Implementation:Explodinggradients Ragas Griptape Integration

Metadata	Value
Source	src/ragas/integrations/griptape.py (Lines 13-61)
Domains	Integration, Griptape
Last Updated	2026-02-10

Overview

Converts Griptape RAG context objects into a Ragas EvaluationDataset, enabling evaluation of Griptape RAG pipelines with Ragas metrics.

Description

The transform_to_ragas_dataset function takes a list of Griptape RagContext objects and optional reference data, then constructs a Ragas EvaluationDataset suitable for metric evaluation. For each RagContext:

• The user_input is extracted from RagContext.query.
• The retrieved_contexts are built by calling to_text() on each text chunk in RagContext.text_chunks.
• The response is produced by joining the text representations of all outputs in RagContext.outputs with newlines.

The function validates that all provided non-None lists have the same length, raising a ValueError on length mismatches. Optional parameters (reference_contexts, references, rubrics) are mapped per-sample when provided or set to None otherwise.

Usage

Use this function when you have Griptape RAG pipeline results and want to evaluate them using Ragas metrics such as faithfulness, context precision, or answer correctness. It bridges the gap between Griptape's data structures and Ragas' evaluation API.

Code Reference

Source Location

Item	Detail
File	src/ragas/integrations/griptape.py
Lines	13-61
Module	ragas.integrations.griptape

Signature

def transform_to_ragas_dataset(
    grip_tape_rag_contexts: List[RagContext],
    reference_contexts: Optional[List[str]] = None,
    references: Optional[List[str]] = None,
    rubrics: Optional[List[Dict[str, str]]] = None,
) -> EvaluationDataset

Import

from ragas.integrations.griptape import transform_to_ragas_dataset

I/O Contract

Inputs

Name	Type	Required	Description
grip_tape_rag_contexts	List[RagContext]	Yes	List of Griptape RAG context objects containing queries, text chunks, and outputs
reference_contexts	Optional[List[str]]	No	Ground-truth reference contexts for each sample
references	Optional[List[str]]	No	Ground-truth reference answers for each sample
rubrics	Optional[List[Dict[str, str]]]	No	Evaluation rubrics for each sample

Outputs

Name	Type	Description
(return)	EvaluationDataset	A Ragas evaluation dataset with samples containing user_input, retrieved_contexts, response, and optional reference fields

Exceptions

Exception	Condition
ValueError	When provided lists have inconsistent lengths
ImportError	When the griptape package is not installed

Usage Examples

Basic Conversion from Griptape RAG Contexts

from ragas.integrations.griptape import transform_to_ragas_dataset

# Assuming you have Griptape RagContext objects from a RAG pipeline
# rag_contexts = [context1, context2, context3]

dataset = transform_to_ragas_dataset(
    grip_tape_rag_contexts=rag_contexts,
)

# The dataset can now be evaluated with Ragas metrics
print(len(dataset.samples))

Conversion with Reference Data

from ragas.integrations.griptape import transform_to_ragas_dataset

# With ground-truth references for metrics like answer correctness
dataset = transform_to_ragas_dataset(
    grip_tape_rag_contexts=rag_contexts,
    references=["Expected answer 1", "Expected answer 2"],
    reference_contexts=["Ground truth context 1", "Ground truth context 2"],
)

# Evaluate with Ragas
from ragas import evaluate
from ragas.metrics import faithfulness, answer_correctness

results = evaluate(dataset=dataset, metrics=[faithfulness, answer_correctness])

Related Pages

• R2R Integration - Similar transform function for R2R response objects
• Amazon Bedrock Integration - Another integration that extracts RAG contexts from agent traces
• LlamaIndex Integration - Evaluates LlamaIndex query engines using the same EvaluationDataset format