wehub-resource-sync
263 lines
8.1 KiB
Python
263 lines
8.1 KiB
Python
# Copyright (c) 2022 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 copy
|
|
import math
|
|
import os
|
|
import re
|
|
import unittest
|
|
|
|
import numpy as np
|
|
from op_test import is_custom_device
|
|
|
|
import paddle
|
|
import paddle.sparse
|
|
from paddle.base import core
|
|
from paddle.base.framework import in_pir_mode
|
|
|
|
|
|
def get_cuda_version():
|
|
result = os.popen("nvcc --version").read()
|
|
regex = r'release (\S+),'
|
|
match = re.search(regex, result)
|
|
if match:
|
|
num = str(match.group(1))
|
|
integer, decimal = num.split('.')
|
|
return int(integer) * 1000 + int(float(decimal) * 10)
|
|
else:
|
|
return -1
|
|
|
|
|
|
@unittest.skipIf(
|
|
not (core.is_compiled_with_cuda() or is_custom_device())
|
|
or get_cuda_version() < 11080,
|
|
"core is not compiled with CUDA and cuda version need larger than or equal to 11.8",
|
|
)
|
|
class TestSparseAttentionAPI1(unittest.TestCase):
|
|
def setUp(self):
|
|
paddle.seed(0)
|
|
self.batch_size = 16
|
|
self.num_heads = 16
|
|
self.seq_len = 128
|
|
self.head_dim = 16
|
|
self.dtype = 'float64'
|
|
self.use_mask = True
|
|
|
|
def test_dygraph(self):
|
|
self.shape = [
|
|
self.batch_size,
|
|
self.num_heads,
|
|
self.seq_len,
|
|
self.head_dim,
|
|
]
|
|
query = paddle.rand(self.shape, self.dtype)
|
|
key = paddle.rand(self.shape, self.dtype)
|
|
value = paddle.rand(self.shape, self.dtype)
|
|
|
|
query.stop_gradient = False
|
|
key.stop_gradient = False
|
|
value.stop_gradient = False
|
|
|
|
mask = paddle.nn.functional.dropout(
|
|
paddle.ones([self.seq_len, self.seq_len]),
|
|
mode='downscale_in_infer',
|
|
)
|
|
mask = mask.expand(
|
|
[self.batch_size, self.num_heads, self.seq_len, self.seq_len]
|
|
)
|
|
sp_mask = mask.reshape([-1, self.seq_len, self.seq_len]).to_sparse_csr()
|
|
|
|
query_sp = copy.deepcopy(query)
|
|
key_sp = copy.deepcopy(key)
|
|
value_sp = copy.deepcopy(value)
|
|
|
|
query_sp.stop_gradient = False
|
|
key_sp.stop_gradient = False
|
|
value_sp.stop_gradient = False
|
|
|
|
if self.use_mask:
|
|
kp_mask = paddle.randint(
|
|
0, 2, [self.batch_size, self.seq_len]
|
|
).astype(self.dtype)
|
|
attn_mask = paddle.randint(
|
|
0, 2, [self.seq_len, self.seq_len]
|
|
).astype(self.dtype)
|
|
|
|
sdd = paddle.matmul(query, key, False, True) / math.sqrt(
|
|
float(self.head_dim)
|
|
)
|
|
sdd = (
|
|
sdd
|
|
+ ((mask * kp_mask.unsqueeze([1, 2]) * attn_mask) - 1.0) * 1e9
|
|
)
|
|
softmax = paddle.nn.functional.softmax(sdd)
|
|
output = paddle.matmul(softmax, value)
|
|
output.backward()
|
|
|
|
output_sp = paddle.sparse.nn.functional.attention(
|
|
query_sp, key_sp, value_sp, sp_mask, kp_mask, attn_mask
|
|
)
|
|
output_sp.backward()
|
|
else:
|
|
sdd = paddle.matmul(query, key, False, True) / math.sqrt(
|
|
float(self.head_dim)
|
|
)
|
|
sdd = sdd + (mask - 1.0) * 1e9
|
|
softmax = paddle.nn.functional.softmax(sdd)
|
|
output = paddle.matmul(softmax, value)
|
|
output.backward()
|
|
|
|
output_sp = paddle.sparse.nn.functional.attention(
|
|
query_sp, key_sp, value_sp, sp_mask
|
|
)
|
|
output_sp.backward()
|
|
|
|
np.testing.assert_allclose(
|
|
output_sp.numpy(), output.numpy(), rtol=1e-05
|
|
)
|
|
np.testing.assert_allclose(
|
|
query_sp.grad.numpy(), query.grad.numpy(), rtol=1e-05
|
|
)
|
|
np.testing.assert_allclose(
|
|
key_sp.grad.numpy(), key.grad.numpy(), rtol=1e-05
