Skip to content
Continuous Eval

Overview of Metrics

Metric Categories

The continuous-eval package offers three categories of metrics based on how they are computed:

  • Deterministic metrics: calculated based on statistical formulas
  • Semantic: calculated using smaller models
  • LLM-based: calculated by an Evaluation LLM with curated prompts

All the metrics comes with pros and cons and there’s not a one-size-fits-all evaluation pipeline that’s optimal for every use case. We aim to provide a wide range of metrics for you to choose from.

The package also offers a way to Ensemble Metrics of different metrics to improve performance on quality and effeciency.


Metric Class

Below is the list of metrics available:

Module Category Metrics
Retrieval Deterministic PrecisionRecallF1, RankedRetrievalMetrics
LLM-based LLMBasedContextPrecision, LLMBasedContextCoverage
Text Generation Deterministic DeterministicAnswerCorrectness, DeterministicFaithfulness, FleschKincaidReadability
Semantic DebertaAnswerScores, BertAnswerRelevance, BertAnswerSimilarity
LLM-based LLMBasedFaithfulness, LLMBasedAnswerCorrectness, LLMBasedAnswerRelevance, LLMBasedStyleConsistency
Classification Deterministic ClassificationAccuracy
Code Generation Deterministic CodeStringMatch, PythonASTSimilarity
LLM-based LLMBasedCodeGeneration
Agent Tools Deterministic ToolSelectionAccuracy
Custom Define your own metrics

Retrieval metrics

Deterministic

PrecisionRecallF1

  • Definition: Rank-agnostic metrics including Precision, Recall, and F1 of Retrieved Contexts
  • Inputs: retrieved_context, ground_truth_context

RankedRetrievalMetrics

  • Definition: Rank-aware metrics including Mean Average Precision (MAP), Mean Reciprical Rank (MRR), NDCG (Normalized Discounted Cumulative Gain) of retrieved contexts
  • Inputs: retrieved_context, ground_truth_context
LLM-based

LLMBasedContextPrecision

  • Definition: Precision and Mean Average Precision (MAP) based on context relevancy classified by LLM
  • Inputs: question, retrieved_context

LLMBasedContextCoverage

  • Definition: Proportion of statements in ground truth answer that can be attributed to Retrieved Contexts calculated by LLM
  • Inputs: question, retrieved_context, ground_truth_answers

Text Generation metrics

Deterministic

DeterministicAnswerRelevance

  • Definition: Includes Token Overlap (Precision, Recall, F1), ROUGE-L (Precision, Recall, F1), and BLEU score of Generated Answer vs. Ground Truth Answer
  • Inputs: question, generated_answer

DeterministicFaithfulness

  • Definition: Proportion of sentences in Answer that can be matched to Retrieved Contexts using ROUGE-L precision, Token Overlap precision, and BLEU score
  • Inputs: retrieved_context, generated_answer

FleschKincaidReadability

  • Definition: How easy or difficult it is to understand the LLM generated answer.
  • Inputs: generated_answer
Semantic

DebertaAnswerScores

  • Definition: Entailment and contradiction scores between the Generated Answer and Ground Truth Answer
  • Inputs: generated_answer, ground_truth_answers

BertAnswerRelevance

  • Definition: Similarity score based on the BERT model between the Generated Answer and Question
  • Inputs: question, generated_answer

BertAnswerSimilarity

  • Definition: Similarity score based on the BERT model between the Generated Answer and Ground Truth Answer
  • Inputs: generated_answer, ground_truth_answers
LLM-based

LLMBasedFaithfulness

  • Definition: Binary classifications of whether the statements in the Generated Answer can be attributed to the Retrieved Contexts by LLM
  • Inputs: question, retrieved_context, generated_answer

LLMBasedAnswerCorrectness

  • Definition: Overall correctness of the Generated Answer based on the Question and Ground Truth Answer calculated by LLM
  • Inputs: question, generated_answer, ground_truth_answers

LLMBasedAnswerRelevance

  • Definition: Relevance of the Generated Answer with respect to the Question
  • Inputs: question, generated_answer

LLMBasedStyleConsistency

  • Definition: Consistency of style between the Generated Answer and the Ground Truth Answer(s)
  • Inputs: generated_answer, ground_truth_answers

Code Generation metrics

Deterministic

DeterministicAnswerRelevance

  • Definition: Includes Token Overlap (Precision, Recall, F1), ROUGE-L (Precision, Recall, F1), and BLEU score of Generated Answer vs. Ground Truth Answer
  • Inputs: question, generated_answer

DeterministicFaithfulness

  • Definition: Proportion of sentences in Answer that can be matched to Retrieved Contexts using ROUGE-L precision, Token Overlap precision, and BLEU score
  • Inputs: retrieved_context, generated_answer

Classification metrics

Deterministic

ClassificationAccuracy

  • Definition: Proportion of correctly identified items out of the total items
  • Inputs: predictions, ground_truth_labels

Code Generation metrics

Deterministic

CodeStringMatch

  • Definition: Exact and fuzzy match scores between generated code strings and the ground truth code strings
  • Inputs: answer, ground_truths

PythonASTSimilarity

  • Definition: Similarity of Abstract Syntax Trees (ASTs) for Python code, comparing the generated code to the ground truth code
  • Inputs: answer, ground_truths

Agent Tools metrics

Deterministic

ToolSelectionAccuracy

  • Definition: Accuracy of selecting the correct tool(s) for a given task by the agent
  • Inputs: tools, ground_truths