chore: address PR feedback for test functionality

Co-Authored-By: Joe Moura <joao@crewai.com>

src/crewai/crew.py

@@ -6,6 +6,7 @@ from concurrent.futures import Future
 from hashlib import md5
 from typing import Any, Callable, Dict, List, Optional, Tuple, Union
 
+from langchain_core.language_models.base import BaseLanguageModel
 from pydantic import (
     UUID4,
     BaseModel,
@@ -1075,19 +1076,39 @@ class Crew(BaseModel):
     def test(
         self,
         n_iterations: int,
-        openai_model_name: Optional[str] = None,
+        llm: Optional[Union[str, InstanceOf[LLM], Any]] = None,
+        openai_model_name: Optional[str] = None,  # Kept for backward compatibility
         inputs: Optional[Dict[str, Any]] = None,
     ) -> None:
-        """Test and evaluate the Crew with the given inputs for n iterations concurrently using concurrent.futures."""
+        """Test and evaluate the Crew with the given inputs for n iterations.
+
+        Args:
+            n_iterations (int): Number of test iterations to run
+            llm (Optional[Union[str, LLM, BaseLanguageModel]]): Language model to use for testing
+            openai_model_name (Optional[str]): Legacy parameter for OpenAI models (deprecated)
+            inputs (Optional[Dict[str, Any]]): Test inputs for the crew
+
+        Raises:
+            ValueError: If n_iterations is less than 1 or if llm type is unsupported
+
+        Returns:
+            None
+        """
+        if n_iterations < 1:
+            raise ValueError("n_iterations must be greater than 0")
+        if llm is not None and not isinstance(llm, (str, LLM, BaseLanguageModel)):
+            raise ValueError(f"Unsupported LLM type: {type(llm)}")
+
         test_crew = self.copy()
+        test_llm = llm if llm is not None else openai_model_name
 
         self._test_execution_span = test_crew._telemetry.test_execution_span(
             test_crew,
             n_iterations,
             inputs,
-            openai_model_name,  # type: ignore[arg-type]
+            test_llm,  # type: ignore[arg-type]
         )  # type: ignore[arg-type]
-        evaluator = CrewEvaluator(test_crew, openai_model_name)  # type: ignore[arg-type]
+        evaluator = CrewEvaluator(test_crew, test_llm)  # type: ignore[arg-type]
 
         for i in range(1, n_iterations + 1):
             evaluator.set_iteration(i)

src/crewai/utilities/evaluators/crew_evaluator_handler.py

@@ -1,14 +1,19 @@
+import os
 from collections import defaultdict
+from typing import Any, Dict, List, Optional, Union
 
-from pydantic import BaseModel, Field
+from langchain_core.language_models.base import BaseLanguageModel
+from pydantic import BaseModel, Field, InstanceOf
 from rich.box import HEAVY_EDGE
 from rich.console import Console
 from rich.table import Table
 
 from crewai.agent import Agent
+from crewai.llm import LLM
 from crewai.task import Task
 from crewai.tasks.task_output import TaskOutput
 from crewai.telemetry import Telemetry
+from crewai.utilities.logger import Logger
 
 
 class TaskEvaluationPydanticOutput(BaseModel):
@@ -22,22 +27,62 @@ class CrewEvaluator:
     A class to evaluate the performance of the agents in the crew based on the tasks they have performed.
 
     Attributes:
-        crew (Crew): The crew of agents to evaluate.
-        openai_model_name (str): The model to use for evaluating the performance of the agents (for now ONLY OpenAI accepted).
-        tasks_scores (defaultdict): A dictionary to store the scores of the agents for each task.
-        iteration (int): The current iteration of the evaluation.
+        crew (Crew): The crew of agents to evaluate
+        llm (Union[str, LLM, BaseLanguageModel]): Language model to use for evaluation
+        tasks_scores (defaultdict): Dictionary to store the scores of the agents for each task
+        iteration (int): Current iteration of the evaluation
+        run_execution_times (defaultdict): Dictionary to store execution times for each run
     """
 
     tasks_scores: defaultdict = defaultdict(list)
     run_execution_times: defaultdict = defaultdict(list)
     iteration: int = 0
 
-    def __init__(self, crew, openai_model_name: str):
+    def __init__(self, crew, llm: Union[str, InstanceOf[LLM], BaseLanguageModel]):
+        """Initialize the CrewEvaluator.
+        
+        Args:
+            crew (Crew): The crew to evaluate
+            llm (Union[str, LLM, BaseLanguageModel]): Language model to use for evaluation
+        
+        Raises:
+            ValueError: If llm is of an unsupported type
+        """
+        if not isinstance(llm, (str, LLM, BaseLanguageModel, type(None))):
+            raise ValueError(f"Unsupported LLM type: {type(llm)}")
+
         self.crew = crew
-        self.openai_model_name = openai_model_name
+        self.llm = llm
         self._telemetry = Telemetry()
+        self._logger = Logger()
+        self._setup_llm()
         self._setup_for_evaluating()
 
