Metrics

MetricType

Bases: Enum

Enumeration of metric types in Ragas.

Attributes:

Name	Type	Description
SINGLE_TURN	str	Represents a single-turn metric type.
MULTI_TURN	str	Represents a multi-turn metric type.

Metric dataclass

Metric(_required_columns: Dict[MetricType, Set[str]] = dict(), name: str = '')

Bases: ABC

Abstract base class for metrics in Ragas.

Attributes:

Name	Type	Description
name	str	The name of the metric.
required_columns	Dict[str, Set[str]]	A dictionary mapping metric type names to sets of required column names. This is a property and raises ValueError if columns are not in VALID_COLUMNS.

score

score(row: Dict, callbacks: Callbacks = None) -> float

Calculates the score for a single row of data.

Note

This method is deprecated and will be removed in 0.3. Please use single_turn_ascore or multi_turn_ascore instead.

Source code in src/ragas/metrics/base.py

@deprecated("0.2", removal="0.3", alternative="single_turn_ascore")
def score(self, row: t.Dict, callbacks: Callbacks = None) -> float:
    """
    Calculates the score for a single row of data.

    Note
    ----
    This method is deprecated and will be removed in 0.3. Please use `single_turn_ascore` or `multi_turn_ascore` instead.
    """
    callbacks = callbacks or []
    rm, group_cm = new_group(
        self.name,
        inputs=row,
        callbacks=callbacks,
        metadata={"type": ChainType.METRIC},
    )
    try:
        if is_event_loop_running():
            try:
                import nest_asyncio

                nest_asyncio.apply()
            except ImportError:
                raise ImportError(
                    "It seems like your running this in a jupyter-like environment. Please install nest_asyncio with `pip install nest_asyncio` to make it work."
                )
        loop = asyncio.get_event_loop()
        score = loop.run_until_complete(self._ascore(row=row, callbacks=group_cm))
    except Exception as e:
        if not group_cm.ended:
            rm.on_chain_error(e)
        raise e
    else:
        if not group_cm.ended:
            rm.on_chain_end({"output": score})
    return score

ascore async

ascore(row: Dict, callbacks: Callbacks = None, timeout: Optional[float] = None) -> float

Asynchronously calculates the score for a single row of data.

Note

This method is deprecated and will be removed in 0.3. Please use single_turn_ascore instead.

Source code in src/ragas/metrics/base.py

@deprecated("0.2", removal="0.3", alternative="single_turn_ascore")
async def ascore(
    self,
    row: t.Dict,
    callbacks: Callbacks = None,
    timeout: t.Optional[float] = None,
) -> float:
    """
    Asynchronously calculates the score for a single row of data.

    Note
    ----
    This method is deprecated and will be removed in 0.3. Please use `single_turn_ascore` instead.
    """
    callbacks = callbacks or []
    rm, group_cm = new_group(
        self.name,
        inputs=row,
        callbacks=callbacks,
        metadata={"type": ChainType.METRIC},
    )
    try:
        score = await asyncio.wait_for(
            self._ascore(row=row, callbacks=group_cm),
            timeout=timeout,
        )
    except Exception as e:
        if not group_cm.ended:
            rm.on_chain_error(e)
        raise e
    else:
        if not group_cm.ended:
            rm.on_chain_end({"output": score})
    return score

MetricWithLLM dataclass

MetricWithLLM(_required_columns: Dict[MetricType, Set[str]] = dict(), name: str = '', llm: Optional[BaseRagasLLM] = None, output_type: Optional[MetricOutputType] = None)

Bases: Metric, PromptMixin

A metric class that uses a language model for evaluation.

Attributes:

Name	Type	Description
llm	Optional[BaseRagasLLM]	The language model used for the metric.

train

train(path: str, demonstration_config: Optional[DemonstrationConfig] = None, instruction_config: Optional[InstructionConfig] = None, callbacks: Optional[Callbacks] = None, run_config: Optional[RunConfig] = None, batch_size: Optional[int] = None, with_debugging_logs=False, raise_exceptions: bool = True) -> None

Train the metric using local JSON data

Parameters:

Name	Type	Description	Default
path	str	Path to local JSON training data file	required
demonstration_config	DemonstrationConfig	Configuration for demonstration optimization	None
instruction_config	InstructionConfig	Configuration for instruction optimization	None
callbacks	Callbacks	List of callback functions	None
run_config	RunConfig	Run configuration	None
batch_size	int	Batch size for training	None
with_debugging_logs	bool	Enable debugging logs	False
raise_exceptions	bool	Whether to raise exceptions during training	True

Raises:

Type	Description
ValueError	If path is not provided or not a JSON file

Source code in src/ragas/metrics/base.py

def train(
    self,
    path: str,
    demonstration_config: t.Optional[DemonstrationConfig] = None,
    instruction_config: t.Optional[InstructionConfig] = None,
    callbacks: t.Optional[Callbacks] = None,
    run_config: t.Optional[RunConfig] = None,
    batch_size: t.Optional[int] = None,
    with_debugging_logs=False,
    raise_exceptions: bool = True,
) -> None:
    """
    Train the metric using local JSON data

    Parameters
    ----------
    path : str
        Path to local JSON training data file
    demonstration_config : DemonstrationConfig, optional
        Configuration for demonstration optimization
    instruction_config : InstructionConfig, optional
        Configuration for instruction optimization
    callbacks : Callbacks, optional
        List of callback functions
    run_config : RunConfig, optional
        Run configuration
    batch_size : int, optional
        Batch size for training
    with_debugging_logs : bool, default=False
        Enable debugging logs
    raise_exceptions : bool, default=True
        Whether to raise exceptions during training

    Raises
    ------
    ValueError
        If path is not provided or not a JSON file
    """
    # Validate input parameters
    if not path:
        raise ValueError("Path to training data file must be provided")

    if not path.endswith(".json"):
        raise ValueError("Train data must be in json format")

    run_config = run_config or RunConfig()
    callbacks = callbacks or []

    # Load the dataset from JSON file
    dataset = MetricAnnotation.from_json(path, metric_name=self.name)

    # only optimize the instruction if instruction_config is provided
    if instruction_config is not None:
        self._optimize_instruction(
            instruction_config=instruction_config,
            dataset=dataset,
            callbacks=callbacks,
            run_config=run_config,
            batch_size=batch_size,
            with_debugging_logs=with_debugging_logs,
            raise_exceptions=raise_exceptions,
        )

    # if demonstration_config is provided, optimize the demonstrations
    if demonstration_config is not None:
        self._optimize_demonstration(
            demonstration_config=demonstration_config,
            dataset=dataset,
        )

SingleTurnMetric dataclass

SingleTurnMetric(_required_columns: Dict[MetricType, Set[str]] = dict(), name: str = '')

Bases: Metric

A metric class for evaluating single-turn interactions.

This class provides methods to score single-turn samples, both synchronously and asynchronously.

single_turn_score

single_turn_score(sample: SingleTurnSample, callbacks: Callbacks = None) -> float

Synchronously score a single-turn sample.

May raise ImportError if nest_asyncio is not installed in a Jupyter-like environment.

Source code in src/ragas/metrics/base.py

def single_turn_score(
    self,
    sample: SingleTurnSample,
    callbacks: Callbacks = None,
) -> float:
    """
    Synchronously score a single-turn sample.

