nq_eval

Functions

compute_f1	Computes F1, precision, recall for a list of answer scores.
compute_final_f1	Computes overall F1 given long and short answers, ignoring scores.
compute_optimal_metrics	Computes overall metrics for long and short answers for their respective optimal thresholds :param long_answer_stats: List of long answer scores.
compute_pr_curves	Computes PR curve and returns R@P for specific targets.
extract_metrics_at_optimal_threshold	param answer_stats one of the dictionaries returned from score_answers
get_metrics_as_dict	Library version of the end-to-end evaluation.
get_metrics_with_answer_stats	Generate metrics dict using long and short answer stats.
load_gt_lookup_as_dict
pretty_print
print_r_at_p_table	Pretty prints the R@P table for default targets.
safe_divide	Compute x / y, but return 0 if y is zero.
score_answers	Scores all answers for all documents.
score_long_answer	Scores a long answer as correct or not.
score_short_answer	Scores a short answer as correct or not.