fix: prefer to use max_completion_tokens rather than max_tokens (#17516)

2026-01-21 20:22:20 -08:00
parent e95668abc7
commit a5bbcda968
1 changed files with 1 additions and 1 deletions
@@ -654,7 +654,7 @@ class ChatCompletionRequest(BaseModel):
        sampling_params = {
            "temperature": get_param("temperature"),
-            "max_new_tokens": self.max_tokens or self.max_completion_tokens,
+            "max_new_tokens": self.max_completion_tokens or self.max_tokens,
            "min_new_tokens": self.min_tokens,
            "stop": stop,
            "stop_token_ids": self.stop_token_ids,