[Docs]fix sampling docs 2.1

docs/features/sampling.md

@@ -98,7 +98,7 @@ curl -X POST "http://0.0.0.0:9222/v1/chat/completions" \
     {"role": "user", "content": "How old are you"}
   ],
   "top_p": 0.8,
-  "top_k": 50
+  "top_k": 20
 }'
 ```
 
@@ -117,7 +117,7 @@ response = client.chat.completions.create(
     ],
     stream=True,
     top_p=0.8,
-    top_k=50
+    extra_body={"top_k": 20, "min_p":0.1}
 )
 for chunk in response:
     if chunk.choices[0].delta:
@@ -159,8 +159,7 @@ response = client.chat.completions.create(
     ],
     stream=True,
     top_p=0.8,
-    top_k=20,
-    min_p=0.1
+    extra_body={"top_k": 20, "min_p":0.1}
 )
 for chunk in response:
     if chunk.choices[0].delta:

docs/offline_inference.md

@@ -183,6 +183,7 @@ For ```LLM``` configuration, refer to [Parameter Documentation](parameters.md).
 * min_p(float): Minimum probability relative to the maximum probability for a token to be considered (>0 filters low-probability tokens to improve quality)
 * max_tokens(int): Maximum generated tokens (input + output)
 * min_tokens(int): Minimum forced generation length
+* bad_words(list[str]): Prohibited words
 
 ### 2.5 fastdeploy.engine.request.RequestOutput
 

docs/zh/features/sampling.md

@@ -98,7 +98,7 @@ curl -X POST "http://0.0.0.0:9222/v1/chat/completions" \
     {"role": "user", "content": "How old are you"}
   ],
   "top_p": 0.8,
-  "top_k": 50
+  "top_k": 20
 }'
 ```
 
@@ -118,7 +118,7 @@ response = client.chat.completions.create(
     ],
     stream=True,
     top_p=0.8,
-    extra_body={"top_k": 50}
+    extra_body={"top_k": 20}
 )
 for chunk in response:
     if chunk.choices[0].delta:
@@ -161,8 +161,7 @@ response = client.chat.completions.create(
     ],
     stream=True,
     top_p=0.8,
-    extra_body={"top_k": 20},
-    min_p=0.1
+    extra_body={"top_k": 20, "min_p": 0.1}
 )
 for chunk in response:
     if chunk.choices[0].delta:

docs/zh/offline_inference.md

@@ -183,6 +183,7 @@ for output in outputs:
 * min_p(float): token入选的最小概率阈值(相对于最高概率token的比值，设为>0可通过过滤低概率token来提升文本生成质量)
 * max_tokens(int): 限制模型生成的最大token数量（包括输入和输出）
 * min_tokens(int): 强制模型生成的最少token数量，避免过早结束
+* bad_words(list[str]): 禁止生成的词列表, 防止模型生成不希望出现的词
 
 ### 2.5 fastdeploy.engine.request.RequestOutput