Failed to convert text into a pydantic model due to the following error: litellm.APIConnectionError: 'arguments'

Crews
,
1

Hello,
After going through the similar topics on this problem i still remain puzzled. As the title suggest lite llm seems to be throwing this strange error. I call no tools on that specific task it just is supposed to aggregate results from multiple reports. Maybe I’m missing something.

Code for my crew :

from crewai import Agent, Crew, Process, Task, LLM
from crewai.project import CrewBase, agent, crew, task
from crewai_tools import FileReadTool

from .firefighter_outputs import *
from .tools.output_validation import OutputValidation

@CrewBase
class FirefighterCrew:
	"""Firefighter Crew"""

	agents_config = 'config/agents.yaml'
	tasks_config = 'config/tasks.yaml'


	def __init__(self, route_tool, inventory_fn):
		"""Initialize the FirefighterCrew class."""
		super().__init__()
		self.route_tool = route_tool
		self.inventory_fn = inventory_fn
		self.llm = LLM(
            model = "ollama/llama3.1:latest",
            api_base = "http://localhost:11434",
            api_key = "NA"
        )

	@agent
	def firefighter_strategist(self) -> Agent:
		return Agent(
			config=self.agents_config['firefighter_strategist'],
    		allow_delegation=False,
			llm=self.llm,
            verbose=True
		)

	@agent
	def equipment_resource_manager(self) -> Agent:
		return Agent(
			config=self.agents_config['equipment_resource_manager'],
    		allow_delegation=False,
			llm=self.llm,
            verbose=True
		)

	@agent
	def unit_deployment_manager(self) -> Agent:
		return Agent(
			config=self.agents_config['unit_deployment_manager'],
    		allow_delegation=False,
			llm=self.llm,
            verbose=True
		)

	@task
	def read_inventory(self) -> Task:
		self.read_inventory_task =  Task(
            config=self.tasks_config['read_inventory'],
            # output_pydantic=ResourceReport,
			# async_execution=True,
			tools=[FileReadTool(self.inventory_fn)],
            # output_file='./reports/firefighter/initial_resource_report.json'
		)
		return self.read_inventory_task

	@task
	def create_resource_report(self) -> Task:
		self.create_resource_report_task =  Task(
            config=self.tasks_config['create_resource_report'],
            output_pydantic=ResourceReport,
			# async_execution=True,
			# tools=[OutputValidation(ResourceReport)],
			context=[self.read_inventory_task],
            output_file='./reports/firefighter/initial_resource_report.json',
			expected_output=f"The output should adhere to the following schema: {ResourceReport.get_schema()}"
		)
		return self.create_resource_report_task

	@task
	def create_firefighting_strategy(self) -> Task:
		self.create_firefighting_strategy_task = Task(
			config=self.tasks_config['create_firefighting_strategy'],
            output_pydantic=StrategyReport,
			# async_execution=True
			# tools=[OutputValidation(StrategyReport)],
			context=[self.create_resource_report_task],
            output_file='./reports/firefighter/firefighting_strategy.json',
			expected_output=f"The output should adhere to the following schema: {StrategyReport.get_schema()}"
		)
		return self.create_firefighting_strategy_task

	@task
	def generate_unit_deployment_report(self) -> Task:
		self.generate_unit_deployment_report_task = Task(
			config=self.tasks_config['generate_unit_deployment_report'],
			output_pydantic=UnitAllocationReport,
			# async_execution=True,
			context=[self.create_firefighting_strategy_task, self.create_resource_report_task],
			tools=[self.route_tool],#, OutputValidation(UnitAllocationReport)],
            output_file='./reports/firefighter/unit_deployment_report.json',
			expected_output=f"The output should adhere to the following schema: {UnitAllocationReport.get_schema()}"
		)
		return self.generate_unit_deployment_report_task