+    def _setup_llm(self):
+        """Set up the LLM following the Agent class pattern.
+        
+        This method initializes the language model based on the provided llm parameter:
+        - If string: creates new LLM instance with model name
+        - If LLM instance: uses as-is
+        - If None: uses default model from environment or "gpt-4"
+        - Otherwise: attempts to extract model name from object attributes
+        """
+        if isinstance(self.llm, str):
+            self.llm = LLM(model=self.llm)
+        elif isinstance(self.llm, LLM):
+            pass
+        elif self.llm is None:
+            model_name = os.environ.get("OPENAI_MODEL_NAME") or "gpt-4"
+            self.llm = LLM(model=model_name)
+        else:
+            llm_params = {
+                "model": getattr(self.llm, "model_name", None)
+                or getattr(self.llm, "deployment_name", None)
+                or str(self.llm),
+            }
+            self.llm = LLM(**llm_params)
+
+
     def _setup_for_evaluating(self) -> None:
         """Sets up the crew for evaluating."""
         for task in self.crew.tasks:
@@ -51,7 +96,7 @@ class CrewEvaluator:
             ),
             backstory="Evaluator agent for crew evaluation with precise capabilities to evaluate the performance of the agents in the crew based on the tasks they have performed",
             verbose=False,
-            llm=self.openai_model_name,
+            llm=self.llm,
         )
 
     def _evaluation_task(
@@ -157,35 +202,48 @@ class CrewEvaluator:
         console.print(table)
 
     def evaluate(self, task_output: TaskOutput):
-        """Evaluates the performance of the agents in the crew based on the tasks they have performed."""
-        current_task = None
-        for task in self.crew.tasks:
-            if task.description == task_output.description:
-                current_task = task
-                break
+        """Evaluates the performance of the agents in the crew based on the tasks they have performed.
+        
+        Args:
+            task_output (TaskOutput): The output from the task to evaluate
 
-        if not current_task or not task_output:
-            raise ValueError(
-                "Task to evaluate and task output are required for evaluation"
+        Raises:
+            ValueError: If task to evaluate or task output is missing, or if evaluation result is invalid
+        """
+        try:
+            current_task = None
+            for task in self.crew.tasks:
+                if task.description == task_output.description:
+                    current_task = task
+                    break
+
+            if not current_task or not task_output:
+                raise ValueError(
+                    "Task to evaluate and task output are required for evaluation"
+                )
+
+            self._logger.log("info", f"Starting evaluation for task: {task_output.description}")
+            evaluator_agent = self._evaluator_agent()
+            evaluation_task = self._evaluation_task(
+                evaluator_agent, current_task, task_output.raw
             )
 
-        evaluator_agent = self._evaluator_agent()
-        evaluation_task = self._evaluation_task(
-            evaluator_agent, current_task, task_output.raw
-        )
+            evaluation_result = evaluation_task.execute_sync()
 
-        evaluation_result = evaluation_task.execute_sync()
-
-        if isinstance(evaluation_result.pydantic, TaskEvaluationPydanticOutput):
-            self._test_result_span = self._telemetry.individual_test_result_span(
-                self.crew,
-                evaluation_result.pydantic.quality,
-                current_task._execution_time,
-                self.openai_model_name,
-            )
-            self.tasks_scores[self.iteration].append(evaluation_result.pydantic.quality)
-            self.run_execution_times[self.iteration].append(
-                current_task._execution_time
-            )
-        else:
-            raise ValueError("Evaluation result is not in the expected format")
+            if isinstance(evaluation_result.pydantic, TaskEvaluationPydanticOutput):
+                self._test_result_span = self._telemetry.individual_test_result_span(
+                    self.crew,
+                    evaluation_result.pydantic.quality,
+                    current_task._execution_time,
+                    self.llm,
+                )
+                self.tasks_scores[self.iteration].append(evaluation_result.pydantic.quality)
+                self.run_execution_times[self.iteration].append(
+                    current_task._execution_time
+                )
+                self._logger.log("info", f"Evaluation completed with score: {evaluation_result.pydantic.quality}")
+            else:
+                raise ValueError("Evaluation result is not in the expected format")
+        except Exception as e:
+            self._logger.log("error", f"Evaluation failed: {str(e)}")
+            raise

