lighteval/pr_1326/en/package_reference/metrics.md

Metrics

Metrics

Metric[[lighteval.metrics.Metric]]

lighteval.metrics.Metric[[lighteval.metrics.Metric]]

lighteval.metrics.Metric(metric_name: str, higher_is_better: bool, category: SamplingMethod, sample_level_fn: lighteval.metrics.metrics_sample.SampleLevelComputation | lighteval.metrics.sample_preparator.Preparator, corpus_level_fn: typing.Union[lighteval.metrics.metrics_corpus.CorpusLevelComputation, typing.Callable], batched_compute: bool = False)

Source

CorpusLevelMetric[[lighteval.metrics.utils.metric_utils.CorpusLevelMetric]]

lighteval.metrics.utils.metric_utils.CorpusLevelMetric[[lighteval.metrics.utils.metric_utils.CorpusLevelMetric]]

lighteval.metrics.utils.metric_utils.CorpusLevelMetric(metric_name: str, higher_is_better: bool, category: SamplingMethod, sample_level_fn: lighteval.metrics.metrics_sample.SampleLevelComputation | lighteval.metrics.sample_preparator.Preparator, corpus_level_fn: typing.Union[lighteval.metrics.metrics_corpus.CorpusLevelComputation, typing.Callable], batched_compute: bool = False)

Source

Metric computed over the whole corpora, with computations happening at the aggregation phase

SampleLevelMetric[[lighteval.metrics.utils.metric_utils.SampleLevelMetric]]

lighteval.metrics.utils.metric_utils.SampleLevelMetric[[lighteval.metrics.utils.metric_utils.SampleLevelMetric]]

lighteval.metrics.utils.metric_utils.SampleLevelMetric(metric_name: str, higher_is_better: bool, category: SamplingMethod, sample_level_fn: lighteval.metrics.metrics_sample.SampleLevelComputation | lighteval.metrics.sample_preparator.Preparator, corpus_level_fn: typing.Union[lighteval.metrics.metrics_corpus.CorpusLevelComputation, typing.Callable], batched_compute: bool = False)

Source

Metric computed per sample, then aggregated over the corpus

MetricGrouping[[lighteval.metrics.utils.metric_utils.MetricGrouping]]

lighteval.metrics.utils.metric_utils.MetricGrouping[[lighteval.metrics.utils.metric_utils.MetricGrouping]]

lighteval.metrics.utils.metric_utils.MetricGrouping(metric_name: list, higher_is_better: dict, category: SamplingMethod, sample_level_fn: lighteval.metrics.metrics_sample.SampleLevelComputation | lighteval.metrics.sample_preparator.Preparator, corpus_level_fn: dict, batched_compute: bool = False)

Source

Some metrics are more advantageous to compute together at once. For example, if a costly preprocessing is the same for all metrics, it makes more sense to compute it once.

CorpusLevelMetricGrouping[[lighteval.metrics.utils.metric_utils.CorpusLevelMetricGrouping]]

lighteval.metrics.utils.metric_utils.CorpusLevelMetricGrouping[[lighteval.metrics.utils.metric_utils.CorpusLevelMetricGrouping]]

lighteval.metrics.utils.metric_utils.CorpusLevelMetricGrouping(metric_name: list, higher_is_better: dict, category: SamplingMethod, sample_level_fn: lighteval.metrics.metrics_sample.SampleLevelComputation | lighteval.metrics.sample_preparator.Preparator, corpus_level_fn: dict, batched_compute: bool = False)

Source

MetricGrouping computed over the whole corpora, with computations happening at the aggregation phase

SampleLevelMetricGrouping[[lighteval.metrics.utils.metric_utils.SampleLevelMetricGrouping]]

lighteval.metrics.utils.metric_utils.SampleLevelMetricGrouping[[lighteval.metrics.utils.metric_utils.SampleLevelMetricGrouping]]

lighteval.metrics.utils.metric_utils.SampleLevelMetricGrouping(metric_name: list, higher_is_better: dict, category: SamplingMethod, sample_level_fn: lighteval.metrics.metrics_sample.SampleLevelComputation | lighteval.metrics.sample_preparator.Preparator, corpus_level_fn: dict, batched_compute: bool = False)

