Results are not in the same order as input test cases

[joaopcm1996]

Evaluation results are out of order, which makes it a pain to use them in a downstream process, or if I'm evaluating 2 models predictions at once. The 2 model's predictions and scores get mixed up.

Steps to reproduce the behavior:

from deepeval import evaluate
from deepeval.metrics import FaithfulnessMetric
from deepeval.test_case import LLMTestCase

cases = []
for idx in range(3):
    cases.append(LLMTestCase(
        input=f"input{idx}",
        actual_output=f'The context says speaker {idx} is a bad person',
        retrieval_context=[f"context{idx}"])
    )

metric = FaithfulnessMetric(model='gpt-4o',include_reason=True)

results = evaluate(cases, [metric], print_results=False)

for idx,i in enumerate(results):
    print(i.input)
    print(i.actual_output)
    print(i.retrieval_context)
    print(cases[idx].retrieval_context)
    print(i.metrics_data[0].verbose_logs)
    print('\n\n')

Expected behavior
The results list should be ordered the same way as cases list.

[raresaxpo]

I encountered the same issue. I am trying to match the metadata of the test cases in the evaluation dataset with the test results. Without the preservation of order, this is impossible..