    May raise ImportError if nest_asyncio is not installed in a Jupyter-like environment.
    """
    callbacks = callbacks or []
    # only get the required columns
    sample = self._only_required_columns_single_turn(sample)
    rm, group_cm = new_group(
        self.name,
        inputs=sample.to_dict(),
        callbacks=callbacks,
        metadata={"type": ChainType.METRIC},
    )
    try:
        if is_event_loop_running():
            try:
                import nest_asyncio

                nest_asyncio.apply()
            except ImportError:
                raise ImportError(
                    "It seems like your running this in a jupyter-like environment. Please install nest_asyncio with `pip install nest_asyncio` to make it work."
                )
        loop = asyncio.get_event_loop()
        score = loop.run_until_complete(
            self._single_turn_ascore(sample=sample, callbacks=group_cm)
        )
    except Exception as e:
        if not group_cm.ended:
            rm.on_chain_error(e)
        raise e
    else:
        if not group_cm.ended:
            rm.on_chain_end({"output": score})

    # track the evaluation event
    _analytics_batcher.add_evaluation(
        EvaluationEvent(
            metrics=[self.name],
            num_rows=1,
            evaluation_type=MetricType.SINGLE_TURN.name,
            language=get_metric_language(self),
        )
    )
    return score

single_turn_ascore async

single_turn_ascore(sample: SingleTurnSample, callbacks: Callbacks = None, timeout: Optional[float] = None) -> float

Asynchronously score a single-turn sample with an optional timeout.

May raise asyncio.TimeoutError if the scoring process exceeds the specified timeout.

Source code in src/ragas/metrics/base.py

async def single_turn_ascore(
    self,
    sample: SingleTurnSample,
    callbacks: Callbacks = None,
    timeout: t.Optional[float] = None,
) -> float:
    """
    Asynchronously score a single-turn sample with an optional timeout.

    May raise asyncio.TimeoutError if the scoring process exceeds the specified timeout.
    """
    callbacks = callbacks or []
    # only get the required columns
    sample = self._only_required_columns_single_turn(sample)
    rm, group_cm = new_group(
        self.name,
        inputs=sample.to_dict(),
        callbacks=callbacks,
        metadata={"type": ChainType.METRIC},
    )
    try:
        score = await asyncio.wait_for(
            self._single_turn_ascore(sample=sample, callbacks=group_cm),
            timeout=timeout,
        )
    except Exception as e:
        if not group_cm.ended:
            rm.on_chain_error(e)
        raise e
    else:
        if not group_cm.ended:
            rm.on_chain_end({"output": score})

    # track the evaluation event
    _analytics_batcher.add_evaluation(
        EvaluationEvent(
            metrics=[self.name],
            num_rows=1,
            evaluation_type=MetricType.SINGLE_TURN.name,
            language=get_metric_language(self),
        )
    )
    return score

MultiTurnMetric dataclass

MultiTurnMetric(_required_columns: Dict[MetricType, Set[str]] = dict(), name: str = '')

Bases: Metric

A metric class for evaluating multi-turn conversations.

This class extends the base Metric class to provide functionality for scoring multi-turn conversation samples.

multi_turn_score

multi_turn_score(sample: MultiTurnSample, callbacks: Callbacks = None) -> float

Score a multi-turn conversation sample synchronously.

May raise ImportError if nest_asyncio is not installed in Jupyter-like environments.

Source code in src/ragas/metrics/base.py

def multi_turn_score(
    self,
    sample: MultiTurnSample,
    callbacks: Callbacks = None,
) -> float:
    """
    Score a multi-turn conversation sample synchronously.

    May raise ImportError if nest_asyncio is not installed in Jupyter-like environments.
    """
    callbacks = callbacks or []
    sample = self._only_required_columns_multi_turn(sample)
    rm, group_cm = new_group(
        self.name,
        inputs=sample.to_dict(),
        callbacks=callbacks,
        metadata={"type": ChainType.METRIC},
    )
    try:
        if is_event_loop_running():
            try:
                import nest_asyncio

                nest_asyncio.apply()
            except ImportError:
                raise ImportError(
                    "It seems like your running this in a jupyter-like environment. Please install nest_asyncio with `pip install nest_asyncio` to make it work."
                )
        loop = asyncio.get_event_loop()
        score = loop.run_until_complete(
            self._multi_turn_ascore(sample=sample, callbacks=group_cm)
        )
    except Exception as e:
        if not group_cm.ended:
            rm.on_chain_error(e)
        raise e
    else:
        if not group_cm.ended:
            rm.on_chain_end({"output": score})

    # track the evaluation event
    _analytics_batcher.add_evaluation(
        EvaluationEvent(
            metrics=[self.name],
            num_rows=1,
            evaluation_type=MetricType.SINGLE_TURN.name,
            language=get_metric_language(self),
        )
    )
    return score

multi_turn_ascore async

multi_turn_ascore(sample: MultiTurnSample, callbacks: Callbacks = None, timeout: Optional[float] = None) -> float

Score a multi-turn conversation sample asynchronously.

May raise asyncio.TimeoutError if the scoring process exceeds the specified timeout.

Source code in src/ragas/metrics/base.py

async def multi_turn_ascore(
    self,
    sample: MultiTurnSample,
    callbacks: Callbacks = None,
    timeout: t.Optional[float] = None,
) -> float:
    """
    Score a multi-turn conversation sample asynchronously.

    May raise asyncio.TimeoutError if the scoring process exceeds the specified timeout.
    """
    callbacks = callbacks or []
    sample = self._only_required_columns_multi_turn(sample)

    rm, group_cm = new_group(
        self.name,
        inputs=sample.to_dict(),
        callbacks=callbacks,
        metadata={"type": ChainType.METRIC},
    )
    try:
        score = await asyncio.wait_for(
            self._multi_turn_ascore(sample=sample, callbacks=group_cm),
            timeout=timeout,
        )
    except Exception as e:
        if not group_cm.ended:
            rm.on_chain_error(e)
        raise e
    else:
        if not group_cm.ended:
            rm.on_chain_end({"output": score})

    # track the evaluation event
    _analytics_batcher.add_evaluation(
        EvaluationEvent(
            metrics=[self.name],
            num_rows=1,
            evaluation_type=MetricType.SINGLE_TURN.name,
            language=get_metric_language(self),
        )
    )

    return score

Ensember

Combine multiple llm outputs for same input (n>1) to a single output

from_discrete

from_discrete(inputs: list[list[Dict]], attribute: str) -> List[Dict]

Simple majority voting for binary values, ie [0,0,1] -> 0 inputs: list of list of dicts each containing verdict for a single input