Source

MetricGrouping are computed per sample, then aggregated over the corpus

Corpus Metrics

CorpusLevelF1Score[[lighteval.metrics.metrics_corpus.CorpusLevelF1Score]]

lighteval.metrics.metrics_corpus.CorpusLevelF1Score[[lighteval.metrics.metrics_corpus.CorpusLevelF1Score]]

lighteval.metrics.metrics_corpus.CorpusLevelF1Score(average: str, num_classes: int = 2)

Source

compute_corpus[[lighteval.metrics.metrics_corpus.CorpusLevelF1Score.compute_corpus]]

compute_corpus(items: list)

Source

Computes the metric score over all the corpus generated items, by using the scikit learn implementation.

CorpusLevelPerplexityMetric[[lighteval.metrics.metrics_corpus.CorpusLevelPerplexityMetric]]

lighteval.metrics.metrics_corpus.CorpusLevelPerplexityMetric[[lighteval.metrics.metrics_corpus.CorpusLevelPerplexityMetric]]

lighteval.metrics.metrics_corpus.CorpusLevelPerplexityMetric(metric_type: str)

Source

compute_corpus[[lighteval.metrics.metrics_corpus.CorpusLevelPerplexityMetric.compute_corpus]]

compute_corpus(items: list)

Source

Computes the metric score over all the corpus generated items.

CorpusLevelTranslationMetric[[lighteval.metrics.metrics_corpus.CorpusLevelTranslationMetric]]

lighteval.metrics.metrics_corpus.CorpusLevelTranslationMetric[[lighteval.metrics.metrics_corpus.CorpusLevelTranslationMetric]]

lighteval.metrics.metrics_corpus.CorpusLevelTranslationMetric(metric_type: str, lang: typing.Literal['zh', 'ja', 'ko', ''] = '')

Source

compute_corpus[[lighteval.metrics.metrics_corpus.CorpusLevelTranslationMetric.compute_corpus]]

compute_corpus(items: list)

Source

Computes the metric score over all the corpus generated items, by using the sacrebleu implementation.

MatthewsCorrCoef[[lighteval.metrics.metrics_corpus.MatthewsCorrCoef]]

lighteval.metrics.metrics_corpus.MatthewsCorrCoef[[lighteval.metrics.metrics_corpus.MatthewsCorrCoef]]

lighteval.metrics.metrics_corpus.MatthewsCorrCoef()

Source

compute_corpus[[lighteval.metrics.metrics_corpus.MatthewsCorrCoef.compute_corpus]]

compute_corpus(items: list)

Source

Parameters:

items (list[dict]) : List of GenerativeCorpusMetricInput

Returns: float

Score

Computes the Matthews Correlation Coefficient, using scikit learn (doc).

Sample Metrics

ExactMatches[[lighteval.metrics.metrics_sample.ExactMatches]]

lighteval.metrics.metrics_sample.ExactMatches[[lighteval.metrics.metrics_sample.ExactMatches]]

lighteval.metrics.metrics_sample.ExactMatches(aggregation_function: typing.Callable[[list[float]], float] = <built-in function max>, normalize_gold: typing.Optional[typing.Callable[[str], str]] = None, normalize_pred: typing.Optional[typing.Callable[[str], str]] = None, strip_strings: bool = False, type_exact_match: str = 'full')

Source

compute[[lighteval.metrics.metrics_sample.ExactMatches.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

doc (Doc) : The document containing gold references.

model_response (ModelResponse) : The model's response containing predictions.

• **kwargs : Additional keyword arguments.

Returns: float

Aggregated score over the current sample's items.

Computes the metric over a list of golds and predictions for one single sample.

compute_one_item[[lighteval.metrics.metrics_sample.ExactMatches.compute_one_item]]

compute_one_item(gold: str, pred: str)

Source

Parameters:

gold (str) : One of the possible references

pred (str) : One of the possible predictions

Returns: float

The exact match score. Will be 1 for a match, 0 otherwise.

Compares two strings only.