|
|
)
|
|
np.testing.assert_allclose(
|
|
value_sp.grad.numpy(), value.grad.numpy(), rtol=1e-05
|
|
)
|
|
|
|
|
|
class TestSparseAttentionAPI2(TestSparseAttentionAPI1):
|
|
def setUp(self):
|
|
super().setUp()
|
|
self.batch_size = 16
|
|
self.num_heads = 16
|
|
self.seq_len = 128
|
|
self.head_dim = 32
|
|
self.dtype = 'float64'
|
|
self.use_mask = False
|
|
|
|
|
|
class TestSparseAttentionAPI3(TestSparseAttentionAPI1):
|
|
def setUp(self):
|
|
super().setUp()
|
|
self.batch_size = 16
|
|
self.num_heads = 16
|
|
self.seq_len = 512
|
|
self.head_dim = 16
|
|
self.dtype = 'float64'
|
|
self.use_mask = True
|
|
|
|
|
|
class TestSparseAttentionAPI4(TestSparseAttentionAPI1):
|
|
def setUp(self):
|
|
super().setUp()
|
|
self.batch_size = 16
|
|
self.num_heads = 16
|
|
self.seq_len = 512
|
|
self.head_dim = 32
|
|
self.dtype = 'float64'
|
|
self.use_mask = False
|
|
|
|
|
|
class TestSparseAttentionAPI5(TestSparseAttentionAPI1):
|
|
def setUp(self):
|
|
super().setUp()
|
|
self.batch_size = 16
|
|
self.num_heads = 16
|
|
self.seq_len = 512
|
|
self.head_dim = 64
|
|
self.dtype = 'float64'
|
|
self.use_mask = True
|
|
|
|
|
|
devices = []
|
|
if paddle.device.get_device() != "cpu":
|
|
devices.append(paddle.device.get_device())
|
|
else:
|
|
devices.append('cpu')
|
|
|
|
|
|
class TestSparseSoftmaxStaticAPI(unittest.TestCase):
|
|
'''
|
|
Test the API paddle.sparse.nn.functional.softmax on some sparse tensors in pir mode in static graph.
|
|
'''
|
|
|
|
def check_result_coo(self, x_shape):
|
|
'''
|
|
x_shape: a tensor shape,
|
|
generate a sparse tensor with shape "x_shape" and compute the output of paddle.sparse.nn.functional.softmax.
|
|
compare the output of paddle.sparse.nn.functional.softmax and the output of paddle.nn.functional.Softmax.
|
|
'''
|
|
for device in devices:
|
|
paddle.device.set_device(device)
|
|
x = paddle.rand(x_shape, dtype='float32')
|
|
indices_data, values_data = (
|
|
x.detach().to_sparse_coo(sparse_dim=len(x_shape)).indices(),
|
|
x.detach().to_sparse_coo(sparse_dim=len(x_shape)).values(),
|
|
)
|
|
|
|
x.stop_gradient = False
|
|
out = paddle.nn.functional.softmax(x)
|
|
|
|
paddle.enable_static()
|
|
with paddle.static.program_guard(
|
|
paddle.static.Program(), paddle.static.Program()
|
|
):
|
|
indices = paddle.static.data(
|
|
name="indices",
|
|
shape=indices_data.shape,
|
|
dtype=indices_data.dtype,
|
|
)
|
|
values = paddle.static.data(
|
|
name="values",
|
|
shape=values_data.shape,
|
|
dtype=values_data.dtype,
|
|
)
|
|
|
|
sp_x = paddle.sparse.sparse_coo_tensor(
|
|
indices,
|
|
values,
|
|
shape=x.shape,
|
|
dtype=x.dtype,
|
|
)
|
|
sp_out = paddle.sparse.nn.functional.softmax(sp_x)
|
|
sp_dense_out = sp_out.to_dense()
|
|
|
|
sp_exe = paddle.static.Executor()
|
|
sp_fetch = sp_exe.run(
|
|
feed={
|
|
"indices": indices_data.numpy(),
|
|
"values": values_data.numpy(),
|
|
},
|
|
fetch_list=[sp_dense_out],
|
|
return_numpy=True,
|
|
)
|
|
np.testing.assert_allclose(out.numpy(), sp_fetch[0], rtol=1e-05)
|
|
paddle.disable_static()
|
|
|
|
def test_softmax_2d(self):
|
|
if in_pir_mode():
|
|
self.check_result_coo([3, 4])
|
|
|
|
def test_softmax_3d(self):
|
|
if in_pir_mode():
|
|
self.check_result_coo([3, 4, 5])
|
|
|
|
def test_softmax_4d(self):
|
|
if in_pir_mode():
|
|
self.check_result_coo([3, 4, 5, 6])
|
|
|
|
|
|
if __name__ == '__main__':
|
|
unittest.main()
|