ragas_rag_pipeline_evaluation_with_faithfulness_and_relevancy_metrics.py

1import os
2from datasets import load_dataset
3from ragas import evaluate
4from ragas.metrics import faithfulness, answer_relevancy, context_precision, context_recall
5
6# 1. Load the sample dataset
7# This dataset contains 'question', 'contexts', 'answer', and 'ground_truth'
8dataset = load_dataset("explodinggradients/fiqa", "ragas_eval")
9
10# 2. Define the metrics you want to use
11metrics = [
12    faithfulness,
13    answer_relevancy,
14    context_precision,
15    context_recall
16]
17
18# 3. Run the evaluation
19# Note: Ensure your OPENAI_API_KEY is set in your environment variables
20# as Ragas uses OpenAI models by default for evaluation.
21result = evaluate(
22    dataset["baseline"],
23    metrics=metrics,
24)
25
26# 4. Export and view the results
27df = result.to_pandas()
28print(df.head())