Merge branch 'wangshangsam/fix-req-timeout' of github.com:CentML/mlperf-inference into wangshangsam/fix-req-timeout

Author: wangshangsam

multimodal/vl2l/src/mlperf_inference_multimodal_vl2l/task.py

@@ -67,7 +67,8 @@ def __init__(
         self.openai_api_client = AsyncOpenAI(
             base_url=endpoint.url,
             http_client=DefaultAioHttpClient(
-                timeout=httpx.Timeout(timeout=request_timeout_seconds, connect=5.0),
+                timeout=httpx.Timeout(
+                    timeout=request_timeout_seconds, connect=5.0),
             ),
             api_key=endpoint.api_key,
             timeout=request_timeout_seconds,
@@ -187,7 +188,9 @@ def estimated_num_performance_samples(self) -> int:
         """
         estimation_indices = random.sample(
             range(self.total_num_samples),
-            k=min(MAX_NUM_ESTIMATION_PERFORMANCE_SAMPLES, self.total_num_samples),
+            k=min(
+                MAX_NUM_ESTIMATION_PERFORMANCE_SAMPLES,
+                self.total_num_samples),
         )
         estimation_samples = [
             self.formulate_loaded_sample(
@@ -274,7 +277,8 @@ def _unload_samples_from_ram(query_sample_indices: list[int]) -> None:
             _unload_samples_from_ram,
         )
 
-    async def _query_endpoint_async_batch(self, query_sample: lg.QuerySample) -> None:
+    async def _query_endpoint_async_batch(
+            self, query_sample: lg.QuerySample) -> None:
         """Query the endpoint through the async OpenAI API client."""
         try:
             sample = self.loaded_samples[query_sample.index]
@@ -351,7 +355,8 @@ async def _query_endpoint_async_batch(self, query_sample: lg.QuerySample) -> Non
                 ],
             )
 
-    async def _query_endpoint_async_stream(self, query_sample: lg.QuerySample) -> None:
+    async def _query_endpoint_async_stream(
+            self, query_sample: lg.QuerySample) -> None:
         """Query the endpoint through the async OpenAI API client."""
         ttft_set = False
         try: