Fix return finish_reason = "tool_calls" for gemini tool calling (#10485)

* fix(vertex_and_google_ai_studio.py): fix finish reason to be 'tool_calls' when tool call returned

Vertex returns 'Stop', openai format is 'tool calls'

* test(base_llm_unit_tests.py): bump test to assert tool calls in finish reason

litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py

@@ -44,6 +44,7 @@ from litellm.types.llms.openai import (
     ChatCompletionToolCallFunctionChunk,
     ChatCompletionToolParamFunctionChunk,
     ChatCompletionUsageBlock,
+    OpenAIChatCompletionFinishReason,
 )
 from litellm.types.llms.vertex_ai import (
     VERTEX_CREDENTIALS_TYPES,
@@ -810,6 +811,22 @@ class VertexGeminiConfig(VertexAIBaseConfig, BaseConfig):
 
         return usage
 
+    def _check_finish_reason(
+        self,
+        chat_completion_message: ChatCompletionResponseMessage,
+        finish_reason: Optional[str],
+    ) -> OpenAIChatCompletionFinishReason:
+        if chat_completion_message.get("function_call"):
+            return "function_call"
+        elif chat_completion_message.get("tool_calls"):
+            return "tool_calls"
+        elif finish_reason and (
+            finish_reason == "SAFETY" or finish_reason == "RECITATION"
+        ):  # vertex ai
+            return "content_filter"
+        else:
+            return "stop"
+
     def _process_candidates(self, _candidates, model_response, litellm_params):
         """Helper method to process candidates and extract metadata"""
         grounding_metadata: List[dict] = []
@@ -865,7 +882,9 @@ class VertexGeminiConfig(VertexAIBaseConfig, BaseConfig):
                 chat_completion_message["function_call"] = functions
 
             choice = litellm.Choices(
-                finish_reason=candidate.get("finishReason", "stop"),
+                finish_reason=self._check_finish_reason(
+                    chat_completion_message, candidate.get("finishReason")
+                ),
                 index=candidate.get("index", idx),
                 message=chat_completion_message,  # type: ignore
                 logprobs=chat_completion_logprobs,

litellm/proxy/_new_secret_config.yaml

@@ -57,6 +57,9 @@ model_list:
     litellm_params:
       model: text-embedding-ada-002
       api_key: os.environ/OPENAI_API_KEY
+  - model_name: gemini/gemini-2.0-flash
+    litellm_params:
+      model: gemini/gemini-2.0-flash
 
 litellm_settings:
   num_retries: 0

litellm/types/llms/openai.py

@@ -824,12 +824,12 @@ class OpenAIChatCompletionChunk(ChatCompletionChunk):
 
 class Hyperparameters(BaseModel):
     batch_size: Optional[Union[str, int]] = None  # "Number of examples in each batch."
-    learning_rate_multiplier: Optional[Union[str, float]] = (
-        None  # Scaling factor for the learning rate
-    )
-    n_epochs: Optional[Union[str, int]] = (
-        None  # "The number of epochs to train the model for"
-    )
+    learning_rate_multiplier: Optional[
+        Union[str, float]
+    ] = None  # Scaling factor for the learning rate
+    n_epochs: Optional[
+        Union[str, int]
+    ] = None  # "The number of epochs to train the model for"
 
 
 class FineTuningJobCreate(BaseModel):
@@ -856,18 +856,18 @@ class FineTuningJobCreate(BaseModel):
 
     model: str  # "The name of the model to fine-tune."
     training_file: str  # "The ID of an uploaded file that contains training data."
-    hyperparameters: Optional[Hyperparameters] = (
-        None  # "The hyperparameters used for the fine-tuning job."
-    )
-    suffix: Optional[str] = (
-        None  # "A string of up to 18 characters that will be added to your fine-tuned model name."
-    )
-    validation_file: Optional[str] = (
-        None  # "The ID of an uploaded file that contains validation data."
-    )
-    integrations: Optional[List[str]] = (
-        None  # "A list of integrations to enable for your fine-tuning job."
-    )
+    hyperparameters: Optional[
+        Hyperparameters
+    ] = None  # "The hyperparameters used for the fine-tuning job."
+    suffix: Optional[
+        str
+    ] = None  # "A string of up to 18 characters that will be added to your fine-tuned model name."
+    validation_file: Optional[
+        str
+    ] = None  # "The ID of an uploaded file that contains validation data."
+    integrations: Optional[
+        List[str]
+    ] = None  # "A list of integrations to enable for your fine-tuning job."
     seed: Optional[int] = None  # "The seed controls the reproducibility of the job."
 
 
@@ -1293,3 +1293,8 @@ class OpenAIModerationResponse(BaseLiteLLMOpenAIResponseObject):
 
     # Define private attributes using PrivateAttr
     _hidden_params: dict = PrivateAttr(default_factory=dict)
+
+
+OpenAIChatCompletionFinishReason = Literal[
+    "stop", "content_filter", "function_call", "tool_calls", "length"
+]

tests/llm_translation/base_llm_unit_tests.py

@@ -917,6 +917,7 @@ class BaseLLMChatTest(ABC):
             assert isinstance(
                 response.choices[0].message.tool_calls[0].function.arguments, str
             )
+            assert response.choices[0].finish_reason == "tool_calls"
             messages.append(
                 response.choices[0].message.model_dump()
             )  # Add assistant tool invokes