fix(vertex_ai.py): map finish reason

litellm/__init__.py

@@ -67,6 +67,7 @@ max_budget: float = 0.0  # set the max budget across all providers
 budget_duration: Optional[
     str
 ] = None  # proxy only - resets budget after fixed duration. You can set duration as seconds ("30s"), minutes ("30m"), hours ("30h"), days ("30d").
+_openai_finish_reasons = ["stop", "length", "function_call", "content_filter", "null"]
 _openai_completion_params = [
     "functions",
     "function_call",

litellm/llms/vertex_ai.py

@@ -4,7 +4,7 @@ from enum import Enum
 import requests
 import time
 from typing import Callable, Optional, Union
-from litellm.utils import ModelResponse, Usage, CustomStreamWrapper
+from litellm.utils import ModelResponse, Usage, CustomStreamWrapper, map_finish_reason
 import litellm, uuid
 import httpx
 
@@ -575,9 +575,9 @@ def completion(
         model_response["model"] = model
         ## CALCULATING USAGE
         if model in litellm.vertex_language_models and response_obj is not None:
-            model_response["choices"][0].finish_reason = response_obj.candidates[
-                0
-            ].finish_reason.name
+            model_response["choices"][0].finish_reason = map_finish_reason(
+                response_obj.candidates[0].finish_reason.name
+            )
             usage = Usage(
                 prompt_tokens=response_obj.usage_metadata.prompt_token_count,
                 completion_tokens=response_obj.usage_metadata.candidates_token_count,
@@ -771,9 +771,9 @@ async def async_completion(
         model_response["model"] = model
         ## CALCULATING USAGE
         if model in litellm.vertex_language_models and response_obj is not None:
-            model_response["choices"][0].finish_reason = response_obj.candidates[
-                0
-            ].finish_reason.name
+            model_response["choices"][0].finish_reason = map_finish_reason(
+                response_obj.candidates[0].finish_reason.name
+            )
             usage = Usage(
                 prompt_tokens=response_obj.usage_metadata.prompt_token_count,
                 completion_tokens=response_obj.usage_metadata.candidates_token_count,

litellm/tests/test_amazing_vertex_completion.py

@@ -123,6 +123,10 @@ def test_vertex_ai():
             print(response)
             assert type(response.choices[0].message.content) == str
             assert len(response.choices[0].message.content) > 1
+            print(
+                f"response.choices[0].finish_reason: {response.choices[0].finish_reason}"
+            )
+            assert response.choices[0].finish_reason in litellm._openai_finish_reasons
         except Exception as e:
             pytest.fail(f"Error occurred: {e}")