def compute_paragraph_score(sample, mode=None):
"""
For each paragraph, compute the f1 score compared with the question
Args:
sample: a sample in the dataset.
Returns:
None
Raises:
None
"""
if mode == "train" and "segmented_answers" in sample:
sentence = sample["segmented_answers"]
sentence += sample["segmented_question"]
else:
sentence = sample["segmented_question"]
for doc in sample['documents']:
doc['segmented_paragraphs_scores'] = []
for p_idx, (para_tokens, p_tokens) in enumerate(
zip(doc['segmented_paragraphs'], doc['paragraphs'])):
if len(sentence) > 0:
if mode == "train":
related_score = metric_max_over_ground_truths(
f1_score, para_tokens, sentence)
else:
related_score = metric_max_over_ground_truths(
f1_score, para_tokens, [sentence])
else:
related_score = 0.0
doc['segmented_paragraphs_scores'].append(related_score)
def compute_paragraph_score(sample):
"""
For each paragraph, compute the f1 score compared with the question
Args:
sample: a sample in the dataset.
Returns:
None
Raises:
None
"""
scores = []
question = sample["segmented_question"]
sample['segmented_paragraphs'] = []
sample['segmented_paragraphs_scores'] = []
for d_idx, doc in enumerate(sample['documents']):
for p_idx, para_tokens in enumerate(doc['segmented_paragraphs']):
if len(question) > 0:
related_score = metric_max_over_ground_truths(
f1_score, para_tokens, [question])
else:
related_score = 0.0
sample['segmented_paragraphs'].append(para_tokens)
sample['segmented_paragraphs_scores'].append(related_score)
scores.append((related_score, d_idx, p_idx))
with open('scores.txt', 'a') as f:
f.write(str(scores) + '\n')
def compute_paragraph_score(sample):
"""
For each paragraph, compute the f1 score compared with the question
Args:
sample: a sample in the dataset.
Returns:
None
Raises:
None
"""
question = sample["segmented_question"]
for doc in sample['documents']:
doc['segmented_paragraphs_scores'] = []
for p_idx, para_tokens in enumerate(doc['segmented_paragraphs']):
if len(question) > 0:
related_score = metric_max_over_ground_truths(
f1_score, para_tokens, [question])
else:
related_score = 0.0
doc['segmented_paragraphs_scores'].append(related_score)