Source code in src/ragas/metrics/base.py

def from_discrete(
    self, inputs: list[list[t.Dict]], attribute: str
) -> t.List[t.Dict]:
    """
    Simple majority voting for binary values, ie [0,0,1] -> 0
    inputs: list of list of dicts each containing verdict for a single input
    """

    if not isinstance(inputs, list):
        inputs = [inputs]

    if not all(len(item) == len(inputs[0]) for item in inputs):
        logger.warning("All inputs must have the same length")
        return inputs[0]

    if not all(attribute in item for input in inputs for item in input):
        logger.warning(f"All inputs must have {attribute} attribute")
        return inputs[0]

    if len(inputs) == 1:
        return inputs[0]

    verdict_agg = []
    for i in range(len(inputs[0])):
        item = inputs[0][i]
        verdicts = [inputs[k][i][attribute] for k in range(len(inputs))]
        verdict_counts = dict(Counter(verdicts).most_common())
        item[attribute] = list(verdict_counts.keys())[0]
        verdict_agg.append(item)

    return verdict_agg

AnswerCorrectness dataclass

AnswerCorrectness(_required_columns: Dict[MetricType, Set[str]] = (lambda: {SINGLE_TURN: {'user_input', 'response', 'reference'}})(), name: str = 'answer_correctness', embeddings: Optional[Union[BaseRagasEmbeddings, BaseRagasEmbedding]] = None, llm: Optional[BaseRagasLLM] = None, output_type: Optional[MetricOutputType] = None, correctness_prompt: PydanticPrompt = CorrectnessClassifier(), statement_generator_prompt: PydanticPrompt = StatementGeneratorPrompt(), weights: list[float] = (lambda: [0.75, 0.25])(), beta: float = 1.0, answer_similarity: Optional[AnswerSimilarity] = None, max_retries: int = 1)

Bases: MetricWithLLM, MetricWithEmbeddings, SingleTurnMetric

Measures answer correctness compared to ground truth as a combination of factuality and semantic similarity.

Attributes:

Name	Type	Description
name	string	The name of the metrics
weights	list[float]	a list of two weights corresponding to factuality and semantic similarity Defaults [0.75, 0.25]
answer_similarity	Optional[AnswerSimilarity]	The AnswerSimilarity object

ResponseRelevancy dataclass

ResponseRelevancy(_required_columns: Dict[MetricType, Set[str]] = (lambda: {SINGLE_TURN: {'user_input', 'response'}})(), name: str = 'answer_relevancy', embeddings: Optional[Union[BaseRagasEmbeddings, BaseRagasEmbedding]] = None, llm: Optional[BaseRagasLLM] = None, output_type: Optional[MetricOutputType] = None, question_generation: PydanticPrompt = ResponseRelevancePrompt(), strictness: int = 3)

Bases: MetricWithLLM, MetricWithEmbeddings, SingleTurnMetric

Scores the relevancy of the answer according to the given question. Answers with incomplete, redundant or unnecessary information is penalized. Score can range from 0 to 1 with 1 being the best.

Attributes:

Name	Type	Description
name	string	The name of the metrics
strictness	int	Here indicates the number questions generated per answer. Ideal range between 3 to 5.
embeddings	Embedding	The langchain wrapper of Embedding object. E.g. HuggingFaceEmbeddings('BAAI/bge-base-en')

SemanticSimilarity dataclass

SemanticSimilarity(_required_columns: Dict[MetricType, Set[str]] = (lambda: {SINGLE_TURN: {'reference', 'response'}})(), name: str = 'semantic_similarity', embeddings: Optional[Union[BaseRagasEmbeddings, BaseRagasEmbedding]] = None, is_cross_encoder: bool = False, threshold: Optional[float] = None)

Bases: MetricWithEmbeddings, SingleTurnMetric

Scores the semantic similarity of ground truth with generated answer. cross encoder score is used to quantify semantic similarity. SAS paper: https://arxiv.org/pdf/2108.06130.pdf

Attributes:

Name	Type	Description
name	str
model_name		The model to be used for calculating semantic similarity Defaults open-ai-embeddings select cross-encoder model for best results https://huggingface.co/spaces/mteb/leaderboard
threshold	Optional[float]	The threshold if given used to map output to binary Default 0.5

AspectCritic

AspectCritic(name: str, definition: str, llm: Optional[BaseRagasLLM] = None, required_columns: Optional[Dict[MetricType, Set[str]]] = None, output_type: Optional[MetricOutputType] = BINARY, single_turn_prompt: Optional[PydanticPrompt] = None, multi_turn_prompt: Optional[PydanticPrompt] = None, strictness: int = 1, max_retries: int = 1)

Bases: MetricWithLLM, SingleTurnMetric, MultiTurnMetric

Judges the submission to give binary results using the criteria specified in the metric definition.

Attributes:

Name	Type	Description
name	str	name of the metrics
definition	str	criteria to judge the submission, example "Is the submission spreading fake information?"
strictness	int	The number of times self consistency checks is made. Final judgement is made using majority vote.

Source code in src/ragas/metrics/_aspect_critic.py

def __init__(
    self,
    name: str,
    definition: str,
    llm: t.Optional[BaseRagasLLM] = None,
    required_columns: t.Optional[t.Dict[MetricType, t.Set[str]]] = None,
    output_type: t.Optional[MetricOutputType] = MetricOutputType.BINARY,
    single_turn_prompt: t.Optional[PydanticPrompt] = None,
    multi_turn_prompt: t.Optional[PydanticPrompt] = None,
    strictness: int = 1,
    max_retries: int = 1,
):
    self._required_columns = required_columns or {
        MetricType.SINGLE_TURN: {
            "user_input:optional",
            "response:optional",
            "retrieved_contexts:optional",
            "reference:optional",
            "reference_contexts:optional",
        },
        MetricType.MULTI_TURN: {
            "user_input:optional",
            "reference:optional",
        },
    }
    super().__init__(
        name=name,
        _required_columns=self._required_columns,
        llm=llm,
        output_type=output_type,
    )

    self._definition = definition
    self.single_turn_prompt = single_turn_prompt or SingleTurnAspectCriticPrompt()
    self.multi_turn_prompt = multi_turn_prompt or MultiTurnAspectCriticPrompt()
    self.max_retries = max_retries

    # update the instruction for the prompts with the definition
    instruction = f"Evaluate the Input based on the criterial defined. Use only 'Yes' (1) and 'No' (0) as verdict.\nCriteria Definition: {self._definition}"
    self.single_turn_prompt.instruction = instruction
    self.multi_turn_prompt.instruction = instruction

    # ensure odd number of checks to avoid tie in majority vote.
    self.strictness = strictness
    self.strictness = (
        self.strictness if self.strictness % 2 != 0 else self.strictness + 1
    )

ContextEntityRecall dataclass

ContextEntityRecall(_required_columns: Dict[MetricType, Set[str]] = (lambda: {SINGLE_TURN: {'reference', 'retrieved_contexts'}})(), name: str = 'context_entity_recall', llm: Optional[BaseRagasLLM] = None, output_type: Optional[MetricOutputType] = None, context_entity_recall_prompt: PydanticPrompt = ExtractEntitiesPrompt(), max_retries: int = 1)

Bases: MetricWithLLM, SingleTurnMetric

Calculates recall based on entities present in ground truth and context. Let CN be the set of entities present in context, GN be the set of entities present in the ground truth.

Then we define can the context entity recall as follows: Context Entity recall = | CN ∩ GN | / | GN |

If this quantity is 1, we can say that the retrieval mechanism has retrieved context which covers all entities present in the ground truth, thus being a useful retrieval. Thus this can be used to evaluate retrieval mechanisms in specific use cases where entities matter, for example, a tourism help chatbot.

Attributes:

Name	Type	Description
name	str
batch_size	int	Batch size for openai completion.

