evaluate-v1 Namespace Reference

Functions
def	normalize_answer (s)
def	f1_score (prediction, ground_truth)
def	exact_match_score (prediction, ground_truth)
def	metric_max_over_ground_truths (metric_fn, prediction, ground_truths)
def	evaluate (dataset, predictions, f1_acc)

Variables
string	expected_version = '1.1'
	parser
	help
	args = parser.parse_args()
	dataset_json = json.load(dataset_file)
	file
	dataset = dataset_json['data']
	predictions = json.load(prediction_file)
	f1_acc = float(args.f1_acc)

Function Documentation

normalize_answer()

def evaluate-v1.normalize_answer

(

s

)

Lower text and remove punctuation, articles and extra whitespace.

Here is the caller graph for this function:

f1_score()

def evaluate-v1.f1_score	(		prediction,
			ground_truth
	)

Here is the call graph for this function:

exact_match_score()

def evaluate-v1.exact_match_score	(		prediction,
			ground_truth
	)

Here is the call graph for this function:

metric_max_over_ground_truths()

def evaluate-v1.metric_max_over_ground_truths	(		metric_fn,
			prediction,
			ground_truths
	)

Here is the caller graph for this function:

evaluate()

def evaluate-v1.evaluate	(		dataset,
			predictions,
			f1_acc
	)

Here is the call graph for this function:

Variable Documentation

expected_version

string evaluate-v1.expected_version = '1.1'

parser

evaluate-v1.parser

Initial value:

=  argparse.ArgumentParser(
        description='Evaluation for SQuAD ' + expected_version)

help

evaluate-v1.help

args

evaluate-v1.args = parser.parse_args()

dataset_json

evaluate-v1.dataset_json = json.load(dataset_file)

file

evaluate-v1.file

dataset

evaluate-v1.dataset = dataset_json['data']

predictions

evaluate-v1.predictions = json.load(prediction_file)

f1_acc

evaluate-v1.f1_acc = float(args.f1_acc)