F1_score[[lighteval.metrics.metrics_sample.F1_score]]

lighteval.metrics.metrics_sample.F1_score[[lighteval.metrics.metrics_sample.F1_score]]

lighteval.metrics.metrics_sample.F1_score(aggregation_function: typing.Callable[[list[float]], float] = <built-in function max>, normalize_gold: typing.Optional[typing.Callable[[str], str]] = None, normalize_pred: typing.Optional[typing.Callable[[str], str]] = None, strip_strings: bool = False)

Source

compute[[lighteval.metrics.metrics_sample.F1_score.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

doc (Doc) : The document containing gold references.

model_response (ModelResponse) : The model's response containing predictions.

• **kwargs : Additional keyword arguments.

Returns: float

Aggregated score over the current sample's items.

Computes the metric over a list of golds and predictions for one single sample.

compute_one_item[[lighteval.metrics.metrics_sample.F1_score.compute_one_item]]

compute_one_item(gold: str, pred: str)

Source

Parameters:

gold (str) : One of the possible references

pred (str) : One of the possible predictions

Returns: float

The f1 score over the bag of words, computed using nltk.

Compares two strings only.

LoglikelihoodAcc[[lighteval.metrics.metrics_sample.LoglikelihoodAcc]]

lighteval.metrics.metrics_sample.LoglikelihoodAcc[[lighteval.metrics.metrics_sample.LoglikelihoodAcc]]

lighteval.metrics.metrics_sample.LoglikelihoodAcc(logprob_normalization: lighteval.metrics.normalizations.LogProbCharNorm | lighteval.metrics.normalizations.LogProbTokenNorm | lighteval.metrics.normalizations.LogProbPMINorm | None = None)

Source

compute[[lighteval.metrics.metrics_sample.LoglikelihoodAcc.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

doc (Doc) : The document containing choices and gold indices.

model_response (ModelResponse) : The model's response containing logprobs.

• **kwargs : Additional keyword arguments.

Returns: int

The eval score: 1 if the best log-prob choice is in gold, 0 otherwise.

Computes the log likelihood accuracy: is the choice with the highest logprob in choices_logprob present in the gold_ixs?

NormalizedMultiChoiceProbability[[lighteval.metrics.metrics_sample.NormalizedMultiChoiceProbability]]

lighteval.metrics.metrics_sample.NormalizedMultiChoiceProbability[[lighteval.metrics.metrics_sample.NormalizedMultiChoiceProbability]]

lighteval.metrics.metrics_sample.NormalizedMultiChoiceProbability(log_prob_normalization: lighteval.metrics.normalizations.LogProbCharNorm | lighteval.metrics.normalizations.LogProbTokenNorm | lighteval.metrics.normalizations.LogProbPMINorm | None = None, aggregation_function: typing.Callable[[numpy.ndarray], float] = <function max at 0x7f69f4fd73f0>)

Source

compute[[lighteval.metrics.metrics_sample.NormalizedMultiChoiceProbability.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

doc (Doc) : The document containing choices and gold indices.

model_response (ModelResponse) : The model's response containing logprobs.

• **kwargs : Additional keyword arguments.

Returns: float

The probability of the best log-prob choice being a gold choice.

Computes the log likelihood probability: chance of choosing the best choice.

Probability[[lighteval.metrics.metrics_sample.Probability]]

lighteval.metrics.metrics_sample.Probability[[lighteval.metrics.metrics_sample.Probability]]

lighteval.metrics.metrics_sample.Probability(normalization: lighteval.metrics.normalizations.LogProbTokenNorm | None = None, aggregation_function: typing.Callable[[numpy.ndarray], float] = <function max at 0x7f69f4fd73f0>)

Source

compute[[lighteval.metrics.metrics_sample.Probability.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

doc (Doc) : The document containing choices and gold indices.

model_response (ModelResponse) : The model's response containing logprobs.

• **kwargs : Additional keyword arguments.

Returns: float

The probability of the best log-prob choice being a gold choice.

Computes the log likelihood probability: chance of choosing the best choice.

