Nous.Eval.Evaluators.FuzzyMatch (nous v0.13.3)

View Source

Evaluator that uses string similarity for matching.

Uses Levenshtein distance to calculate similarity between strings.

Configuration

  • :threshold - Minimum similarity (0.0 to 1.0, default: 0.8)
  • :normalize - Normalize strings before comparison (default: true)
  • :case_insensitive - Ignore case (default: true)

Examples

TestCase.new(
  id: "fuzzy",
  input: "What is the capital of France?",
  expected: "Paris is the capital of France",
  eval_type: :fuzzy_match,
  eval_config: %{threshold: 0.7}
)

Summary

Functions

calculate_similarity(s1, s2)

Calculate similarity between two strings using Levenshtein distance.

levenshtein_distance(s1, s2)

Calculate the Levenshtein distance between two strings.

Functions

calculate_similarity(s1, s2)

@spec calculate_similarity(String.t(), String.t()) :: float()

Calculate similarity between two strings using Levenshtein distance.

Returns a value between 0.0 (completely different) and 1.0 (identical).

levenshtein_distance(s1, s2)

@spec levenshtein_distance(String.t(), String.t()) :: non_neg_integer()

Calculate the Levenshtein distance between two strings.