The MultimodalContextualRecallMetric is a multimodal metric designed to evaluate the recall of contextual information in generated outputs. It assesses how well the generated output recalls relevant information from the context, serving as a proxy for evaluating the performance of models that generate text based on multimodal inputs.

​
Required Arguments

  • input: A list containing the prompt or question related to the image or context.
  • actual_output: A list containing the generated textual description or answer.
  • expected_output: A list containing the expected textual description or answer.
  • retrieval_context: A list containing the contextual information or description of the image.

​
Optional Arguments

  • threshold: A float representing the minimum passing threshold, defaulted to 0.5.
  • model: A string specifying which of OpenAI’s GPT models to use, or any custom LLM model of type DeepEvalBaseLLM. Defaulted to ‘gpt-4o’.
  • include_reason: A boolean which, when set to True, includes a reason for its evaluation score. Defaulted to True.
  • strict_mode: A boolean which, when set to True, enforces a binary metric score: 1 for perfection, 0 otherwise. It also overrides the current threshold and sets it to 1. Defaulted to False.
  • async_mode: A boolean which, when set to True, enables concurrent execution within the measure() method. Defaulted to True.
  • verbose_mode: A boolean which, when set to True, prints the intermediate steps used to calculate the metric to the console. Defaulted to False.

​
Usage Example

import sys
import os

from agensight.eval.metrics import MultimodalContextualRecallMetric
from agensight.eval.test_case import MLLMTestCase
from agensight.eval.test_case import MLLMImage

input_data = ["Describe the image."]
actual_output = ["A cat is sitting on a windowsill."]
expected_output = ["A cat is sitting on a windowsill, looking outside."]
retrieval_context = ["A photo of a cat on a windowsill."]

metric = MultimodalContextualRecallMetric(model="gpt-4o", threshold=0.5)
test_case = MLLMTestCase(input=input_data, actual_output=actual_output, expected_output=expected_output, retrieval_context=retrieval_context)

metric.measure(test_case)
print(f"Score: {metric.score}")
print(f"Reason: {metric.reason}")