wassname/Castor
1
0
Fork 0
mirror of https://github.com/wassname/Castor.git synced 2026-09-10 11:40:44 +08:00
Code Issues Packages Projects Releases Wiki Activity

Add twitter url dataset (#145)

Browse Source 
* add prediction/qrel files dump option

* fix comment

* add twitter-url dataset, minor refactor

* fix minor error
This commit is contained in:
Victor Yang
2018-09-08 11:16:23 +08:00
committed by GitHub
parent 5afb845a4b
commit 10e4e56817
4 changed files with 14 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files
common/dataset.py
+6 -3
View File
@@ -51,18 +51,21 @@ class DatasetFactory(object):
return TRECQA, embedding, train_loader, test_loader, dev_loader
elif dataset_name == 'wikiqa':
if not os.path.exists(os.path.join(castor_dir, utils_trecqa)):
raise FileNotFoundError('TrecQA requires the trec_eval tool to run. Please run get_trec_eval.sh inside Castor/utils (as working directory) before continuing.')
raise FileNotFoundError('WikiQA requires the trec_eval tool to run. Please run get_trec_eval.sh inside Castor/utils (as working directory) before continuing.')
dataset_root = os.path.join(castor_dir, os.pardir, 'Castor-data', 'datasets', 'WikiQA/')
train_loader, dev_loader, test_loader = WikiQA.iters(dataset_root, word_vectors_file, word_vectors_dir, batch_size, device=device, unk_init=UnknownWordVecCache.unk)
embedding = nn.Embedding.from_pretrained(WikiQA.TEXT_FIELD.vocab.vectors)
return WikiQA, embedding, train_loader, test_loader, dev_loader
elif dataset_name == 'pit2015':
if not os.path.exists(os.path.join(castor_dir, utils_trecqa)):
raise FileNotFoundError('TrecQA requires the trec_eval tool to run. Please run get_trec_eval.sh inside Castor/utils (as working directory) before continuing.')
dataset_root = os.path.join(castor_dir, os.pardir, 'Castor-data', 'datasets', 'SemEval-PIT2015/')
train_loader, dev_loader, test_loader = PIT2015.iters(dataset_root, word_vectors_file, word_vectors_dir, batch_size, device=device, unk_init=UnknownWordVecCache.unk)
embedding = nn.Embedding.from_pretrained(PIT2015.TEXT_FIELD.vocab.vectors)
return PIT2015, embedding, train_loader, test_loader, dev_loader
elif dataset_name == 'twitterurl':
dataset_root = os.path.join(castor_dir, os.pardir, 'Castor-data', 'datasets', 'Twitter-URL/')
train_loader, dev_loader, test_loader = PIT2015.iters(dataset_root, word_vectors_file, word_vectors_dir, batch_size, device=device, unk_init=UnknownWordVecCache.unk)
embedding = nn.Embedding.from_pretrained(PIT2015.TEXT_FIELD.vocab.vectors)
return PIT2015, embedding, train_loader, test_loader, dev_loader
else:
raise ValueError('{} is not a valid dataset.'.format(dataset_name))
common/evaluation.py
+2 -1
View File
@@ -19,7 +19,8 @@ class EvaluatorFactory(object):
'SST-2': SSTEvaluator,
'trecqa': TRECQAEvaluator,
'wikiqa': WikiQAEvaluator,
'pit2015': PIT2015Evaluator
'pit2015': PIT2015Evaluator,
'twitterurl': PIT2015Evaluator
}
evaluator_map_nce = {
common/evaluators/pit2015_evaluator.py
+4 -2
View File
@@ -21,8 +21,10 @@ class PIT2015Evaluator(Evaluator):
n_dev_correct += (prediction == gold_label).sum().item()
acc_total += ((prediction == batch.label.data) * (prediction == 1)).sum().item()
total_loss += F.nll_loss(scores, batch.label, size_average=False).item()
rel_total += batch.label.data.sum().item()
pre_total += torch.max(scores, 1)[1].view(batch.label.size()).data.sum().item()
rel_total += gold_label.sum().item()
pre_total += prediction.sum().item()
del scores
precision = acc_total / pre_total
recall = acc_total / rel_total
common/train.py
+2 -1
View File
@@ -19,7 +19,8 @@ class TrainerFactory(object):
'SST-2': SSTTrainer,
'trecqa': TRECQATrainer,
'wikiqa': WikiQATrainer,
'pit2015': PIT2015Trainer
'pit2015': PIT2015Trainer,
'twitterurl': PIT2015Trainer
}
trainer_map_nce = {