	@task
	def update_resource_report(self) -> Task:
		self.update_resource_report_task = Task(
			config=self.tasks_config['update_resource_report'],
			output_pydantic=UpdatedResourceReport,
			# async_execution=True,
			# tools=[OutputValidation(UpdatedResourceReport)],
			context=[self.create_firefighting_strategy_task, self.generate_unit_deployment_report_task],
            output_file='./reports/firefighter/updated_resource_report.json',
			expected_output=f"The output should adhere to the following schema: {UpdatedResourceReport.get_schema()}"
		)
		return self.update_resource_report_task

	@task
	def combine_reports(self) -> Task:
		return Task(
			config=self.tasks_config['combine_reports'],
			output_pydantic=FirefighterReports,
			context=[self.create_firefighting_strategy_task, self.update_resource_report_task, self.generate_unit_deployment_report_task],
            output_file='./reports/firefighter/full_final_report.json',
			expected_output=f"The output should adhere to the following schema: {FirefighterReports.get_schema()}"
		)

	@crew
	def crew(self) -> Crew:
		"""Creates the Research Crew"""
		return Crew(
			agents=self.agents, # Automatically created by the @agent decorator
			tasks=self.tasks, # Automatically created by the @task decorator
			process=Process.sequential,
			verbose=False,
			# planning=True,
			# planning_llm='ollama/llama3.1'
		)

Error log:

Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.


Provider List: https://docs.litellm.ai/docs/providers


Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.


Provider List: https://docs.litellm.ai/docs/providers


Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.


Provider List: https://docs.litellm.ai/docs/providers


Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.


Provider List: https://docs.litellm.ai/docs/providers


Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.


Provider List: https://docs.litellm.ai/docs/providers


Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.


Provider List: https://docs.litellm.ai/docs/providers


Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.


Provider List: https://docs.litellm.ai/docs/providers


Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.


Provider List: https://docs.litellm.ai/docs/providers


Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.


Provider List: https://docs.litellm.ai/docs/providers

 Failed to convert text into a pydantic model due to the following error: litellm.APIConnectionError: 'arguments'
Traceback (most recent call last):
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/crewai/utilities/converter.py", line 100, in convert_to_model
    escaped_result = json.dumps(json.loads(result, strict=False))
                                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/.pyenv/versions/3.11.0/lib/python3.11/json/__init__.py", line 359, in loads
    return cls(**kw).decode(s)
           ^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/.pyenv/versions/3.11.0/lib/python3.11/json/decoder.py", line 337, in decode
    obj, end = self.raw_decode(s, idx=_w(s, 0).end())
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/.pyenv/versions/3.11.0/lib/python3.11/json/decoder.py", line 355, in raw_decode
    raise JSONDecodeError("Expecting value", s, err.value) from None
json.decoder.JSONDecodeError: Expecting value: line 1 column 1 (char 0)

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/main.py", line 2735, in completion
    response = base_llm_http_handler.completion(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/llms/custom_httpx/llm_http_handler.py", line 225, in completion
    return provider_config.transform_response(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/llms/ollama/completion/transformation.py", line 264, in transform_response
    "arguments": json.dumps(function_call["arguments"]),
                            ~~~~~~~~~~~~~^^^^^^^^^^^^^
KeyError: 'arguments'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/instructor/retry.py", line 156, in retry_sync
    response = func(*args, **kwargs)
               ^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/utils.py", line 987, in wrapper
    raise e
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/utils.py", line 868, in wrapper
    result = original_function(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/main.py", line 3012, in completion
    raise exception_type(
          ^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/litellm_core_utils/exception_mapping_utils.py", line 2146, in exception_type
    raise e
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/litellm_core_utils/exception_mapping_utils.py", line 2122, in exception_type
    raise APIConnectionError(
litellm.exceptions.APIConnectionError: litellm.APIConnectionError: 'arguments'
Traceback (most recent call last):
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/crewai/utilities/converter.py", line 100, in convert_to_model
    escaped_result = json.dumps(json.loads(result, strict=False))
                                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/.pyenv/versions/3.11.0/lib/python3.11/json/__init__.py", line 359, in loads
    return cls(**kw).decode(s)
           ^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/.pyenv/versions/3.11.0/lib/python3.11/json/decoder.py", line 337, in decode
    obj, end = self.raw_decode(s, idx=_w(s, 0).end())
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/.pyenv/versions/3.11.0/lib/python3.11/json/decoder.py", line 355, in raw_decode
    raise JSONDecodeError("Expecting value", s, err.value) from None
json.decoder.JSONDecodeError: Expecting value: line 1 column 1 (char 0)

