wehub-resource-sync
152 lines
5.0 KiB
Python
152 lines
5.0 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.
|
|
|
|
# Unit test for paddle.nn.functional.embedding
|
|
# Target: cover embedding related code paths
|
|
|
|
import unittest
|
|
|
|
import paddle
|
|
import paddle.nn.functional as F
|
|
from paddle import nn
|
|
|
|
|
|
class TestEmbedding(unittest.TestCase):
|
|
"""Test embedding function."""
|
|
|
|
def setUp(self):
|
|
paddle.disable_static()
|
|
|
|
def test_embedding_basic(self):
|
|
"""Basic embedding lookup."""
|
|
x = paddle.to_tensor([[1, 2, 3], [4, 5, 6]], dtype='int64')
|
|
w = paddle.randn([10, 32])
|
|
out = F.embedding(x, w)
|
|
self.assertEqual(out.shape, [2, 3, 32])
|
|
|
|
def test_embedding_with_padding_idx(self):
|
|
"""Embedding with padding_idx."""
|
|
x = paddle.to_tensor([[0, 1, 2], [0, 3, 4]], dtype='int64')
|
|
w = paddle.randn([10, 32])
|
|
out = F.embedding(x, w, padding_idx=0)
|
|
self.assertEqual(out.shape, [2, 3, 32])
|
|
|
|
def test_embedding_sparse_gradient(self):
|
|
"""Embedding with sparse gradient."""
|
|
x = paddle.to_tensor([[1, 2, 3]], dtype='int64')
|
|
w = paddle.randn([10, 32])
|
|
w.stop_gradient = False
|
|
out = F.embedding(x, w, sparse=True)
|
|
self.assertEqual(out.shape, [1, 3, 32])
|
|
|
|
def test_embedding_float16_weight(self):
|
|
"""Embedding with float16 weight."""
|
|
x = paddle.to_tensor([[1, 2, 3]], dtype='int64')
|
|
w = paddle.randn([10, 32], dtype='float16')
|
|
out = F.embedding(x, w)
|
|
self.assertEqual(out.dtype, paddle.float16)
|
|
|
|
def test_embedding_int32_input(self):
|
|
"""Embedding with int32 input."""
|
|
x = paddle.to_tensor([[1, 2, 3]], dtype='int32')
|
|
w = paddle.randn([10, 32])
|
|
out = F.embedding(x, w)
|
|
self.assertEqual(out.shape, [1, 3, 32])
|
|
|
|
def test_embedding_max_norm(self):
|
|
"""Embedding with max_norm."""
|
|
x = paddle.to_tensor([[1, 2, 3]], dtype='int64')
|
|
w = paddle.randn([10, 32]) * 10 # large values
|
|
out = F.embedding(x, w, max_norm=1.0)
|
|
self.assertEqual(out.shape, [1, 3, 32])
|
|
# Check that norms are bounded
|
|
norms = paddle.norm(out, p=2, axis=-1)
|
|
self.assertTrue(paddle.all(norms <= 1.0 + 1e-5))
|
|
|
|
|
|
class TestOneHot(unittest.TestCase):
|
|
"""Test one_hot function.
|
|
F.one_hot(x, num_classes) - no dtype parameter.
|
|
"""
|
|
|
|
def setUp(self):
|
|
paddle.disable_static()
|
|
|
|
def test_one_hot_basic(self):
|
|
"""Basic one_hot encoding."""
|
|
x = paddle.to_tensor([0, 1, 2, 3], dtype='int64')
|
|
out = F.one_hot(x, num_classes=5)
|
|
self.assertEqual(out.shape, [4, 5])
|
|
# Check one-hot encoding
|
|
result = out.numpy()
|
|
for i in range(4):
|
|
self.assertEqual(result[i, i], 1)
|
|
# All other positions should be 0
|
|
for j in range(5):
|
|
if j != i:
|
|
self.assertEqual(result[i, j], 0)
|
|
|
|
def test_one_hot_int32(self):
|
|
"""One_hot with int32 input."""
|
|
x = paddle.to_tensor([0, 1], dtype='int32')
|
|
out = F.one_hot(x, num_classes=3)
|
|
self.assertEqual(out.shape, [2, 3])
|
|
|
|
def test_one_hot_2d(self):
|
|
"""One_hot with 2D input."""
|
|
x = paddle.to_tensor([[0, 1], [2, 0]], dtype='int64')
|
|
out = F.one_hot(x, num_classes=4)
|
|
self.assertEqual(out.shape, [2, 2, 4])
|
|
|
|
def test_one_hot_output_dtype(self):
|
|
"""One_hot default dtype is float32."""
|
|
x = paddle.to_tensor([0, 1], dtype='int64')
|
|
out = F.one_hot(x, num_classes=3)
|
|
# Default output dtype
|
|
self.assertEqual(out.shape, [2, 3])
|
|
|
|
|
|
class TestEmbeddingLayer(unittest.TestCase):
|
|
"""Test nn.Embedding layer."""
|
|
|
|
def setUp(self):
|
|
paddle.disable_static()
|
|
|
|
def test_embedding_layer_basic(self):
|
|
"""nn.Embedding basic usage."""
|
|
layer = nn.Embedding(100, 32)
|
|
x = paddle.to_tensor([[1, 2, 3]], dtype='int64')
|
|
out = layer(x)
|
|
self.assertEqual(out.shape, [1, 3, 32])
|
|
|
|
def test_embedding_layer_with_padding_idx(self):
|
|
"""nn.Embedding with padding_idx."""
|
|
layer = nn.Embedding(100, 32, padding_idx=0)
|
|
x = paddle.to_tensor([[0, 1, 2]], dtype='int64')
|
|
out = layer(x)
|
|
self.assertEqual(out.shape, [1, 3, 32])
|
|
|
|
def test_embedding_layer_sparse(self):
|
|
"""nn.Embedding with sparse gradient."""
|
|
layer = nn.Embedding(100, 32, sparse=True)
|
|
x = paddle.to_tensor([[1, 2, 3]], dtype='int64')
|
|
out = layer(x)
|
|
loss = out.mean()
|
|
loss.backward()
|
|
# Should not crash
|
|
|
|
|
|
if __name__ == '__main__':
|
|
unittest.main()
|