fix(logging_utils.py): revert change

litellm/litellm_core_utils/logging_utils.py

@@ -77,7 +77,8 @@ def _assemble_complete_response_from_streaming_chunks(
     complete_streaming_response: Optional[
         Union[ModelResponse, TextCompletionResponse]
     ] = None
-    if getattr(result, "usage", None) is not None:  # if it's the last chunk
+
+    if result.choices[0].finish_reason is not None:  # if it's the last chunk
         streaming_chunks.append(result)
         try:
             complete_streaming_response = litellm.stream_chunk_builder(

litellm/litellm_core_utils/streaming_handler.py

@@ -1555,10 +1555,11 @@ class CustomStreamWrapper:
                     if response is None:
                         continue
                     ## LOGGING
-                    threading.Thread(
-                        target=self.run_success_logging_and_cache_storage,
-                        args=(response, cache_hit),
-                    ).start()  # log response
+                    executor.submit(
+                        self.run_success_logging_and_cache_storage,
+                        response,
+                        cache_hit,
+                    )  # log response
                     choice = response.choices[0]
                     if isinstance(choice, StreamingChoices):
                         self.response_uptil_now += choice.delta.get("content", "") or ""
@@ -1628,10 +1629,11 @@ class CustomStreamWrapper:
                     usage = calculate_total_usage(chunks=self.chunks)
                     processed_chunk._hidden_params["usage"] = usage
                 ## LOGGING
-                threading.Thread(
-                    target=self.run_success_logging_and_cache_storage,
-                    args=(processed_chunk, cache_hit),
-                ).start()  # log response
+                executor.submit(
+                    self.run_success_logging_and_cache_storage,
+                    processed_chunk,
+                    cache_hit,
+                )  # log response
                 return processed_chunk
         except Exception as e:
             traceback_exception = traceback.format_exc()

tests/logging_callback_tests/test_assemble_streaming_responses.py

@@ -26,7 +26,7 @@ from respx import MockRouter
 import litellm
 from litellm import Choices, Message, ModelResponse, TextCompletionResponse, TextChoices
 
-from litellm.litellm_core_utils.litellm_logging import (
+from litellm.litellm_core_utils.logging_utils import (
     _assemble_complete_response_from_streaming_chunks,
 )