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/main.py", line 2735, in completion
    response = base_llm_http_handler.completion(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/llms/custom_httpx/llm_http_handler.py", line 225, in completion
    return provider_config.transform_response(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/llms/ollama/completion/transformation.py", line 264, in transform_response
    "arguments": json.dumps(function_call["arguments"]),
                            ~~~~~~~~~~~~~^^^^^^^^^^^^^
KeyError: 'arguments'


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

Traceback (most recent call last):
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/instructor/retry.py", line 151, in retry_sync
    for attempt in max_retries:
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/tenacity/__init__.py", line 443, in __iter__
    do = self.iter(retry_state=retry_state)
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/tenacity/__init__.py", line 376, in iter
    result = action(retry_state)
             ^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/tenacity/__init__.py", line 419, in exc_check
    raise retry_exc from fut.exception()
tenacity.RetryError: RetryError[<Future at 0x30065e8d0 state=finished raised APIConnectionError>]

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

Traceback (most recent call last):
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/crewai/utilities/converter.py", line 27, in to_pydantic
    return self._create_instructor().to_pydantic()
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/crewai/utilities/internal_instructor.py", line 45, in to_pydantic
    model = self._client.chat.completions.create(
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/instructor/client.py", line 176, in create
    return self.create_fn(
           ^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/instructor/patch.py", line 193, in new_create_sync
    response = retry_sync(
               ^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/instructor/retry.py", line 181, in retry_sync
    raise InstructorRetryException(
instructor.exceptions.InstructorRetryException: litellm.APIConnectionError: 'arguments'
Traceback (most recent call last):
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/crewai/utilities/converter.py", line 100, in convert_to_model
    escaped_result = json.dumps(json.loads(result, strict=False))
                                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/.pyenv/versions/3.11.0/lib/python3.11/json/__init__.py", line 359, in loads
    return cls(**kw).decode(s)
           ^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/.pyenv/versions/3.11.0/lib/python3.11/json/decoder.py", line 337, in decode
    obj, end = self.raw_decode(s, idx=_w(s, 0).end())
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/.pyenv/versions/3.11.0/lib/python3.11/json/decoder.py", line 355, in raw_decode
    raise JSONDecodeError("Expecting value", s, err.value) from None
json.decoder.JSONDecodeError: Expecting value: line 1 column 1 (char 0)

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/main.py", line 2735, in completion
    response = base_llm_http_handler.completion(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/llms/custom_httpx/llm_http_handler.py", line 225, in completion
    return provider_config.transform_response(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/llms/ollama/completion/transformation.py", line 264, in transform_response
    "arguments": json.dumps(function_call["arguments"]),
                            ~~~~~~~~~~~~~^^^^^^^^^^^^^
KeyError: 'arguments'