LLMContextPrecisionWithReference dataclass

LLMContextPrecisionWithReference(_required_columns: Dict[MetricType, Set[str]] = (lambda: {SINGLE_TURN: {'user_input', 'retrieved_contexts', 'reference'}})(), name: str = 'llm_context_precision_with_reference', llm: Optional[BaseRagasLLM] = None, output_type: Optional[MetricOutputType] = None, context_precision_prompt: PydanticPrompt = ContextPrecisionPrompt(), max_retries: int = 1)

Bases: MetricWithLLM, SingleTurnMetric

Average Precision is a metric that evaluates whether all of the relevant items selected by the model are ranked higher or not.

Attributes:

Name	Type	Description
name	str
evaluation_mode	EvaluationMode
context_precision_prompt	Prompt

LLMContextRecall dataclass

LLMContextRecall(_required_columns: Dict[MetricType, Set[str]] = (lambda: {SINGLE_TURN: {'user_input', 'retrieved_contexts', 'reference'}})(), name: str = 'context_recall', llm: Optional[BaseRagasLLM] = None, output_type: Optional[MetricOutputType] = CONTINUOUS, context_recall_prompt: PydanticPrompt = ContextRecallClassificationPrompt(), max_retries: int = 1)

Bases: MetricWithLLM, SingleTurnMetric

Estimates context recall by estimating TP and FN using annotated answer and retrieved context.

Attributes:

Name	Type	Description
name	str

FactualCorrectness dataclass

FactualCorrectness(_required_columns: Dict[MetricType, Set[str]] = (lambda: {SINGLE_TURN: {'response', 'reference'}})(), name: str = 'factual_correctness', llm: Optional[BaseRagasLLM] = None, output_type: Optional[MetricOutputType] = CONTINUOUS, mode: Literal['precision', 'recall', 'f1'] = 'f1', beta: float = 1.0, atomicity: Literal['low', 'high'] = 'low', coverage: Literal['low', 'high'] = 'low', claim_decomposition_prompt: PydanticPrompt = ClaimDecompositionPrompt(), nli_prompt: PydanticPrompt = NLIStatementPrompt(), language: str = 'english')

Bases: MetricWithLLM, SingleTurnMetric

FactualCorrectness is a metric class that evaluates the factual correctness of responses generated by a language model. It uses claim decomposition and natural language inference (NLI) to verify the claims made in the responses against reference texts.

Attributes: name (str): The name of the metric, default is "factual_correctness". _required_columns (Dict[MetricType, Set[str]]): A dictionary specifying the required columns for each metric type. Default is {"SINGLE_TURN": {"response", "reference"}}. mode (Literal["precision", "recall", "f1"]): The mode of evaluation, can be "precision", "recall", or "f1". Default is "f1". beta (float): The beta value used for the F1 score calculation. A beta > 1 gives more weight to recall, while beta < 1 favors precision. Default is 1.0. atomicity (Literal["low", "high"]): The level of atomicity for claim decomposition. Default is "low". coverage (Literal["low", "high"]): The level of coverage for claim decomposition. Default is "low". claim_decomposition_prompt (PydanticPrompt): The prompt used for claim decomposition. nli_prompt (PydanticPrompt): The prompt used for natural language inference (NLI).

Faithfulness dataclass

Faithfulness(_required_columns: Dict[MetricType, Set[str]] = (lambda: {SINGLE_TURN: {'user_input', 'response', 'retrieved_contexts'}})(), name: str = 'faithfulness', llm: Optional[BaseRagasLLM] = None, output_type: Optional[MetricOutputType] = CONTINUOUS, nli_statements_prompt: PydanticPrompt = NLIStatementPrompt(), statement_generator_prompt: PydanticPrompt = StatementGeneratorPrompt(), max_retries: int = 1)

Bases: MetricWithLLM, SingleTurnMetric

FaithfulnesswithHHEM dataclass

FaithfulnesswithHHEM(_required_columns: Dict[MetricType, Set[str]] = (lambda: {SINGLE_TURN: {'user_input', 'response', 'retrieved_contexts'}})(), name: str = 'faithfulness_with_hhem', llm: Optional[BaseRagasLLM] = None, output_type: Optional[MetricOutputType] = CONTINUOUS, nli_statements_prompt: PydanticPrompt = NLIStatementPrompt(), statement_generator_prompt: PydanticPrompt = StatementGeneratorPrompt(), max_retries: int = 1, device: str = 'cpu', batch_size: int = 10)

Bases: Faithfulness

NoiseSensitivity dataclass

NoiseSensitivity(_required_columns: Dict[MetricType, Set[str]] = (lambda: {SINGLE_TURN: {'user_input', 'response', 'reference', 'retrieved_contexts'}})(), name: str = 'noise_sensitivity', llm: Optional[BaseRagasLLM] = None, output_type: Optional[MetricOutputType] = CONTINUOUS, mode: Literal['relevant', 'irrelevant'] = 'relevant', nli_statements_prompt: PydanticPrompt = NLIStatementPrompt(), statement_generator_prompt: PydanticPrompt = StatementGeneratorPrompt(), max_retries: int = 1)

Bases: MetricWithLLM, SingleTurnMetric

AnswerAccuracy dataclass

AnswerAccuracy(_required_columns: Dict[MetricType, Set[str]] = (lambda: {SINGLE_TURN: {'user_input', 'response', 'reference'}})(), name: str = 'nv_accuracy', llm: Optional[BaseRagasLLM] = None, output_type: Optional[MetricOutputType] = None)

Bases: MetricWithLLM, SingleTurnMetric

Measures answer accuracy compared to ground truth given a user_input. This metric averages two distinct judge prompts to evaluate.

Top10, Zero-shoot LLM-as-a-Judge Leaderboard: 1)- mistralai/mixtral-8x22b-instruct-v0.1 2)- mistralai/mixtral-8x7b-instruct-v0.1 3)- meta/llama-3.1-70b-instruct 4)- meta/llama-3.3-70b-instruct 5)- meta/llama-3.1-405b-instruct 6)- mistralai/mistral-nemo-12b-instruct 7)- nvidia/llama-3.1-nemotron-70b-instruct 8)- meta/llama-3.1-8b-instruct 9)- google/gemma-2-2b-it 10)- nvidia/nemotron-mini-4b-instruct The top1 LB model have high correlation with human judges (~0.90).

Attributes:

Name	Type	Description
name	string	The name of the metrics
answer_accuracy		The AnswerAccuracy object

ContextRelevance dataclass

ContextRelevance(_required_columns: Dict[MetricType, Set[str]] = (lambda: {SINGLE_TURN: {'user_input', 'retrieved_contexts'}})(), name: str = 'nv_context_relevance', llm: Optional[BaseRagasLLM] = None, output_type: Optional[MetricOutputType] = None)

Bases: MetricWithLLM, SingleTurnMetric

Parameters: Score the relevance of the retrieved contexts be based on the user input.

Input: data: list of Dicts with keys: user_input, retrieved_contexts Output: 0.0: retrieved_contexts is not relevant for the user_input 0.5: retrieved_contexts is partially relevant for the user_input 1.0: retrieved_contexts is fully relevant for the user_input

ResponseGroundedness dataclass

ResponseGroundedness(_required_columns: Dict[MetricType, Set[str]] = (lambda: {SINGLE_TURN: {'response', 'retrieved_contexts'}})(), name: str = 'nv_response_groundedness', llm: Optional[BaseRagasLLM] = None, output_type: Optional[MetricOutputType] = None)

Bases: MetricWithLLM, SingleTurnMetric

Parameters: Score the groundedness of the response based on the retrieved contexts.

