Reinstate best_of for V0 (#14356)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>

vllm/entrypoints/openai/protocol.py

@@ -242,6 +242,7 @@ class ChatCompletionRequest(OpenAIBaseModel):
     user: Optional[str] = None
 
     # doc: begin-chat-completion-sampling-params
+    best_of: Optional[int] = None
     use_beam_search: bool = False
     top_k: Optional[int] = None
     min_p: Optional[float] = None
@@ -478,6 +479,7 @@ class ChatCompletionRequest(OpenAIBaseModel):
 
         return SamplingParams.from_optional(
             n=self.n,
+            best_of=self.best_of,
             presence_penalty=self.presence_penalty,
             frequency_penalty=self.frequency_penalty,
             repetition_penalty=repetition_penalty,
@@ -648,6 +650,7 @@ class CompletionRequest(OpenAIBaseModel):
     # https://platform.openai.com/docs/api-reference/completions/create
     model: Optional[str] = None
     prompt: Union[list[int], list[list[int]], str, list[str]]
+    best_of: Optional[int] = None
     echo: Optional[bool] = False
     frequency_penalty: Optional[float] = 0.0
     logit_bias: Optional[dict[str, float]] = None
@@ -845,6 +848,7 @@ class CompletionRequest(OpenAIBaseModel):
 
         return SamplingParams.from_optional(
             n=self.n,
+            best_of=self.best_of,
             presence_penalty=self.presence_penalty,
             frequency_penalty=self.frequency_penalty,
             repetition_penalty=repetition_penalty,