fix: increase agent execute and chat completions timeouts to ≥600s

vlmrun/client/agent.py

@@ -320,7 +320,7 @@ def completions(self):
         openai_client = OpenAI(
             api_key=self._client.api_key,
             base_url=base_url,
-            timeout=self._client.timeout,
+            timeout=self._client.timeout if self._client.timeout is None else max(self._client.timeout, 600),
             max_retries=self._client.max_retries,
         )
 
@@ -374,7 +374,7 @@ async def main():
         async_openai_client = AsyncOpenAI(
             api_key=self._client.api_key,
             base_url=base_url,
-            timeout=self._client.timeout,
+            timeout=self._client.timeout if self._client.timeout is None else max(self._client.timeout, 600),
             max_retries=self._client.max_retries,
         )
 

vlmrun/client/executions.py

@@ -57,7 +57,7 @@ def get(self, id: str) -> AgentExecutionResponse:
         return AgentExecutionResponse(**response)
 
     def wait(
-        self, id: str, timeout: int = 300, sleep: int = 5
+        self, id: str, timeout: int = 600, sleep: int = 5
     ) -> AgentExecutionResponse:
         """Wait for execution to complete.
 

vlmrun/client/predictions.py

@@ -129,7 +129,7 @@ def get(self, id: str) -> PredictionResponse:
 
         return prediction
 
-    def wait(self, id: str, timeout: int = 300, sleep: int = 5) -> PredictionResponse:
+    def wait(self, id: str, timeout: int = 600, sleep: int = 5) -> PredictionResponse:
         """Wait for prediction to complete.
 
         Args: