update

app/core/runner/llm_backend.py

@@ -29,7 +29,7 @@ class LLMBackend:
         temperature=None,
         top_p=None,
         response_format=None,
-        parallel_tool_calls=False,
+        parallel_tool_calls=True,
         audio=None,
         modalities=None,
     ) -> ChatCompletion | Stream[ChatCompletionChunk]:
@@ -40,7 +40,8 @@ class LLMBackend:
             "messages": messages,
             "model": model,
             "stream": stream,
-            "presence_penalty": None,
+            "presence_penalty": 0,
+            "frequency_penalty": 0
             # "parallel_tool_calls": parallel_tool_calls,
         }
         if extra_body:
@@ -84,6 +85,7 @@ class LLMBackend:
             if "content" not in message:
                 message["content"] = ""
         chat_params["timeout"] = 300
+
         logging.info("chat_params: %s", chat_params)
         response = self.client.chat.completions.create(**chat_params)
         logging.info("chat_response: %s", response)

app/core/runner/thread_runner.py

@@ -195,7 +195,7 @@ class ThreadRunner:
             stream=self.stream,
             stream_options=run.stream_options,
             extra_body=run.extra_body,
-            temperature=run.temperature,
+            temperature=run.temperature or 1.0,
             top_p=run.top_p,
             response_format=run.response_format,
             parallel_tool_calls=run.parallel_tool_calls,