Make return-exceptions as default behavior in litellm.batch_completion

litellm/main.py

@@ -2332,11 +2332,6 @@ def batch_completion(
     """
     args = locals()
 
-    # extra kw for dealing with exceptions
-    return_exceptions = args.get("kwargs").get("return_exceptions", False)
-    if "return_exceptions" in args.get("kwargs"):
-        args.get("kwargs").pop("return_exceptions")
-
     batch_messages = messages
     completions = []
     model = model
@@ -2391,15 +2386,14 @@ def batch_completion(
 
         # Retrieve the results from the futures
         # results = [future.result() for future in completions]
-        if return_exceptions:
-            results = []
-            for future in completions:
-                try:
-                    results.append(future.result())
-                except Exception as exc:
-                    results.append(exc)
-        else:  # original
-           results = [future.result() for future in completions]
+        # return exceptions if any
+        results = []
+        for future in completions:
+            try:
+                results.append(future.result())
+            except Exception as exc:
+                results.append(exc)
+
     return results
 
 

litellm/tests/test_batch_completion_return_exceptions.py

@@ -1,30 +1,17 @@
 """https://github.com/BerriAI/litellm/pull/3397/commits/a7ec1772b1457594d3af48cdcb0a382279b841c7#diff-44852387ceb00aade916d6b314dfd5d180499e54f35209ae9c07179febe08b4b."""
 """Test batch_completion's return_exceptions."""
-import pytest
 import litellm
 
 msg1 = [{"role": "user", "content": "hi 1"}]
 msg2 = [{"role": "user", "content": "hi 2"}]
 
 
-def test_batch_completion_return_exceptions_default():
-    """Test batch_completion's return_exceptions."""
-    with pytest.raises(Exception):
-        _ = litellm.batch_completion(
-            model="gpt-3.5-turbo",
-            messages=[msg1, msg2],
-            api_key="sk_xxx",  # deliberately set invalid key
-            # return_exceptions=False,
-        )
-
-
 def test_batch_completion_return_exceptions_true():
     """Test batch_completion's return_exceptions."""
     res = litellm.batch_completion(
         model="gpt-3.5-turbo",
         messages=[msg1, msg2],
         api_key="sk_xxx",  # deliberately set invalid key
-        return_exceptions=True,
     )
 
     assert isinstance(res[0], litellm.exceptions.AuthenticationError)