---8<-- "vllm/entrypoints/openai/protocol.py:score-pooling-params"
+--8<-- "vllm/pooling_params.py:common-pooling-params"
+--8<-- "vllm/pooling_params.py:classification-pooling-params"
 ```
 
 The following extra parameters are supported:
@@ -815,10 +818,11 @@ Result documents will be sorted by relevance, and the `index` property can be us
 
 #### Extra parameters
 
-The following [pooling parameters][pooling-params] are supported.
+The following [pooling parameters][vllm.PoolingParams] are supported.
 
 ```python
---8<-- "vllm/entrypoints/openai/protocol.py:rerank-pooling-params"
+--8<-- "vllm/pooling_params.py:common-pooling-params"
+--8<-- "vllm/pooling_params.py:classification-pooling-params"
 ```
 
 The following extra parameters are supported:
diff --git a/vllm/pooling_params.py b/vllm/pooling_params.py
index 6672392b8d080..a6313367457a4 100644
--- a/vllm/pooling_params.py
+++ b/vllm/pooling_params.py
@@ -20,25 +20,33 @@ class PoolingParams(
     """API parameters for pooling models.
 
     Attributes:
+        truncate_prompt_tokens: Controls prompt truncation.
+            Set to -1 to use the model's default truncation size.
+            Set to k to keep only the last k tokens (left truncation).
+            Set to None to disable truncation.         
         normalize: Whether to normalize the embeddings outputs.
         dimensions: Reduce the dimensions of embeddings
-                    if model support matryoshka representation.
+            if model support matryoshka representation.
         activation: Whether to apply activation function to
-                    the classification outputs.
+            the classification outputs.
         softmax: Whether to apply softmax to the reward outputs.
     """
+
+    # --8<-- [start:common-pooling-params]
     truncate_prompt_tokens: Optional[Annotated[int,
                                                msgspec.Meta(ge=-1)]] = None
-    """If set to -1, will use the truncation size supported by the model. If
-    set to an integer k, will use only the last k tokens from the prompt
-    (i.e., left truncation). If set to `None`, truncation is disabled."""
+    # --8<-- [end:common-pooling-params]
 
     ## for embeddings models
+    # --8<-- [start:embedding-pooling-params]
     dimensions: Optional[int] = None
     normalize: Optional[bool] = None
+    # --8<-- [end:embedding-pooling-params]
 
-    ## for classification models
+    ## for classification, scoring and rerank
+    # --8<-- [start:classification-pooling-params]
     activation: Optional[bool] = None
+    # --8<-- [end:classification-pooling-params]
 
     ## for reward models
     softmax: Optional[bool] = None