Input: data: list of Dicts with keys: response, retrieved contexts Output: 0.0: response is not grounded in the retrieved contexts 0.5: response is partially grounded in the retrieved contexts 1.0: response is fully grounded in the retrieved contexts

SimpleCriteriaScore

SimpleCriteriaScore(name: str, definition: str, llm: Optional[BaseRagasLLM] = None, required_columns: Optional[Dict[MetricType, Set[str]]] = None, output_type: Optional[MetricOutputType] = DISCRETE, single_turn_prompt: Optional[PydanticPrompt] = None, multi_turn_prompt: Optional[PydanticPrompt] = None, strictness: int = 1)

Bases: MetricWithLLM, SingleTurnMetric, MultiTurnMetric

Judges the submission to give binary results using the criteria specified in the metric definition.

Attributes:

Name	Type	Description
name	str	name of the metrics
definition	str	criteria to score the submission
strictness	int	The number of times self consistency checks is made. Final judgement is made using majority vote.

Source code in src/ragas/metrics/_simple_criteria.py

def __init__(
    self,
    name: str,
    definition: str,
    llm: t.Optional[BaseRagasLLM] = None,
    required_columns: t.Optional[t.Dict[MetricType, t.Set[str]]] = None,
    output_type: t.Optional[MetricOutputType] = MetricOutputType.DISCRETE,
    single_turn_prompt: t.Optional[PydanticPrompt] = None,
    multi_turn_prompt: t.Optional[PydanticPrompt] = None,
    strictness: int = 1,
):
    if required_columns is None:
        required_columns = {
            MetricType.SINGLE_TURN: {
                "user_input:optional",
                "response:optional",
                "retrieved_contexts:optional",
                "reference:optional",
                "reference_contexts:optional",
            },
            MetricType.MULTI_TURN: {
                "user_input:optional",
                "reference:optional",
            },
        }
    super().__init__(
        name=name,
        llm=llm,
        _required_columns=required_columns,
        output_type=output_type,
    )

    self._definition = definition
    self.single_turn_prompt = single_turn_prompt or SingleTurnSimpleCriteriaPrompt()
    self.multi_turn_prompt = multi_turn_prompt or MultiTurnSimpleCriteriaPrompt()

    # update the instruction for the prompts with the definition
    instruction = f"Evaluate the input based on the criteria defined.\nCriteria Definition: {self._definition}"
    self.single_turn_prompt.instruction = instruction
    self.multi_turn_prompt.instruction = instruction

    # ensure odd number of checks to avoid tie in majority vote.
    self.strictness = strictness
    self.strictness = (
        self.strictness if self.strictness % 2 != 0 else self.strictness + 1
    )

Metric dataclass

Metric(_required_columns: Dict[MetricType, Set[str]] = dict(), name: str = '')

Bases: ABC

Abstract base class for metrics in Ragas.

Attributes:

Name	Type	Description
name	str	The name of the metric.
required_columns	Dict[str, Set[str]]	A dictionary mapping metric type names to sets of required column names. This is a property and raises ValueError if columns are not in VALID_COLUMNS.

score

score(row: Dict, callbacks: Callbacks = None) -> float

Calculates the score for a single row of data.

Note

This method is deprecated and will be removed in 0.3. Please use single_turn_ascore or multi_turn_ascore instead.

Source code in src/ragas/metrics/base.py

@deprecated("0.2", removal="0.3", alternative="single_turn_ascore")
def score(self, row: t.Dict, callbacks: Callbacks = None) -> float:
    """
    Calculates the score for a single row of data.

    Note
    ----
    This method is deprecated and will be removed in 0.3. Please use `single_turn_ascore` or `multi_turn_ascore` instead.
    """
    callbacks = callbacks or []
    rm, group_cm = new_group(
        self.name,
        inputs=row,
        callbacks=callbacks,
        metadata={"type": ChainType.METRIC},
    )
    try:
        if is_event_loop_running():
            try:
                import nest_asyncio

                nest_asyncio.apply()
            except ImportError:
                raise ImportError(
                    "It seems like your running this in a jupyter-like environment. Please install nest_asyncio with `pip install nest_asyncio` to make it work."
                )
        loop = asyncio.get_event_loop()
        score = loop.run_until_complete(self._ascore(row=row, callbacks=group_cm))
    except Exception as e:
        if not group_cm.ended:
            rm.on_chain_error(e)
        raise e
    else:
        if not group_cm.ended:
            rm.on_chain_end({"output": score})
    return score

ascore async

ascore(row: Dict, callbacks: Callbacks = None, timeout: Optional[float] = None) -> float

Asynchronously calculates the score for a single row of data.

Note

This method is deprecated and will be removed in 0.3. Please use single_turn_ascore instead.

Source code in src/ragas/metrics/base.py

@deprecated("0.2", removal="0.3", alternative="single_turn_ascore")
async def ascore(
    self,
    row: t.Dict,
    callbacks: Callbacks = None,
    timeout: t.Optional[float] = None,
) -> float:
    """
    Asynchronously calculates the score for a single row of data.

    Note
    ----
    This method is deprecated and will be removed in 0.3. Please use `single_turn_ascore` instead.
    """
    callbacks = callbacks or []
    rm, group_cm = new_group(
        self.name,
        inputs=row,
        callbacks=callbacks,
        metadata={"type": ChainType.METRIC},
    )
    try:
        score = await asyncio.wait_for(
            self._ascore(row=row, callbacks=group_cm),
            timeout=timeout,
        )
    except Exception as e:
        if not group_cm.ended:
            rm.on_chain_error(e)
        raise e
    else:
        if not group_cm.ended:
            rm.on_chain_end({"output": score})
    return score

MetricType

Bases: Enum

Enumeration of metric types in Ragas.

Attributes:

Name	Type	Description
SINGLE_TURN	str	Represents a single-turn metric type.
MULTI_TURN	str	Represents a multi-turn metric type.

MetricWithLLM dataclass

MetricWithLLM(_required_columns: Dict[MetricType, Set[str]] = dict(), name: str = '', llm: Optional[BaseRagasLLM] = None, output_type: Optional[MetricOutputType] = None)

Bases: Metric, PromptMixin

A metric class that uses a language model for evaluation.

