Including more than 1 tool for a Gemini Model results in 400 - Request contains an invalid argument

[expresspotato]

Environment details

• OS type and version: Mac OS / Python

Steps to reproduce

1. Include more than one Tool object
2. Generate

Code example

'''
Created on May 10, 2024

@author: Kevin
'''

from django.core.management.base import BaseCommand

import vertexai
from vertexai.generative_models import GenerativeModel, Tool
from vertexai.preview import generative_models as preview_generative_models
from vertexai.generative_models import (
    Content,
    FunctionDeclaration,
    GenerationConfig,
    GenerativeModel,
    Part,
    Tool,
)
from pprint import pprint

PROJECT_ID = ""
LOCATION_ID = ""
AGENT_ID = ""

MODEL_ID = 'gemini-1.5-pro-preview-0409'

class ModelFunction():
    def __init__(self):
        self.function_name = self.__class__.__name__
        self.data = {}

class MFGetCurrentWeather(ModelFunction):
    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self.description = "Get the current weather in a given location"
        self.parameters = [("location", "string", "Location"), ]

    def set_parameter_values(self, location=None):
        self.data['location'] = location

    def get_functions_gemini(self):
        properties = {}
        for item in self.parameters: properties[item[0]] = {"type": item[1], "description": item[2]}

        return [
            FunctionDeclaration(
                name=self.function_name,
                description=self.description,
                parameters={
                    "type": "object",
                    "properties": properties,
                },
            )
        ]

    def get_response_gemini(self):
        return """{ "location": "Boston, MA", "temperature": 38, "description": "Partly Cloudy", "icon": "partly-cloudy", "humidity": 65, "wind": { "speed": 10, "direction": "NW" } }"""

class MFGetCurrentTimezone(ModelFunction):
    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self.description = "Get the current timezone in a given location"
        self.parameters = [("location", "string", "Location"), ]

    def set_parameter_values(self, location=None):
        self.data['location'] = location

    def get_functions_gemini(self):
        properties = {}
        for item in self.parameters: properties[item[0]] = {"type": item[1], "description": item[2]}

        return [
            FunctionDeclaration(
                name=self.function_name,
                description=self.description,
                parameters={
                    "type": "object",
                    "properties": properties,
                },
            )
        ]

    def get_response_gemini(self):
        return """{ "location": "Boston, MA", "timeZone": {"name": "US / Eastern", "friendlyName": "Eastern Time"}} """

class Command(BaseCommand):
    help = 'Test basic VertexAI / Gemini functionality'

    def handle(self, *args, **options):
        print('--> Test...')

        generation_config = {
            'candidate_count': 1,
            'temperature': 0.1,
            'top_k': 5,
            'top_p': 0.90,
            'max_output_tokens': 1000,
        }

        messages = []

        prompt = 'What is the timezone in Boston?'

        messages.append({'role': 'user', 'parts': [{'text': prompt}]})

        mfgcw = MFGetCurrentWeather()
        mfgct = MFGetCurrentTimezone()

        weather_tool = Tool(function_declarations=mfgcw.get_functions_gemini())
        timezone_tool = Tool(function_declarations=mfgct.get_functions_gemini())

        # import ipdb; ipdb.set_trace();

        vertexai.init(project=PROJECT_ID, location=LOCATION_ID)
        gemini_model = GenerativeModel(MODEL_ID, system_instruction=[])

        model_response = gemini_model.generate_content(messages, tools=[timezone_tool, weather_tool], generation_config=generation_config)

        pprint(model_response)

        function_call = model_response.candidates[0].function_calls[0]
        print(f'--> Model requests function call: \n{function_call}')

        if function_call:
            if function_call.name == mfgcw.function_name:
                mfgcw.set_parameter_values(location=function_call.args['location'])
                function_response = mfgcw.get_response_gemini()
            elif function_call.name == mfgct.function_name:
                mfgct.set_parameter_values(location=function_call.args['location'])
                function_response = mfgct.get_response_gemini()

            # Return the API response to Gemini so it can generate a model response or request another function call
            response = gemini_model.generate_content(
                [
                    Content(role="user", parts=[Part.from_text(prompt)]),
                    model_response.candidates[0].content,  # Function call response
                    Content(
                        parts=[
                            Part.from_function_response(
                                name=mfgcw.function_name,
                                response={
                                    "content": function_response,  # Return the API response to Gemini
                                },
                            ),
                        ],
                    ),
                ],
                tools=[weather_tool],
            )

            pprint(response)
            pprint(response.candidates[0].content.parts[0].text)

Stack trace

google.api_core.exceptions.InvalidArgument: 400 Request contains an invalid argument.

Tried both tools on their own and the response is fine.