Recall[[lighteval.metrics.metrics_sample.Recall]]

lighteval.metrics.metrics_sample.Recall[[lighteval.metrics.metrics_sample.Recall]]

lighteval.metrics.metrics_sample.Recall(k: int)

Source

compute[[lighteval.metrics.metrics_sample.Recall.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

doc (Doc) : The document containing choices and gold indices.

model_response (ModelResponse) : The model's response containing logprobs.

• **kwargs : Additional keyword arguments.

Returns: int

Score: 1 if one of the top level predicted choices was correct, 0 otherwise.

Computes the recall at the requested depth level: looks at the n best predicted choices (with the highest log probabilities) and see if there is an actual gold among them.

MRR[[lighteval.metrics.metrics_sample.MRR]]

lighteval.metrics.metrics_sample.MRR[[lighteval.metrics.metrics_sample.MRR]]

lighteval.metrics.metrics_sample.MRR(length_normalization: bool = False)

Source

compute[[lighteval.metrics.metrics_sample.MRR.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

model_response (ModelResponse) : The model's response containing logprobs.

doc (Doc) : The document containing choices and gold indices.

• **kwargs : Additional keyword arguments.

Returns: float

MRR score.

Mean reciprocal rank. Measures the quality of a ranking of choices (ordered by correctness).

ROUGE[[lighteval.metrics.metrics_sample.ROUGE]]

lighteval.metrics.metrics_sample.ROUGE[[lighteval.metrics.metrics_sample.ROUGE]]

lighteval.metrics.metrics_sample.ROUGE(methods: str | list[str], multiple_golds: bool = False, bootstrap: bool = False, normalize_gold: typing.Optional[typing.Callable] = None, normalize_pred: typing.Optional[typing.Callable] = None, aggregation_function: typing.Optional[typing.Callable] = None, tokenizer: object = None)

Source

compute[[lighteval.metrics.metrics_sample.ROUGE.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

doc (Doc) : The document containing gold references.

model_response (ModelResponse) : The model's response containing predictions.

• **kwargs : Additional keyword arguments.

Returns: float or dict

Aggregated score over the current sample's items. If several rouge functions have been selected, returns a dict which maps name and scores.

Computes the metric(s) over a list of golds and predictions for one single sample.

BertScore[[lighteval.metrics.metrics_sample.BertScore]]

lighteval.metrics.metrics_sample.BertScore[[lighteval.metrics.metrics_sample.BertScore]]

lighteval.metrics.metrics_sample.BertScore(normalize_gold: typing.Optional[typing.Callable] = None, normalize_pred: typing.Optional[typing.Callable] = None)

Source

compute[[lighteval.metrics.metrics_sample.BertScore.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

doc (Doc) : The document containing gold references.

model_response (ModelResponse) : The model's response containing predictions.

• **kwargs : Additional keyword arguments.

Returns: dict

Scores over the current sample's items.

Computes the prediction, recall and f1 score using the bert scorer.

Extractiveness[[lighteval.metrics.metrics_sample.Extractiveness]]

lighteval.metrics.metrics_sample.Extractiveness[[lighteval.metrics.metrics_sample.Extractiveness]]

lighteval.metrics.metrics_sample.Extractiveness(normalize_input: callable = <function remove_braces at 0x7f68fefa1120>, normalize_pred: callable = <function remove_braces_and_strip at 0x7f68fefa11b0>, input_column: str = 'text', language: typing.Literal['en', 'de', 'fr', 'it'] = 'en')

Source

compute[[lighteval.metrics.metrics_sample.Extractiveness.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

doc (Doc) : The document containing input text.

model_response (ModelResponse) : The model's response containing predictions.

• **kwargs : Additional keyword arguments.

Returns: dict[str, float]

The extractiveness scores.

Compute the extractiveness of the predictions.

This method calculates coverage, density, and compression scores for a single prediction against the input text.

Faithfulness[[lighteval.metrics.metrics_sample.Faithfulness]]

lighteval.metrics.metrics_sample.Faithfulness[[lighteval.metrics.metrics_sample.Faithfulness]]

lighteval.metrics.metrics_sample.Faithfulness(normalize_input: typing.Callable = <function remove_braces at 0x7f68fefa1120>, normalize_pred: typing.Callable = <function remove_braces_and_strip at 0x7f68fefa11b0>, input_column: str = 'text')

Source

compute[[lighteval.metrics.metrics_sample.Faithfulness.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

doc (Doc) : The document containing input text.

model_response (ModelResponse) : The model's response containing predictions.

