99 lines
3.1 KiB
Python
99 lines
3.1 KiB
Python
# Copyright (c) 2025 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.
|
|
|
|
|
|
import unittest
|
|
|
|
import paddle
|
|
from paddle import nn
|
|
|
|
|
|
class TestLSTMCellCompat(unittest.TestCase):
|
|
def test_bias_false(self):
|
|
cell = nn.LSTMCell(10, 20, bias=False)
|
|
self.assertFalse(hasattr(cell, 'bias_ih'))
|
|
self.assertFalse(hasattr(cell, 'bias_hh'))
|
|
|
|
# Verify forward pass works without bias
|
|
x = paddle.randn([4, 10])
|
|
h = paddle.randn([4, 20])
|
|
c = paddle.randn([4, 20])
|
|
y, (new_h, new_c) = cell(x, (h, c))
|
|
self.assertEqual(y.shape, [4, 20])
|
|
self.assertEqual(new_h.shape, [4, 20])
|
|
self.assertEqual(new_c.shape, [4, 20])
|
|
|
|
def test_bias_true(self):
|
|
cell = nn.LSTMCell(10, 20, bias=True)
|
|
self.assertTrue(hasattr(cell, 'bias_ih'))
|
|
self.assertTrue(hasattr(cell, 'bias_hh'))
|
|
self.assertFalse(cell.bias_ih.stop_gradient)
|
|
self.assertFalse(cell.bias_hh.stop_gradient)
|
|
|
|
def test_dtype(self):
|
|
cell = nn.LSTMCell(10, 20, dtype='float64')
|
|
self.assertEqual(cell.weight_ih.dtype, paddle.float64)
|
|
self.assertEqual(cell.weight_hh.dtype, paddle.float64)
|
|
self.assertEqual(cell.bias_ih.dtype, paddle.float64)
|
|
self.assertEqual(cell.bias_hh.dtype, paddle.float64)
|
|
|
|
x = paddle.randn([4, 10]).astype('float64')
|
|
h = paddle.randn([4, 20]).astype('float64')
|
|
c = paddle.randn([4, 20]).astype('float64')
|
|
y, (new_h, new_c) = cell(x, (h, c))
|
|
self.assertEqual(y.dtype, paddle.float64)
|
|
|
|
def test_device(self):
|
|
# Only test if gpu is available, otherwise cpu
|
|
device = 'gpu' if paddle.is_compiled_with_cuda() else 'cpu'
|
|
cell = nn.LSTMCell(10, 20, device=device)
|
|
|
|
# We can just check if it runs without error on the specified device
|
|
x = paddle.randn([4, 10])
|
|
h = paddle.randn([4, 20])
|
|
c = paddle.randn([4, 20])
|
|
if device == 'gpu':
|
|
x = x.cuda()
|
|
h = h.cuda()
|
|
c = c.cuda()
|
|
y, (new_h, new_c) = cell(x, (h, c))
|
|
|
|
# Also test explicit cpu on gpu machine if possible, but 'cpu' is always safe
|
|
cell_cpu = nn.LSTMCell(10, 20, device='cpu')
|
|
self.assertTrue(cell_cpu.weight_ih.place.is_cpu_place())
|
|
|
|
def test_keyword_only_args(self):
|
|
# weight_ih_attr is keyword-only
|
|
with self.assertRaises(TypeError):
|
|
nn.LSTMCell(10, 20, paddle.ParamAttr())
|
|
|
|
# This should work
|
|
nn.LSTMCell(10, 20, weight_ih_attr=paddle.ParamAttr())
|
|
|
|
|
|
if __name__ == '__main__':
|
|
unittest.main()
|