wehub-resource-sync
33 lines
693 B
Python
33 lines
693 B
Python
"""
|
|
Simple RAG evaluation example using DeepEval.
|
|
Runs Answer Relevancy metric on sample outputs.
|
|
"""
|
|
|
|
import json
|
|
import os
|
|
from deepeval.test_case import LLMTestCase
|
|
from deepeval.metrics import AnswerRelevancyMetric
|
|
|
|
|
|
BASE_DIR = os.path.dirname(__file__)
|
|
DATA_PATH = os.path.join(BASE_DIR, "outputs.json")
|
|
|
|
with open(DATA_PATH) as f:
|
|
data = json.load(f)
|
|
|
|
metric = AnswerRelevancyMetric(threshold=0.7)
|
|
|
|
print("\n--- DeepEval Results ---\n")
|
|
|
|
for item in data:
|
|
|
|
test_case = LLMTestCase(
|
|
input=item["input"], actual_output=item["actual_output"]
|
|
)
|
|
|
|
score = metric.measure(test_case)
|
|
|
|
print(f"Input: {item['input']}")
|
|
print(f"Score: {score}")
|
|
print("-" * 40)
|