• **kwargs : Additional keyword arguments.

Returns: dict[str, float]

The faithfulness scores.

Compute the faithfulness of the predictions.

The SummaCZS (Summary Content Zero-Shot) model is used with configurable granularity and model variation.

BLEURT[[lighteval.metrics.metrics_sample.BLEURT]]

lighteval.metrics.metrics_sample.BLEURT[[lighteval.metrics.metrics_sample.BLEURT]]

lighteval.metrics.metrics_sample.BLEURT()

Source

compute[[lighteval.metrics.metrics_sample.BLEURT.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

doc (Doc) : The document containing gold references.

model_response (ModelResponse) : The model's response containing predictions.

• **kwargs : Additional keyword arguments.

Returns: float

Score over the current sample's items.

Uses the stored BLEURT scorer to compute the score on the current sample.

BLEU[[lighteval.metrics.metrics_sample.BLEU]]

lighteval.metrics.metrics_sample.BLEU[[lighteval.metrics.metrics_sample.BLEU]]

lighteval.metrics.metrics_sample.BLEU(n_gram: int)

Source

compute[[lighteval.metrics.metrics_sample.BLEU.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

doc (Doc) : The document containing gold references.

model_response (ModelResponse) : The model's response containing predictions.

• **kwargs : Additional keyword arguments.

Returns: float

Score over the current sample's items.

Computes the sentence level BLEU between the golds and each prediction, then takes the average.

StringDistance[[lighteval.metrics.metrics_sample.StringDistance]]

lighteval.metrics.metrics_sample.StringDistance[[lighteval.metrics.metrics_sample.StringDistance]]

lighteval.metrics.metrics_sample.StringDistance(metric_types: list[str] | str, strip_prediction: bool = True)

Source

compute[[lighteval.metrics.metrics_sample.StringDistance.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

doc (Doc) : The document containing gold references.

model_response (ModelResponse) : The model's response containing predictions.

• **kwargs : Additional keyword arguments.

Returns: dict

The different scores computed

Computes all the requested metrics on the golds and prediction.

edit_similarity[[lighteval.metrics.metrics_sample.StringDistance.edit_similarity]]

edit_similarity(s1, s2)

Source

Returns: float

Edit similarity score between 0 and 1

Compute the edit similarity between two lists of strings.

Edit similarity is also used in the paper Lee, Katherine, et al. "Deduplicating training data makes language models better." arXiv preprint arXiv:2107.06499 (2021).

longest_common_prefix_length[[lighteval.metrics.metrics_sample.StringDistance.longest_common_prefix_length]]

longest_common_prefix_length(s1: ndarray, s2: ndarray)

Source

Compute the length of the longest common prefix.

Metrics allowing sampling

PassAtK[[lighteval.metrics.metrics_sample.PassAtK]]

lighteval.metrics.metrics_sample.PassAtK[[lighteval.metrics.metrics_sample.PassAtK]]

lighteval.metrics.metrics_sample.PassAtK(k: int | None = None, n: int | None = None, **kwargs)

Source

compute[[lighteval.metrics.metrics_sample.PassAtK.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

doc (Doc) : The document containing gold references.

model_response (ModelResponse) : The model's response containing predictions.

• **kwargs : Additional keyword arguments.

Returns: float

Aggregated score over the current sample's items.

Computes the metric over a list of golds and predictions for one single item with possibly many samples. It applies normalisation (if needed) to model prediction and gold, computes their per prediction score, then aggregates the scores over the samples using a pass@k.

pass_at_k[[lighteval.metrics.metrics_sample.PassAtK.pass_at_k]]

pass_at_k(all_scores: list)

Source

Algo from https://arxiv.org/pdf/2107.03374

MajAtN[[lighteval.metrics.metrics_sample.MajAtN]]

lighteval.metrics.metrics_sample.MajAtN[[lighteval.metrics.metrics_sample.MajAtN]]

lighteval.metrics.metrics_sample.MajAtN(n: int | None = None, **kwargs)

Source

compute[[lighteval.metrics.metrics_sample.MajAtN.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

doc (Doc) : The document containing gold references.

model_response (ModelResponse) : The model's response containing predictions.

