-
Notifications
You must be signed in to change notification settings - Fork 44
/
Copy pathagreement_test.py
82 lines (70 loc) · 2.74 KB
/
agreement_test.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
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
# Copyright 2024 DeepMind Technologies Limited
#
# 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.
# ==============================================================================
"""Tests for agreement computation."""
from absl.testing import absltest
from absl.testing import parameterized
import jax.numpy as jnp
import numpy as np
import agreement
class AgreementTest(parameterized.TestCase):
def test_leave_one_reader_out_agreement(self):
rankings = jnp.array([[[0, 1, 2]] * 3] * 2)
groups = jnp.array([
[
[0, -1, -1],
[0, 1, -1],
[0, 0, -1],
],
[
[0, -1, -1],
[0, 1, -1],
[-1, -1, -1],
],
])
num_readers = jnp.array([3, 2])
expected_group_overlap = jnp.array([
[1, 1.5, 1.5],
[0.5, 0.5, 0], # 0.5 because the last empty reader is still included.
])
# Basically looks at how many non-negative entries in groups overlap
# between readers and averages that across readers.
def average_group_overlap(
left_out_rankings,
left_out_groups, # pylint: disable=unused-argument
other_rankings,
other_groups, # pylint: disable=unused-argument
num_readers,
): # pylint: disable=unused-argument
left_out_groups = jnp.repeat(left_out_groups, 2, axis=1)
group_overlap = jnp.logical_and(left_out_groups >= 0, other_groups >= 0)
return jnp.mean(jnp.sum(group_overlap, axis=2), axis=1)
group_overlap = agreement.leave_one_reader_out_agreement(
rankings, groups, num_readers, average_group_overlap
)
np.testing.assert_array_almost_equal(group_overlap, expected_group_overlap)
def test_leave_one_reader_out_coverage_agreement(self):
rankings = jnp.array([[[0, 1, 2]] * 3] * 2)
groups = jnp.array([[[0, 1, -1]] * 3] * 2)
num_readers = jnp.array([3, 2])
standard_accuracy_agreement = (
agreement.leave_one_reader_out_coverage_agreement(
rankings, groups, num_readers
)
)
np.testing.assert_array_equal(standard_accuracy_agreement.shape, (2, 3))
self.assertTrue(jnp.all(standard_accuracy_agreement <= 1))
self.assertTrue(jnp.all(standard_accuracy_agreement >= 0))
if __name__ == '__main__':
absltest.main()