/
reparameterized_sampling.py
51 lines (44 loc) · 1.72 KB
/
reparameterized_sampling.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
# coding=utf-8
# Copyright 2020 The TF-Agents Authors.
#
# 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
#
# https://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.
"""Helper function to do reparameterized sampling if the distributions supports it."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import tensorflow_probability as tfp
from tf_agents.distributions import gumbel_softmax
def sample(distribution, reparam=False, **kwargs):
"""Sample from distribution either with reparameterized sampling or regular sampling.
Args:
distribution: A `tfp.distributions.Distribution` instance.
reparam: Whether to use reparameterized sampling.
**kwargs: Parameters to be passed to distribution's sample() fucntion.
Returns:
"""
if reparam:
if (
distribution.reparameterization_type
!= tfp.distributions.FULLY_REPARAMETERIZED
):
raise ValueError(
'This distribution cannot be reparameterized: {}'.format(distribution)
)
else:
return distribution.sample(**kwargs)
else:
if isinstance(distribution, gumbel_softmax.GumbelSoftmax):
samples = distribution.sample(**kwargs)
return distribution.convert_to_one_hot(samples)
else:
return distribution.sample(**kwargs)