• **kwargs : Additional keyword arguments.

Returns: float

Aggregated score over the current sample's items.

Computes the metric over a list of golds and predictions for one single sample. It applies normalisation (if needed) to model prediction and gold, and takes the most frequent answer of all the available ones, then compares it to the gold.

AvgAtN[[lighteval.metrics.metrics_sample.AvgAtN]]

lighteval.metrics.metrics_sample.AvgAtN[[lighteval.metrics.metrics_sample.AvgAtN]]

lighteval.metrics.metrics_sample.AvgAtN(n: int | None = None, **kwargs)

Source

compute[[lighteval.metrics.metrics_sample.AvgAtN.compute]]

compute(doc: Doc, model_response: ModelResponse, **kwargs)

Source

Parameters:

model_response (ModelResponse) : The model's response containing predictions.

doc (Doc) : The document containing gold references.

• **kwargs : Additional keyword arguments.

Returns: float

Aggregated score over the current sample's items.

Computes the metric over a list of golds and predictions for one single sample. It applies normalisation (if needed) to model prediction and gold, and takes the most frequent answer of all the available ones, then compares it to the gold.

LLM-as-a-Judge

JudgeLM[[lighteval.metrics.utils.llm_as_judge.JudgeLM]]

lighteval.metrics.utils.llm_as_judge.JudgeLM[[lighteval.metrics.utils.llm_as_judge.JudgeLM]]

lighteval.metrics.utils.llm_as_judge.JudgeLM(model: str, templates: typing.Callable, process_judge_response: typing.Callable, judge_backend: typing.Literal['litellm', 'openai', 'transformers', 'tgi', 'vllm', 'inference-providers'], url: str | None = None, api_key: str | None = None, max_tokens: int | None = None, response_format: BaseModel = None, hf_provider: typing.Optional[typing.Literal['black-forest-labs', 'cerebras', 'cohere', 'fal-ai', 'fireworks-ai', 'inference-providers', 'hyperbolic', 'nebius', 'novita', 'openai', 'replicate', 'sambanova', 'together']] = None, backend_options: dict | None = None)

Source

Parameters:

model (str) : The name of the model.

templates (Callable) : A function taking into account the question, options, answer, and gold and returning the judge prompt.

process_judge_response (Callable) : A function for processing the judge's response.

judge_backend (Literal["litellm", "openai", "transformers", "tgi", "vllm", "inference-providers"]) : The backend for the judge.

url (str | None) : The URL for the OpenAI API.

api_key (str | None) : The API key for the OpenAI API (either OpenAI or HF key). Stored internally as a SecretStr so it is masked in logs, reprs, and serialized configs.

max_tokens (int) : The maximum number of tokens to generate. Defaults to 512.

response_format (BaseModel | None) : The format of the response from the API, used for the OpenAI and TGI backend.

hf_provider (Literal["black-forest-labs", "cerebras", "cohere", "fal-ai", "fireworks-ai", : "inference-providers", "hyperbolic", "nebius", "novita", "openai", "replicate", "sambanova", "together"] | None): The HuggingFace provider when using the inference-providers backend.

backend_options (dict | None) : Options for the backend. Currently only supported for litellm.

A class representing a judge for evaluating answers using either the chosen backend.

Methods: evaluate_answer: Evaluates an answer using the OpenAI API or Transformers library. __lazy_load_client: Lazy loads the OpenAI client or Transformers pipeline. __call_api: Calls the API to get the judge's response. __call_transformers: Calls the Transformers pipeline to get the judge's response. __call_vllm: Calls the VLLM pipeline to get the judge's response.

dict_of_lists_to_list_of_dicts[[lighteval.metrics.utils.llm_as_judge.JudgeLM.dict_of_lists_to_list_of_dicts]]

dict_of_lists_to_list_of_dicts(dict_of_lists)

Source

Parameters:

dict_of_lists : A dictionary where each value is a list. All lists are expected to have the same length.