Attributes:

Name	Type	Description
llm	Optional[BaseRagasLLM]	The language model used for the metric.

train

train(path: str, demonstration_config: Optional[DemonstrationConfig] = None, instruction_config: Optional[InstructionConfig] = None, callbacks: Optional[Callbacks] = None, run_config: Optional[RunConfig] = None, batch_size: Optional[int] = None, with_debugging_logs=False, raise_exceptions: bool = True) -> None

Train the metric using local JSON data

Parameters:

Name	Type	Description	Default
path	str	Path to local JSON training data file	required
demonstration_config	DemonstrationConfig	Configuration for demonstration optimization	None
instruction_config	InstructionConfig	Configuration for instruction optimization	None
callbacks	Callbacks	List of callback functions	None
run_config	RunConfig	Run configuration	None
batch_size	int	Batch size for training	None
with_debugging_logs	bool	Enable debugging logs	False
raise_exceptions	bool	Whether to raise exceptions during training	True

Raises:

Type	Description
ValueError	If path is not provided or not a JSON file

Source code in src/ragas/metrics/base.py

def train(
    self,
    path: str,
    demonstration_config: t.Optional[DemonstrationConfig] = None,
    instruction_config: t.Optional[InstructionConfig] = None,
    callbacks: t.Optional[Callbacks] = None,
    run_config: t.Optional[RunConfig] = None,
    batch_size: t.Optional[int] = None,
    with_debugging_logs=False,
    raise_exceptions: bool = True,
) -> None:
    """
    Train the metric using local JSON data

    Parameters
    ----------
    path : str
        Path to local JSON training data file
    demonstration_config : DemonstrationConfig, optional
        Configuration for demonstration optimization
    instruction_config : InstructionConfig, optional
        Configuration for instruction optimization
    callbacks : Callbacks, optional
        List of callback functions
    run_config : RunConfig, optional
        Run configuration
    batch_size : int, optional
        Batch size for training
    with_debugging_logs : bool, default=False
        Enable debugging logs
    raise_exceptions : bool, default=True
        Whether to raise exceptions during training

    Raises
    ------
    ValueError
        If path is not provided or not a JSON file
    """
    # Validate input parameters
    if not path:
        raise ValueError("Path to training data file must be provided")

    if not path.endswith(".json"):
        raise ValueError("Train data must be in json format")

    run_config = run_config or RunConfig()
    callbacks = callbacks or []

    # Load the dataset from JSON file
    dataset = MetricAnnotation.from_json(path, metric_name=self.name)

    # only optimize the instruction if instruction_config is provided
    if instruction_config is not None:
        self._optimize_instruction(
            instruction_config=instruction_config,
            dataset=dataset,
            callbacks=callbacks,
            run_config=run_config,
            batch_size=batch_size,
            with_debugging_logs=with_debugging_logs,
            raise_exceptions=raise_exceptions,
        )

    # if demonstration_config is provided, optimize the demonstrations
    if demonstration_config is not None:
        self._optimize_demonstration(
            demonstration_config=demonstration_config,
            dataset=dataset,
        )

MultiTurnMetric dataclass

MultiTurnMetric(_required_columns: Dict[MetricType, Set[str]] = dict(), name: str = '')

Bases: Metric

A metric class for evaluating multi-turn conversations.

This class extends the base Metric class to provide functionality for scoring multi-turn conversation samples.

multi_turn_score

multi_turn_score(sample: MultiTurnSample, callbacks: Callbacks = None) -> float

Score a multi-turn conversation sample synchronously.

May raise ImportError if nest_asyncio is not installed in Jupyter-like environments.

Source code in src/ragas/metrics/base.py

def multi_turn_score(
    self,
    sample: MultiTurnSample,
    callbacks: Callbacks = None,
) -> float:
    """
    Score a multi-turn conversation sample synchronously.

    May raise ImportError if nest_asyncio is not installed in Jupyter-like environments.
    """
    callbacks = callbacks or []
    sample = self._only_required_columns_multi_turn(sample)
    rm, group_cm = new_group(
        self.name,
        inputs=sample.to_dict(),
        callbacks=callbacks,
        metadata={"type": ChainType.METRIC},
    )
    try:
        if is_event_loop_running():
            try:
                import nest_asyncio

                nest_asyncio.apply()
            except ImportError:
                raise ImportError(
                    "It seems like your running this in a jupyter-like environment. Please install nest_asyncio with `pip install nest_asyncio` to make it work."
                )
        loop = asyncio.get_event_loop()
        score = loop.run_until_complete(
            self._multi_turn_ascore(sample=sample, callbacks=group_cm)
        )
    except Exception as e:
        if not group_cm.ended:
            rm.on_chain_error(e)
        raise e
    else:
        if not group_cm.ended:
            rm.on_chain_end({"output": score})

    # track the evaluation event
    _analytics_batcher.add_evaluation(
        EvaluationEvent(
            metrics=[self.name],
            num_rows=1,
            evaluation_type=MetricType.SINGLE_TURN.name,
            language=get_metric_language(self),
        )
    )
    return score

multi_turn_ascore async

multi_turn_ascore(sample: MultiTurnSample, callbacks: Callbacks = None, timeout: Optional[float] = None) -> float

Score a multi-turn conversation sample asynchronously.

May raise asyncio.TimeoutError if the scoring process exceeds the specified timeout.

Source code in src/ragas/metrics/base.py

async def multi_turn_ascore(
    self,
    sample: MultiTurnSample,
    callbacks: Callbacks = None,
    timeout: t.Optional[float] = None,
) -> float:
    """
    Score a multi-turn conversation sample asynchronously.

    May raise asyncio.TimeoutError if the scoring process exceeds the specified timeout.
    """
    callbacks = callbacks or []
    sample = self._only_required_columns_multi_turn(sample)

    rm, group_cm = new_group(
        self.name,
        inputs=sample.to_dict(),
        callbacks=callbacks,
        metadata={"type": ChainType.METRIC},
    )
    try:
        score = await asyncio.wait_for(
            self._multi_turn_ascore(sample=sample, callbacks=group_cm),
            timeout=timeout,
        )
    except Exception as e:
        if not group_cm.ended:
            rm.on_chain_error(e)
        raise e
    else:
        if not group_cm.ended:
            rm.on_chain_end({"output": score})

    # track the evaluation event
    _analytics_batcher.add_evaluation(
        EvaluationEvent(
            metrics=[self.name],
            num_rows=1,
            evaluation_type=MetricType.SINGLE_TURN.name,
            language=get_metric_language(self),
        )
    )

    return score

SingleTurnMetric dataclass

SingleTurnMetric(_required_columns: Dict[MetricType, Set[str]] = dict(), name: str = '')

Bases: Metric

A metric class for evaluating single-turn interactions.

This class provides methods to score single-turn samples, both synchronously and asynchronously.

single_turn_score

single_turn_score(sample: SingleTurnSample, callbacks: Callbacks = None) -> float

Synchronously score a single-turn sample.

May raise ImportError if nest_asyncio is not installed in a Jupyter-like environment.

Source code in src/ragas/metrics/base.py

def single_turn_score(
    self,
    sample: SingleTurnSample,
    callbacks: Callbacks = None,
) -> float:
    """
    Synchronously score a single-turn sample.

    May raise ImportError if nest_asyncio is not installed in a Jupyter-like environment.
    """
    callbacks = callbacks or []
    # only get the required columns
    sample = self._only_required_columns_single_turn(sample)
    rm, group_cm = new_group(
        self.name,
        inputs=sample.to_dict(),
        callbacks=callbacks,
        metadata={"type": ChainType.METRIC},
    )
    try:
        if is_event_loop_running():
            try:
                import nest_asyncio

                nest_asyncio.apply()
            except ImportError:
                raise ImportError(
                    "It seems like your running this in a jupyter-like environment. Please install nest_asyncio with `pip install nest_asyncio` to make it work."
                )
        loop = asyncio.get_event_loop()
        score = loop.run_until_complete(
            self._single_turn_ascore(sample=sample, callbacks=group_cm)
        )
    except Exception as e:
        if not group_cm.ended:
            rm.on_chain_error(e)
        raise e
    else:
        if not group_cm.ended:
            rm.on_chain_end({"output": score})

    # track the evaluation event
    _analytics_batcher.add_evaluation(
        EvaluationEvent(
            metrics=[self.name],
            num_rows=1,
            evaluation_type=MetricType.SINGLE_TURN.name,
            language=get_metric_language(self),
        )
    )
    return score

single_turn_ascore async

single_turn_ascore(sample: SingleTurnSample, callbacks: Callbacks = None, timeout: Optional[float] = None) -> float

Asynchronously score a single-turn sample with an optional timeout.

