-
Notifications
You must be signed in to change notification settings - Fork 386
/
test_encoder_model_ance.py
58 lines (46 loc) · 2.08 KB
/
test_encoder_model_ance.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
#
# Pyserini: Reproducible IR research with sparse and dense representations
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
import unittest
from itertools import islice
import numpy as np
from pyserini.encode import QueryEncoder, AnceQueryEncoder
from pyserini.search import get_topics
class TestEncodeAnce(unittest.TestCase):
def test_ance_encoded_queries(self):
encoded = QueryEncoder.load_encoded_queries('ance-msmarco-passage-dev-subset')
topics = get_topics('msmarco-passage-dev-subset')
for t in topics:
self.assertTrue(topics[t]['title'] in encoded.embedding)
encoded = QueryEncoder.load_encoded_queries('ance-dl19-passage')
topics = get_topics('dl19-passage')
for t in topics:
self.assertTrue(topics[t]['title'] in encoded.embedding)
encoded = QueryEncoder.load_encoded_queries('ance-dl20')
topics = get_topics('dl20')
for t in topics:
self.assertTrue(topics[t]['title'] in encoded.embedding)
def test_ance_encoder(self):
encoder = AnceQueryEncoder('castorini/ance-msmarco-passage')
cached_encoder = QueryEncoder.load_encoded_queries('ance-dl20')
topics = get_topics('dl20')
# Just test the first 10 topics
for t in dict(islice(topics.items(), 10)):
cached_vector = np.array(cached_encoder.encode(topics[t]['title']))
encoded_vector = np.array(encoder.encode(topics[t]['title']))
l1 = np.sum(np.abs(cached_vector - encoded_vector))
self.assertTrue(l1 < 0.0005)
if __name__ == '__main__':
unittest.main()