[said-rasidin]

I also get the same error for using example grounding using Google search with the example provided in this, already using vertexai version 1.51.0

import vertexai

from vertexai.generative_models import grounding
from vertexai.generative_models import GenerationConfig, GenerativeModel, Tool

# TODO(developer): Update and un-comment below line
project_id = "PROJECT ID"

vertexai.init(project=project_id, location="us-central1")

model = GenerativeModel(model_name="gemini-1.0-pro-002")

# Use Google Search for grounding
tool = Tool.from_google_search_retrieval(grounding.GoogleSearchRetrieval())

prompt = "When is the next total solar eclipse in US?"
response = model.generate_content(
    prompt,
    tools=[tool],
    generation_config=GenerationConfig(
        temperature=0.0,
    ),
)
print(response)

error return:

_InactiveRpcError                         Traceback (most recent call last)
[/usr/local/lib/python3.10/dist-packages/google/api_core/grpc_helpers.py](https://localhost:8080/#) in error_remapped_callable(*args, **kwargs)
     71         try:
---> 72             return callable_(*args, **kwargs)
     73         except grpc.RpcError as exc:

7 frames
_InactiveRpcError: <_InactiveRpcError of RPC that terminated with:
    status = StatusCode.INVALID_ARGUMENT
    details = "Request contains an invalid argument."
    debug_error_string = "UNKNOWN:Error received from peer ipv4:142.251.163.95:443 {grpc_message:"Request contains an invalid argument.", grpc_status:3, created_time:"2024-05-11T14:06:26.807301202+00:00"}"
>

The above exception was the direct cause of the following exception:

InvalidArgument                           Traceback (most recent call last)
[/usr/local/lib/python3.10/dist-packages/google/api_core/grpc_helpers.py](https://localhost:8080/#) in error_remapped_callable(*args, **kwargs)
     72             return callable_(*args, **kwargs)
     73         except grpc.RpcError as exc:
---> 74             raise exceptions.from_grpc_error(exc) from exc
     75 
     76     return error_remapped_callable

InvalidArgument: 400 Request contains an invalid argument.

[expresspotato]

I also get the same error for using example grounding using Google search with the example provided in this, already using vertexai version 1.51.0

import vertexai

from vertexai.generative_models import grounding
from vertexai.generative_models import GenerationConfig, GenerativeModel, Tool

# TODO(developer): Update and un-comment below line
project_id = "PROJECT ID"

vertexai.init(project=project_id, location="us-central1")

model = GenerativeModel(model_name="gemini-1.0-pro-002")

# Use Google Search for grounding
tool = Tool.from_google_search_retrieval(grounding.GoogleSearchRetrieval())

prompt = "When is the next total solar eclipse in US?"
response = model.generate_content(
    prompt,
    tools=[tool],
    generation_config=GenerationConfig(
        temperature=0.0,
    ),
)
print(response)

error return:

_InactiveRpcError                         Traceback (most recent call last)
[/usr/local/lib/python3.10/dist-packages/google/api_core/grpc_helpers.py](https://localhost:8080/#) in error_remapped_callable(*args, **kwargs)
     71         try:
---> 72             return callable_(*args, **kwargs)
     73         except grpc.RpcError as exc:

7 frames
_InactiveRpcError: <_InactiveRpcError of RPC that terminated with:
  status = StatusCode.INVALID_ARGUMENT
  details = "Request contains an invalid argument."
  debug_error_string = "UNKNOWN:Error received from peer ipv4:142.251.163.95:443 {grpc_message:"Request contains an invalid argument.", grpc_status:3, created_time:"2024-05-11T14:06:26.807301202+00:00"}"
>

The above exception was the direct cause of the following exception:

InvalidArgument                           Traceback (most recent call last)
[/usr/local/lib/python3.10/dist-packages/google/api_core/grpc_helpers.py](https://localhost:8080/#) in error_remapped_callable(*args, **kwargs)
     72             return callable_(*args, **kwargs)
     73         except grpc.RpcError as exc:
---> 74             raise exceptions.from_grpc_error(exc) from exc
     75 
     76     return error_remapped_callable

InvalidArgument: 400 Request contains an invalid argument.

Your code seems valid, but I suspect you've not agreed to the terms for grounding. See here: https://cloud.google.com/vertex-ai/generative-ai/docs/multimodal/ground-gemini

[Ark-kun]

@expresspotato Can you please try putting all function declarations in a single Tool?

P.S. AFAIK, some of the most recent models might support multiple tools.

[koverholt]

Sorry to hear about your troubles here. We've (Google DevRel) also heard reports from users who want to use different types of tools (e.g., grounding/retrieval tools along with FunctionDeclarations).

There is a related issue in the generative-ai repo at https://github.com/GoogleCloudPlatform/generative-ai/issues/636, and they've opened an issue in the public issue tracker for Vertex AI here: https://issuetracker.google.com/issues/340729475. Feel free to add more details and/or +1 the latter issue in the public tracker to increase the signal to the product teams. Thanks!