Returns:

A list of dictionaries.

Transform a dictionary of lists into a list of dictionaries.

Each dictionary in the output list will contain one element from each list in the input dictionary, with the same keys as the input dictionary.

Example:

dict_of_lists_to_list_of_dicts({'k': [1, 2, 3], 'k2': ['a', 'b', 'c']}) [{'k': 1, 'k2': 'a'}, {'k': 2, 'k2': 'b'}, {'k': 3, 'k2': 'c'}]

evaluate_answer[[lighteval.metrics.utils.llm_as_judge.JudgeLM.evaluate_answer]]

evaluate_answer(question: str, answer: str, options: list[str] | None = None, gold: str | None = None)

Source

Parameters:

question (str) : The prompt asked to the evaluated model.

answer (str) : Answer given by the evaluated model.

options (list[str] | None) : Optional list of answer options.

gold (str | None) : Optional reference answer.

Returns:

A tuple containing the score, prompts, and judgment.

Evaluates an answer using either Transformers or OpenAI API.

JudgeLLM[[lighteval.metrics.metrics_sample.JudgeLLM]]

lighteval.metrics.metrics_sample.JudgeLLM[[lighteval.metrics.metrics_sample.JudgeLLM]]

lighteval.metrics.metrics_sample.JudgeLLM(judge_model_name: str, template: typing.Callable, process_judge_response: typing.Callable, judge_backend: typing.Literal['litellm', 'openai', 'transformers', 'vllm', 'tgi', 'inference-providers'], short_judge_name: str | None = None, response_format: pydantic.main.BaseModel | None = None, url: str | None = None, api_key: str | None = None, hf_provider: str | None = None, max_tokens: int | None = None, backend_options: dict | None = None)

Source

JudgeLLMMTBench[[lighteval.metrics.metrics_sample.JudgeLLMMTBench]]

lighteval.metrics.metrics_sample.JudgeLLMMTBench[[lighteval.metrics.metrics_sample.JudgeLLMMTBench]]

lighteval.metrics.metrics_sample.JudgeLLMMTBench(judge_model_name: str, template: typing.Callable, process_judge_response: typing.Callable, judge_backend: typing.Literal['litellm', 'openai', 'transformers', 'vllm', 'tgi', 'inference-providers'], short_judge_name: str | None = None, response_format: pydantic.main.BaseModel | None = None, url: str | None = None, api_key: str | None = None, hf_provider: str | None = None, max_tokens: int | None = None, backend_options: dict | None = None)

Source

compute[[lighteval.metrics.metrics_sample.JudgeLLMMTBench.compute]]

compute(model_response: list, doc: list, **kwargs)

Source

Compute the score of a generative task using a llm as a judge. The generative task can be multiturn with 2 turns max, in that case, we return scores for turn 1 and 2. Also returns user_prompt and judgement which are ignored later by the aggregator.

JudgeLLMMixEval[[lighteval.metrics.metrics_sample.JudgeLLMMixEval]]

lighteval.metrics.metrics_sample.JudgeLLMMixEval[[lighteval.metrics.metrics_sample.JudgeLLMMixEval]]

lighteval.metrics.metrics_sample.JudgeLLMMixEval(judge_model_name: str, template: typing.Callable, process_judge_response: typing.Callable, judge_backend: typing.Literal['litellm', 'openai', 'transformers', 'vllm', 'tgi', 'inference-providers'], short_judge_name: str | None = None, response_format: pydantic.main.BaseModel | None = None, url: str | None = None, api_key: str | None = None, hf_provider: str | None = None, max_tokens: int | None = None, backend_options: dict | None = None)

Source

compute[[lighteval.metrics.metrics_sample.JudgeLLMMixEval.compute]]

compute(responses: list, docs: list, **kwargs)

Source

Compute the score of a generative task using a llm as a judge. The generative task can be multiturn with 2 turns max, in that case, we return scores for turn 1 and 2. Also returns user_prompt and judgement which are ignored later by the aggregator.