[Frontend] Support min_p in the Responses API (#48084)

Signed-off-by: sungbin1015 <sbin@solbox.com>
Co-authored-by: Flora Feng <4florafeng@gmail.com>
This commit is contained in:
sungbin1015
2026-10-08 20:38:17 +00:00
committed by GitHub
co-authored by Flora Feng
parent c5c1161382
commit 2ddde19e8e
5 changed files with 27 additions and 8 deletions
@@ -259,6 +259,7 @@ class ResponsesRequest(OpenAIBaseModel):
top_logprobs: TopLogprobsParam = 0
top_p: float | None = None
top_k: int | None = None
min_p: float | None = None
truncation: Literal["auto", "disabled"] | None = "disabled"
user: str | None = None
skip_special_tokens: bool = True
@@ -444,6 +445,7 @@ class ResponsesRequest(OpenAIBaseModel):
"temperature": 1.0,
"top_p": 1.0,
"top_k": 0,
"min_p": 0.0,
}
def extract_structured_outputs(self) -> StructuredOutputsParams | None:
@@ -494,6 +496,10 @@ class ResponsesRequest(OpenAIBaseModel):
top_k = default_sampling_params.get(
"top_k", self._DEFAULT_SAMPLING_PARAMS["top_k"]
)
if (min_p := self.min_p) is None:
min_p = default_sampling_params.get(
"min_p", self._DEFAULT_SAMPLING_PARAMS["min_p"]
)
if (repetition_penalty := self.repetition_penalty) is None:
repetition_penalty = default_sampling_params.get("repetition_penalty", 1.0)
@@ -519,6 +525,7 @@ class ResponsesRequest(OpenAIBaseModel):
watermarking=self.watermarking,
top_p=top_p,
top_k=top_k,
min_p=min_p,
max_tokens=max_tokens,
logprobs=self.top_logprobs if self.is_include_output_logprobs() else None,
stop=stop,