During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/main.py", line 2735, in completion
    response = base_llm_http_handler.completion(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/llms/custom_httpx/llm_http_handler.py", line 225, in completion
    return provider_config.transform_response(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/llms/ollama/completion/transformation.py", line 264, in transform_response
    "arguments": json.dumps(function_call["arguments"]),
                            ~~~~~~~~~~~~~^^^^^^^^^^^^^
KeyError: 'arguments'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/instructor/retry.py", line 156, in retry_sync
    response = func(*args, **kwargs)
               ^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/utils.py", line 987, in wrapper
    raise e
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/utils.py", line 868, in wrapper
    result = original_function(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/main.py", line 3012, in completion
    raise exception_type(
          ^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/litellm_core_utils/exception_mapping_utils.py", line 2146, in exception_type
    raise e
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/litellm_core_utils/exception_mapping_utils.py", line 2122, in exception_type
    raise APIConnectionError(
litellm.exceptions.APIConnectionError: litellm.APIConnectionError: 'arguments'
Traceback (most recent call last):
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/crewai/utilities/converter.py", line 100, in convert_to_model
    escaped_result = json.dumps(json.loads(result, strict=False))
                                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/.pyenv/versions/3.11.0/lib/python3.11/json/__init__.py", line 359, in loads
    return cls(**kw).decode(s)
           ^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/.pyenv/versions/3.11.0/lib/python3.11/json/decoder.py", line 337, in decode
    obj, end = self.raw_decode(s, idx=_w(s, 0).end())
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/.pyenv/versions/3.11.0/lib/python3.11/json/decoder.py", line 355, in raw_decode
    raise JSONDecodeError("Expecting value", s, err.value) from None
json.decoder.JSONDecodeError: Expecting value: line 1 column 1 (char 0)

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/main.py", line 2735, in completion
    response = base_llm_http_handler.completion(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/llms/custom_httpx/llm_http_handler.py", line 225, in completion
    return provider_config.transform_response(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/llms/ollama/completion/transformation.py", line 264, in transform_response
    "arguments": json.dumps(function_call["arguments"]),
                            ~~~~~~~~~~~~~^^^^^^^^^^^^^
KeyError: 'arguments'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/instructor/retry.py", line 156, in retry_sync
    response = func(*args, **kwargs)
               ^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/utils.py", line 987, in wrapper
    raise e
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/utils.py", line 868, in wrapper
    result = original_function(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/evaveli/Desktop/MAI-URV-IMAS/emergency_response/.venv/lib/python3.11/site-packages/litellm/main.py", line 3012, in completion
    raise exception_type(
          ^^^^^^^^^^^^^^
KeyError: 'arguments'
 Using raw output instead.

I would be happy to provide any further code if needed.

Thank you.

2

i have the same question

3

According to the documentation on LLM providers, your code should look like this:

self.llm = LLM(
    model="ollama/llama3.1:latest",
    base_url="http://localhost:11434"
)
4

I have a similar quetion, i got similar error but with json_output:

this is the code:

class Analysis(BaseModel):
    analysis: str

class Research(BaseModel):
    research: str

@CrewBase
class MarketAnalysis():

    def __init__(self):
        load_dotenv()

    @agent
    def researcher(self) -> Agent:
        """
        Restituisce un'istanza di un agente configurato come "researcher".
        La funzione crea e restituisce un oggetto di tipo `Agent` utilizzando
        la configurazione specifica per il ruolo di "researcher". L'agente
        viene inizializzato con un'opzione di verbosità attivata.        
        """
        
        return Agent(
            config=self.agents_config["researcher"],
            verbose=False,
        )
    
    @agent
    def analyst(self) -> Agent:
        """
        Restituisce un'istanza di un agente configurato come "analyst".
        La funzione crea e restituisce un oggetto di tipo `Agent` utilizzando
        la configurazione specifica per il ruolo di "analyst". L'agente
        viene inizializzato con un'opzione di verbosità attivata.        
        """
        
        return Agent(
            config=self.agents_config['analyst'],
            verbose=False,
        )
    
    @task
    def research(self) -> Task:
        """
        Restituisce un'istanza di un task configurato come "research".
        La funzione crea e restituisce un oggetto di tipo `Task` utilizzando
        la configurazione specifica per il task di "research". Il task
        viene inizializzato con un'opzione di verbosità attivata.        
        """
        
        return Task(
            config=self.tasks_config['research'],
            verbose=False,
            output_pydantic=Research,
        )
    
    @task
    def analysis(self) -> Task:
        """
        Restituisce un'istanza di un task configurato come "analysis".
        La funzione crea e restituisce un oggetto di tipo `Task` utilizzando
        la configurazione specifica per il task di "analysis". Il task
        viene inizializzato con un'opzione di verbosità attivata.        
        """
        
        return Task(
            config=self.tasks_config['analysis'],
            verbose=False,
            output_json=Analysis
        )
        