May raise asyncio.TimeoutError if the scoring process exceeds the specified timeout.

Source code in src/ragas/metrics/base.py

async def single_turn_ascore(
    self,
    sample: SingleTurnSample,
    callbacks: Callbacks = None,
    timeout: t.Optional[float] = None,
) -> float:
    """
    Asynchronously score a single-turn sample with an optional timeout.

    May raise asyncio.TimeoutError if the scoring process exceeds the specified timeout.
    """
    callbacks = callbacks or []
    # only get the required columns
    sample = self._only_required_columns_single_turn(sample)
    rm, group_cm = new_group(
        self.name,
        inputs=sample.to_dict(),
        callbacks=callbacks,
        metadata={"type": ChainType.METRIC},
    )
    try:
        score = await asyncio.wait_for(
            self._single_turn_ascore(sample=sample, callbacks=group_cm),
            timeout=timeout,
        )
    except Exception as e:
        if not group_cm.ended:
            rm.on_chain_error(e)
        raise e
    else:
        if not group_cm.ended:
            rm.on_chain_end({"output": score})

    # track the evaluation event
    _analytics_batcher.add_evaluation(
        EvaluationEvent(
            metrics=[self.name],
            num_rows=1,
            evaluation_type=MetricType.SINGLE_TURN.name,
            language=get_metric_language(self),
        )
    )
    return score

DiscreteMetric dataclass

DiscreteMetric(name: str, prompt: Optional[Union[str, Prompt]] = None, allowed_values: List[str] = (lambda: ['pass', 'fail'])())

Bases: LLMMetric

get_correlation

get_correlation(gold_labels: List[str], predictions: List[str]) -> float

Calculate the correlation between gold labels and predictions. This is a placeholder method and should be implemented based on the specific metric.

Source code in src/ragas/metrics/discrete.py

def get_correlation(
    self, gold_labels: t.List[str], predictions: t.List[str]
) -> float:
    """
    Calculate the correlation between gold labels and predictions.
    This is a placeholder method and should be implemented based on the specific metric.
    """
    try:
        from sklearn.metrics import cohen_kappa_score
    except ImportError:
        raise ImportError(
            "scikit-learn is required for correlation calculation. "
            "Please install it with `pip install scikit-learn`."
        )
    return cohen_kappa_score(gold_labels, predictions)

BaseLLMMetric dataclass

BaseLLMMetric(name: str)

Bases: ABC

Base class for simple LLM-based metrics that return MetricResult objects.

LLMMetric dataclass

LLMMetric(name: str, prompt: Optional[Union[str, Prompt]] = None)

Bases: BaseLLMMetric

LLM-based metric that uses prompts to generate structured responses.

get_correlation abstractmethod

get_correlation(gold_labels: List[str], predictions: List[str]) -> float

Calculate the correlation between gold scores and predicted scores. This is a placeholder method and should be implemented based on the specific metric.

Source code in src/ragas/metrics/llm_based.py

@abstractmethod
def get_correlation(
    self, gold_labels: t.List[str], predictions: t.List[str]
) -> float:
    """
    Calculate the correlation between gold scores and predicted scores.
    This is a placeholder method and should be implemented based on the specific metric.
    """
    pass

align_and_validate

align_and_validate(dataset: Dataset, embedding_model: Union[BaseRagasEmbeddings, BaseRagasEmbedding], llm: InstructorBaseRagasLLM, test_size: float = 0.2, random_state: int = 42, **kwargs: Dict[str, Any])

Args: dataset: experiment to align the metric with. embedding_model: The embedding model used for dynamic few-shot prompting. llm: The LLM instance to use for scoring.

Align the metric with the specified experiments and validate it against a gold standard experiment. This method combines alignment and validation into a single step.

Source code in src/ragas/metrics/llm_based.py

def align_and_validate(
    self,
    dataset: "Dataset",
    embedding_model: t.Union[BaseRagasEmbeddings, BaseRagasEmbedding],
    llm: BaseRagasLLM,
    test_size: float = 0.2,
    random_state: int = 42,
    **kwargs: t.Dict[str, t.Any],
):
    """
    Args:
        dataset: experiment to align the metric with.
        embedding_model: The embedding model used for dynamic few-shot prompting.
        llm: The LLM instance to use for scoring.

    Align the metric with the specified experiments and validate it against a gold standard experiment.
    This method combines alignment and validation into a single step.
    """
    train_dataset, test_dataset = dataset.train_test_split(
        test_size=test_size, random_state=random_state
    )

    self.align(train_dataset, embedding_model, **kwargs)  # type: ignore
    return self.validate_alignment(llm, test_dataset)  # type: ignore

align

align(train_dataset: Dataset, embedding_model: Union[BaseRagasEmbeddings, BaseRagasEmbedding], **kwargs: Dict[str, Any])

Args: train_dataset: train_dataset to align the metric with. embedding_model: The embedding model used for dynamic few-shot prompting.

Align the metric with the specified experiments by different optimization methods.

Source code in src/ragas/metrics/llm_based.py

def align(
    self,
    train_dataset: "Dataset",
    embedding_model: t.Union[BaseRagasEmbeddings, BaseRagasEmbedding],
    **kwargs: t.Dict[str, t.Any],
):
    """
    Args:
        train_dataset: train_dataset to align the metric with.
        embedding_model: The embedding model used for dynamic few-shot prompting.

    Align the metric with the specified experiments by different optimization methods.
    """

    # get prompt
    if not self.prompt:
        raise Exception("prompt not passed")
    self.prompt = (
        self.prompt if isinstance(self.prompt, Prompt) else Prompt(self.prompt)
    )
    # Extract specific parameters for from_prompt method
    max_similar_examples_val = kwargs.get("max_similar_examples", 3)
    similarity_threshold_val = kwargs.get("similarity_threshold", 0.7)
    max_similar_examples = (
        int(max_similar_examples_val)
        if isinstance(max_similar_examples_val, (int, str))
        else 3
    )
    similarity_threshold = (
        float(similarity_threshold_val)
        if isinstance(similarity_threshold_val, (int, float, str))
        else 0.7
    )
    # Convert BaseRagasEmbeddings to BaseRagasEmbedding if needed
    if isinstance(embedding_model, BaseRagasEmbeddings):
        # For legacy BaseRagasEmbeddings, we need to wrap it
        # Create a wrapper that implements BaseRagasEmbedding interface
        class EmbeddingWrapper(BaseRagasEmbedding):
            def __init__(self, legacy_embedding):
                self.legacy_embedding = legacy_embedding