tests/crew_test.py

@@ -24,6 +24,36 @@ from crewai.types.usage_metrics import UsageMetrics
 from crewai.utilities import Logger
 from crewai.utilities.rpm_controller import RPMController
 from crewai.utilities.task_output_storage_handler import TaskOutputStorageHandler
+from crewai.llm import LLM
+
+class MockLLM(LLM):
+    """Mock LLM for testing."""
+    def __init__(self):
+        super().__init__(model="gpt-4")  # Use a known model name
+        
+    def chat_completion(self, messages, tools=None, tool_choice=None, **kwargs):
+        # Mock a proper response that matches the expected format
+        if tools and any('output' in tool.get('function', {}).get('name', '') for tool in tools):
+            return {
+                "choices": [{
+                    "message": {
+                        "content": None,
+                        "role": "assistant",
+                        "function_call": {
+                            "name": "output",
+                            "arguments": '{"quality": 8.5}'
+                        }
+                    }
+                }]
+            }
+        return {
+            "choices": [{
+                "message": {
+                    "content": "Mock LLM Response",
+                    "role": "assistant"
+                }
+            }]
+        }
 
 ceo = Agent(
     role="CEO",
@@ -47,6 +77,60 @@ writer = Agent(
 )
 
 
+def test_crew_test_with_custom_llm():
+    """Test that Crew.test() works with a custom LLM implementation."""
+    task = Task(
+        description="Test task",
+        expected_output="Test output",
+        agent=researcher,
+    )
+    crew = Crew(agents=[researcher], tasks=[task])
+    
+    # Test with custom LLM
+    custom_llm = MockLLM()
+    crew.test(n_iterations=1, llm=custom_llm)
+    # No assertion needed as we just verify it runs without errors
+
+def test_crew_test_backward_compatibility():
+    """Test that Crew.test() maintains backward compatibility with openai_model_name."""
+    task = Task(
+        description="Test task",
+        expected_output="Test output",
+        agent=researcher,
+    )
+    crew = Crew(agents=[researcher], tasks=[task])
+    
+    # Test with openai_model_name
+    crew.test(n_iterations=1, openai_model_name="gpt-4")
+    # No assertion needed as we just verify it runs without errors
+
+def test_crew_test_with_invalid_llm():
+    """Test that Crew.test() properly handles invalid LLM inputs."""
+    task = Task(
+        description="Test task",
+        expected_output="Test output",
+        agent=researcher,
+    )
+    crew = Crew(agents=[researcher], tasks=[task])
+    
+    # Test with invalid LLM type
+    with pytest.raises(ValueError, match="Unsupported LLM type"):
+        crew.test(n_iterations=1, llm=123)  # type: ignore
+
+def test_crew_test_with_invalid_iterations():
+    """Test that Crew.test() validates n_iterations."""
+    task = Task(
+        description="Test task",
+        expected_output="Test output",
+        agent=researcher,
+    )
+    crew = Crew(agents=[researcher], tasks=[task])
+    
+    # Test with invalid n_iterations
+    with pytest.raises(ValueError, match="n_iterations must be greater than 0"):
+        crew.test(n_iterations=0, llm=MockLLM())
+
+
 def test_crew_config_conditional_requirement():
     with pytest.raises(ValueError):
         Crew(process=Process.sequential)
@@ -1123,7 +1207,7 @@ def test_kickoff_for_each_empty_input():
     assert results == []
 
 
-@pytest.mark.vcr(filter_headers=["authorization"])
+@pytest.mark.vcr(filter_headeruvs=["authorization"])
 def test_kickoff_for_each_invalid_input():
     """Tests if kickoff_for_each raises TypeError for invalid input types."""
 
@@ -3125,4 +3209,4 @@ def test_multimodal_agent_live_image_analysis():
     # Verify we got a meaningful response
     assert isinstance(result.raw, str)
     assert len(result.raw) > 100  # Expecting a detailed analysis
-    assert "error" not in result.raw.lower()  # No error messages in response
+    assert "error" not in result.raw.lower()  # No error messages in response

tests/utilities/evaluators/test_crew_evaluator_handler.py

@@ -23,7 +23,7 @@ class TestCrewEvaluator:
         )
         crew = Crew(agents=[agent], tasks=[task])
 
-        return CrewEvaluator(crew, openai_model_name="gpt-4o-mini")
+        return CrewEvaluator(crew, llm="openai/gpt-4o-mini")
 
     def test_setup_for_evaluating(self, crew_planner):
         crew_planner._setup_for_evaluating()
@@ -45,7 +45,7 @@ class TestCrewEvaluator:
             == "Evaluator agent for crew evaluation with precise capabilities to evaluate the performance of the agents in the crew based on the tasks they have performed"
         )
         assert agent.verbose is False
-        assert agent.llm.model == "gpt-4o-mini"
+        assert agent.llm.model == "openai/gpt-4o-mini"
 
     def test_evaluation_task(self, crew_planner):
         evaluator_agent = Agent(