Files
2026-07-13 12:40:42 +08:00

209 lines
6.8 KiB
Python

# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
#
# 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.
# [AUTO-GENERATED] Unit test for paddle.tensor.random (bernoulli, binomial, standard_gamma)
# 自动生成的单测,覆盖 paddle.tensor.random 模块中未覆盖的代码路径
# Target: cover uncovered lines 130-142, 240-262, 301 in paddle/python/paddle/tensor/random.py
# 目标:覆盖 random.py 中 bernoulli 的静态图分支、binomial 的静态图分支
"""
This test covers the following modules and code paths:
这个测试覆盖以下模块和代码路径:
1. bernoulli() - 动态图路径 (lines 127-142)
2. bernoulli_() - inplace 版本 (lines 146-191)
3. binomial() - 动态图路径 (lines 235-262)
4. standard_gamma() - 基本功能 (lines 265-301)
5. multinomial() - 基本功能
"""
import unittest
import numpy as np
import paddle
class TestBernoulli(unittest.TestCase):
"""Test bernoulli() distribution.
测试 bernoulli() 分布采样。
覆盖 random.py 第 127-142 行。
"""
def setUp(self):
paddle.disable_static()
paddle.seed(42)
def test_bernoulli_default_p(self):
"""bernoulli with default p=0.5."""
x = paddle.zeros([2, 3], dtype='float32')
out = paddle.bernoulli(x)
self.assertEqual(out.shape, [2, 3])
# All values should be 0 or 1
result = out.numpy()
self.assertTrue(np.all((result == 0) | (result == 1)))
def test_bernoulli_custom_p(self):
"""bernoulli with custom probability."""
x = paddle.full([10, 10], 0.8, dtype='float32')
out = paddle.bernoulli(x)
result = out.numpy()
# Most values should be 1
self.assertTrue(np.mean(result) > 0.5)
def test_bernoulli_float16(self):
"""bernoulli with float16 input."""
x = paddle.full([5, 5], 0.5, dtype='float16')
out = paddle.bernoulli(x)
self.assertEqual(out.shape, [5, 5])
def test_bernoulli_float64(self):
"""bernoulli with float64 input."""
x = paddle.full([5, 5], 0.5, dtype='float64')
out = paddle.bernoulli(x)
self.assertEqual(out.shape, [5, 5])
def test_bernoulli_p_scalar(self):
"""bernoulli with scalar p parameter."""
x = paddle.zeros([2, 3])
out = paddle.bernoulli(x, p=0.8)
self.assertEqual(out.shape, [2, 3])
# Most values should be 1 since p=0.8
result = out.numpy()
self.assertTrue(np.mean(result) > 0.4)
def test_bernoulli_with_name(self):
"""bernoulli with name parameter."""
x = paddle.zeros([2, 3])
out = paddle.bernoulli(x, name='test_bernoulli')
self.assertEqual(out.shape, [2, 3])
class TestBernoulliInplace(unittest.TestCase):
"""Test bernoulli_() inplace operation.
测试 bernoulli_() 就地操作。
"""
def setUp(self):
paddle.disable_static()
paddle.seed(42)
def test_bernoulli_inplace(self):
"""bernoulli_ should modify tensor in-place."""
x = paddle.randn([3, 4])
x_id = id(x)
out = paddle.bernoulli_(x)
self.assertIs(out, x)
result = out.numpy()
self.assertTrue(np.all((result == 0) | (result == 1)))
def test_bernoulli_inplace_with_p(self):
"""bernoulli_ with custom p."""
x = paddle.randn([3, 4])
out = paddle.bernoulli_(x, p=0.9)
result = out.numpy()
self.assertTrue(np.mean(result) > 0.5)
class TestBinomial(unittest.TestCase):
"""Test binomial() distribution.
测试 binomial() 分布采样。
覆盖 random.py 第 235-262 行。
"""
def setUp(self):
paddle.disable_static()
paddle.seed(42)
def test_binomial_basic(self):
"""binomial basic usage."""
count = paddle.full([2, 3], 10, dtype='int32')
prob = paddle.full([2, 3], 0.5, dtype='float32')
out = paddle.binomial(count, prob)
self.assertEqual(out.shape, [2, 3])
result = out.numpy()
# Values should be between 0 and count
self.assertTrue(np.all(result >= 0))
self.assertTrue(np.all(result <= 10))
def test_binomial_float64(self):
"""binomial with float64 probability."""
count = paddle.full([5], 20, dtype='int64')
prob = paddle.full([5], 0.3, dtype='float64')
out = paddle.binomial(count, prob)
self.assertEqual(out.shape, [5])
def test_binomial_broadcast(self):
"""binomial with broadcastable shapes."""
count = paddle.full([2, 1], 10, dtype='int32')
prob = paddle.full([1, 3], 0.5, dtype='float32')
out = paddle.binomial(count, prob)
self.assertEqual(out.shape, [2, 3])
def test_binomial_high_prob(self):
"""binomial with high probability."""
count = paddle.full([100], 10, dtype='int32')
prob = paddle.full([100], 0.99, dtype='float32')
out = paddle.binomial(count, prob)
result = out.numpy()
self.assertTrue(np.mean(result) > 8.0)
class TestStandardGamma(unittest.TestCase):
"""Test standard_gamma() distribution.
测试 standard_gamma() 分布采样。
覆盖 random.py 第 265-301 行。
"""
def setUp(self):
paddle.disable_static()
paddle.seed(42)
def test_standard_gamma_basic(self):
"""standard_gamma basic usage."""
x = paddle.uniform([2, 3], min=1.0, max=5.0)
out = paddle.standard_gamma(x)
self.assertEqual(out.shape, [2, 3])
# Gamma distribution with alpha > 0 should produce positive values
result = out.numpy()
self.assertTrue(np.all(result >= 0))
def test_standard_gamma_float64(self):
"""standard_gamma with float64."""
x = paddle.uniform([2, 3], min=1.0, max=5.0, dtype='float64')
out = paddle.standard_gamma(x)
self.assertEqual(out.dtype, paddle.float64)
class TestPoisson(unittest.TestCase):
"""Test poisson() distribution.
测试 poisson() 分布采样。
"""
def setUp(self):
paddle.disable_static()
paddle.seed(42)
def test_poisson_basic(self):
"""poisson basic usage."""
x = paddle.full([2, 3], 5.0)
out = paddle.poisson(x)
self.assertEqual(out.shape, [2, 3])
result = out.numpy()
self.assertTrue(np.all(result >= 0))
if __name__ == '__main__':
unittest.main()