    @crew
    def crew(self) -> Crew:
        """
        Restituisce un'istanza di un equipaggio configurato come "crew".
        La funzione crea e restituisce un oggetto di tipo `Crew` utilizzando
        la configurazione specifica per l'equipaggio di "crew". L'equipaggio
        viene inizializzato con un'opzione di verbosità attivata.        
        """
        
        return Crew(
            agents=self.agents,
            tasks=self.tasks,
            verbose=False
        )
    
    def get_research_output(self) -> TaskOutput:
        """
        Restituisce l'output del task di ricerca.
        """
        
        raw_output = self.analysis().output.raw
        pydantic_output = self.research().output.pydantic
        description = self.research().output.description
        summary = self.research().output.summary

        return {
            "raw_output" : raw_output, 
            "pydantic_output": pydantic_output, 
            "description": description, 
            "summary": summary
        }
    
    def get_analysis_output(self) -> TaskOutput:
        """
        Restituisce l'output del task di analisi.
        """
        
        raw_output = self.analysis().output.raw
        json_output = json.dumps(self.analysis().output.json_dict, indent=2)
        description = self.analysis().output.description
        summary = self.analysis().output.summary

        return {
            "raw_output" : raw_output, 
            "json_output": json_output, 
            "description": description, 
            "summary": summary
        }

This is task.yaml:

research:
  description: >
   Di cercare lo scenario attuale nel mercato per soluzioni nell'ambito {topic} basate sull'intelligenza artificiale
  expected_output: >
    Dati di mercato completi inclusi principali attori, dimensione del mercato e tendenze di crescita
  agent: researcher

analysis:
  description: >
    Analizzare i dati di mercato e identificare le {number} principali opportunità di investimento
  expected_output: >
    Report analitico con {number} opportunità di investimento consigliate e le relative motivazioni
    Formattato in Markdown senza '```'.
  agent: analyst

This is agents.yaml

researcher:
  role: >
    Specialista in ricerche di mercato nell'ambito: {topic}
  goal: >
    Trovare dati di mercato completi in ambito {topic}
  backstory: >
    Sei un esperto nell'individuare le tendenze di mercato nel raccogliere dati
  llm: together_ai/meta-llama/Llama-3.3-70B-Instruct-Turbo-Free

analyst:
  role: >
    Analista di Mercato
  goal: >
    Analizzare i dati di mercato e identificare le principali opportunità
  backstory: >
    Eccelle nell'interpretare i dati di mercato e nel cogliere intuizioni di valore
  llm: together_ai/meta-llama/Llama-3.3-70B-Instruct-Turbo-Free

I got the error:
Failed to convert text into JSON, error: litellm.BadRequestError: Together_aiException - invalid tools grammar: Model supports only tool_choice auto. Using raw output instead.

5

Hey Simone,

So, here’s a tip: when you use Task.output_json (or even Task.output_pydantic), CrewAI actually beefs up the prompt behind the scenes with extra instructions to make generating that JSON smoother. That means for your Task.expected_output, you should be more concise – just get straight to the point.

I went ahead and removed that extra instruction from your config/tasks.yaml. Here’s how it looks now:

analysis:
  description: >
    Analizzare i dati di mercato e identificare le {number} principali
    opportunità di investimento.
  expected_output: >
    Report analitico con {number} opportunità di investimento consigliate e
    le relative motivazioni.
  agent: analyst

With that tweak, just run your Crew, and you should be golden:

market_analysis = MarketAnalysis()

market_analysis.crew().kickoff(
    inputs={
        "topic": "Calcolo Quantistico",
        "number": "cinque"
    }
)

print(f"{market_analysis.get_research_output()=}")
print(f"{market_analysis.get_analysis_output()=}")
6

This topic was automatically closed 24 hours after the last reply. New replies are no longer allowed.