Creates a metric function that calculates the F1 score between predicted
and expected text based on token overlap.
Usage
metric_f1(field = NULL, normalize = TRUE)
Arguments
- field
Optional field name to extract from structured outputs
- normalize
Logical, whether to normalize text before tokenization
Value
A function with signature function(prediction, expected) -> numeric
Examples
# Token-based F1 score
metric <- metric_f1()
metric("the quick brown fox", "the fast brown fox") # 0.75
#> [1] 0.75
# Field extraction
metric <- metric_f1(field = "answer")
metric(
list(answer = "the quick brown fox"),
list(answer = "the fast brown fox")
)
#> [1] 0.75