            def embed_text(self, text: str, **kwargs) -> t.List[float]:
                return self.legacy_embedding.embed_query(text)

            async def aembed_text(self, text: str, **kwargs) -> t.List[float]:
                return await self.legacy_embedding.aembed_query(text)

        actual_embedding_model = EmbeddingWrapper(embedding_model)
    else:
        # Already BaseRagasEmbedding
        actual_embedding_model = embedding_model

    self.prompt = DynamicFewShotPrompt.from_prompt(
        self.prompt,
        actual_embedding_model,
        max_similar_examples,
        similarity_threshold,
    )
    train_dataset.reload()
    total_items = len(train_dataset)
    input_vars = self.get_variables()
    output_vars = [self.name, f"{self.name}_reason"]

    with Progress() as progress:
        task = progress.add_task("Processing examples", total=total_items)
        for row in train_dataset:
            inputs = {
                var: train_dataset.get_row_value(row, var) for var in input_vars
            }
            inputs = {k: v for k, v in inputs.items() if v is not None}
            output = {
                var: train_dataset.get_row_value(row, var) for var in output_vars
            }
            output = {k: v for k, v in output.items() if v is not None}

            if output:
                self.prompt.add_example(inputs, output)
            progress.update(task, advance=1)

validate_alignment

validate_alignment(llm: InstructorBaseRagasLLM, test_dataset: Dataset, mapping: Dict[str, str] = {})

Args: llm: The LLM instance to use for scoring. test_dataset: An Dataset instance containing the gold standard scores. mapping: A dictionary mapping variable names expected by metrics to their corresponding names in the gold experiment.

Validate the alignment of the metric by comparing the scores against a gold standard experiment. This method computes the Cohen's Kappa score and agreement rate between the gold standard scores and the predicted scores from the metric.

Source code in src/ragas/metrics/llm_based.py

def validate_alignment(
    self,
    llm: BaseRagasLLM,
    test_dataset: "Dataset",
    mapping: t.Dict[str, str] = {},
):
    """
    Args:
        llm: The LLM instance to use for scoring.
        test_dataset: An Dataset instance containing the gold standard scores.
        mapping: A dictionary mapping variable names expected by metrics to their corresponding names in the gold experiment.

    Validate the alignment of the metric by comparing the scores against a gold standard experiment.
    This method computes the Cohen's Kappa score and agreement rate between the gold standard scores and
    the predicted scores from the metric.
    """

    test_dataset.reload()
    gold_scores_raw = [
        test_dataset.get_row_value(row, self.name) for row in test_dataset
    ]
    pred_scores = []
    for row in test_dataset:
        values = {
            v: (
                test_dataset.get_row_value(row, v)
                if v not in mapping
                else test_dataset.get_row_value(row, mapping.get(v, v))
            )
            for v in self.get_variables()
        }
        score = self.score(llm=llm, **values)
        pred_scores.append(score.value)

    # Convert to strings for correlation calculation, filtering out None values
    gold_scores = [str(score) for score in gold_scores_raw if score is not None]
    pred_scores_str = [str(score) for score in pred_scores if score is not None]

    df = test_dataset.to_pandas()
    df[f"{self.name}_pred"] = pred_scores
    correlation = self.get_correlation(gold_scores, pred_scores_str)
    agreement_rate = sum(
        x == y for x, y in zip(gold_scores, pred_scores_str)
    ) / len(gold_scores)
    return {
        "correlation": correlation,
        "agreement_rate": agreement_rate,
        "df": df,
    }

NumericMetric dataclass

NumericMetric(name: str, prompt: Optional[Union[str, Prompt]] = None, allowed_values: Union[Tuple[float, float], range] = (0.0, 1.0))

Bases: LLMMetric

get_correlation

get_correlation(gold_labels: List[str], predictions: List[str]) -> float

Calculate the correlation between gold labels and predictions. This is a placeholder method and should be implemented based on the specific metric.

Source code in src/ragas/metrics/numeric.py

def get_correlation(
    self, gold_labels: t.List[str], predictions: t.List[str]
) -> float:
    """
    Calculate the correlation between gold labels and predictions.
    This is a placeholder method and should be implemented based on the specific metric.
    """
    try:
        from scipy.stats import pearsonr
    except ImportError:
        raise ImportError(
            "scipy is required for correlation calculation. "
            "Please install it with `pip install scipy`."
        )
    # Convert strings to floats for correlation calculation
    gold_floats = [float(x) for x in gold_labels]
    pred_floats = [float(x) for x in predictions]
    result = pearsonr(gold_floats, pred_floats)
    # pearsonr returns (correlation, p-value) tuple
    correlation = t.cast(float, result[0])
    return correlation

RankingMetric dataclass

RankingMetric(name: str, prompt: Optional[Union[str, Prompt]] = None, allowed_values: int = 2)

Bases: LLMMetric

get_correlation

get_correlation(gold_labels: List[str], predictions: List[str]) -> float

Calculate the correlation between gold labels and predictions. This is a placeholder method and should be implemented based on the specific metric.

Source code in src/ragas/metrics/ranking.py

def get_correlation(
    self, gold_labels: t.List[str], predictions: t.List[str]
) -> float:
    """
    Calculate the correlation between gold labels and predictions.
    This is a placeholder method and should be implemented based on the specific metric.
    """
    try:
        from sklearn.metrics import cohen_kappa_score
    except ImportError:
        raise ImportError(
            "scikit-learn is required for correlation calculation. "
            "Please install it with `pip install scikit-learn`."
        )

    kappa_scores = []
    for gold_item, prediction in zip(gold_labels, predictions):
        kappa = cohen_kappa_score(gold_item, prediction, weights="quadratic")
        kappa_scores.append(kappa)

    return sum(kappa_scores) / len(kappa_scores) if kappa_scores else 0.0

MetricResult

MetricResult(value: Any, reason: Optional[str] = None, traces: Optional[Dict[str, Any]] = None)

Class to hold the result of a metric evaluation.

This class behaves like its underlying result value but still provides access to additional metadata like reasoning.

Works with: - DiscreteMetrics (string results) - NumericMetrics (float/int results) - RankingMetrics (list results)

Source code in src/ragas/metrics/result.py

def __init__(
    self,
    value: t.Any,
    reason: t.Optional[str] = None,
    traces: t.Optional[t.Dict[str, t.Any]] = None,
):
    if traces is not None:
        invalid_keys = [
            key for key in traces.keys() if key not in {"input", "output"}
        ]
        if invalid_keys:
            raise ValueError(
                f"Invalid keys in traces: {invalid_keys}. Allowed keys are 'input' and 'output'."
            )
    self._value = value
    self.reason = reason
    self.traces = traces

value property

value

Get the raw result value.

to_dict

to_dict()

Convert the result to a dictionary.

Source code in src/ragas/metrics/result.py

def to_dict(self):
    """Convert the result to a dictionary."""
    return {"result": self._value, "reason": self.reason}

validate classmethod

validate(value: Any, info: ValidationInfo)

Provide compatibility with older Pydantic versions.

Source code in src/ragas/metrics/result.py

@classmethod
def validate(cls, value: t.Any, info: ValidationInfo):
    """Provide compatibility with older Pydantic versions."""
    if isinstance(value, MetricResult):
        return value
    return cls(value=value)