wehub-resource-sync
428 lines
14 KiB
Python
428 lines
14 KiB
Python
# Copyright (c) 2020 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 numpy as np
|
|
from op_test import (
|
|
OpTest,
|
|
convert_float_to_uint16,
|
|
get_device_place,
|
|
get_devices,
|
|
is_custom_device,
|
|
)
|
|
|
|
import paddle
|
|
import paddle.nn.functional as F
|
|
from paddle.base import core
|
|
|
|
np.random.seed(10)
|
|
|
|
|
|
def ref_log_softmax(x):
|
|
shiftx = x - np.max(x)
|
|
out = shiftx - np.log(np.exp(shiftx).sum())
|
|
return out
|
|
|
|
|
|
def ref_log_softmax_grad(x, axis):
|
|
if axis < 0:
|
|
axis += len(x.shape)
|
|
out = np.apply_along_axis(ref_log_softmax, axis, x)
|
|
axis_dim = x.shape[axis]
|
|
dout = np.full_like(x, fill_value=1.0 / x.size)
|
|
dx = dout - np.exp(out) * dout.copy().sum(axis=axis, keepdims=True).repeat(
|
|
axis_dim, axis=axis
|
|
)
|
|
return dx
|
|
|
|
|
|
class TestLogSoftmaxOp(OpTest):
|
|
def setUp(self):
|
|
self.op_type = 'log_softmax'
|
|
self.prim_op_type = "comp"
|
|
self.python_api = F.log_softmax
|
|
self.public_python_api = F.log_softmax
|
|
self.dtype = 'float64'
|
|
self.shape = [2, 3, 4, 5]
|
|
self.axis = -1
|
|
self.set_attrs()
|
|
|
|
x = np.random.uniform(0.1, 1.0, self.shape).astype(self.dtype)
|
|
out = np.apply_along_axis(ref_log_softmax, self.axis, x)
|
|
self.x_grad = ref_log_softmax_grad(x, self.axis)
|
|
|
|
self.inputs = {'X': x}
|
|
self.outputs = {'Out': out}
|
|
self.attrs = {'axis': self.axis}
|
|
|
|
def set_attrs(self):
|
|
pass
|
|
|
|
def test_check_output(self):
|
|
self.check_output(check_pir=True, check_prim_pir=True)
|
|
|
|
def test_check_grad(self):
|
|
self.check_grad(
|
|
['X'], ['Out'], user_defined_grads=[self.x_grad], check_pir=True
|
|
)
|
|
|
|
|
|
class TestLogSoftmaxOp_ZeroDim(TestLogSoftmaxOp):
|
|
def setUp(self):
|
|
self.op_type = 'log_softmax'
|
|
self.prim_op_type = "comp"
|
|
self.python_api = F.log_softmax
|
|
self.public_python_api = F.log_softmax
|
|
self.dtype = 'float64'
|
|
|
|
x = np.random.uniform(0.1, 1.0, []).astype(self.dtype)
|
|
out = np.array(0.0).astype(self.dtype)
|
|
|
|
self.inputs = {'X': x}
|
|
self.outputs = {'Out': out}
|
|
self.attrs = {'axis': -1}
|
|
|
|
def test_check_output(self):
|
|
self.check_output(check_pir=True, check_prim_pir=True)
|
|
|
|
def test_check_grad(self):
|
|
self.check_grad(['X'], ['Out'], check_pir=True)
|
|
|
|
|
|
class TestLogSoftmaxShape(TestLogSoftmaxOp):
|
|
def set_attrs(self):
|
|
self.shape = [12, 10]
|
|
|
|
|
|
class TestLogSoftmaxAxis(TestLogSoftmaxOp):
|
|
def set_attrs(self):
|
|
self.axis = 1
|
|
|
|
|
|
class TestLogSoftmaxFP16OP(TestLogSoftmaxOp):
|
|
def set_attrs(self):
|
|
self.dtype = np.float16
|
|
|
|
def test_check_output(self):
|
|
self.check_output(atol=1e-3, check_pir=True, check_prim_pir=True)
|
|
|
|
def test_check_grad(self):
|
|
self.check_grad(['X'], ['Out'], max_relative_error=1e-2, check_pir=True)
|
|
|
|
|
|
class TestLogSoftmaxShapeFP16OP(TestLogSoftmaxFP16OP):
|
|
def set_attrs(self):
|
|
self.dtype = np.float16
|
|
self.shape = [12, 10]
|
|
|
|
|
|
class TestLogSoftmaxAxisFP16OP(TestLogSoftmaxFP16OP):
|
|
def set_attrs(self):
|
|
self.dtype = np.float16
|
|
self.axis = 1
|
|
|
|
|
|
@unittest.skipIf(
|
|
not (core.is_compiled_with_cuda() or is_custom_device()),
|
|
"core is not compiled with CUDA",
|
|
)
|
|
class TestLogSoftmaxBF16Op(OpTest):
|
|
def setUp(self):
|
|
self.op_type = 'log_softmax'
|
|
self.prim_op_type = "comp"
|
|
self.python_api = F.log_softmax
|
|
self.public_python_api = F.log_softmax
|
|
self.dtype = np.uint16
|
|
self.shape = [2, 3, 4, 5]
|
|
self.axis = -1
|
|
|
|
x = np.random.uniform(0.1, 1.0, self.shape).astype(np.float32)
|
|
out = np.apply_along_axis(ref_log_softmax, self.axis, x)
|
|
self.x_grad = ref_log_softmax_grad(x, self.axis)
|
|
|
|
self.inputs = {'X': convert_float_to_uint16(x)}
|
|
self.outputs = {'Out': convert_float_to_uint16(out)}
|
|
self.attrs = {'axis': self.axis}
|
|
|
|
def test_check_output(self):
|
|
place = get_device_place()
|
|
self.check_output_with_place(place, check_pir=True, check_prim_pir=True)
|
|
|
|
def test_check_grad(self):
|
|
place = get_device_place()
|
|
self.check_grad_with_place(
|
|
place,
|
|
['X'],
|
|
['Out'],
|
|
user_defined_grads=[self.x_grad],
|
|
check_pir=True,
|
|
)
|
|
|
|
|
|
class TestLogSoftmaxLargeDimFP16OP(TestLogSoftmaxOp):
|
|
def set_attrs(self):
|
|
self.dtype = np.float16
|
|
self.shape = [16, 100000]
|
|
|
|
|
|
class TestNNLogSoftmaxAPI(unittest.TestCase):
|
|
def setUp(self):
|
|
self.x_shape = [2, 3, 4, 5]
|
|
self.x = np.random.uniform(-1.0, 1.0, self.x_shape).astype(np.float32)
|
|
self.place = get_device_place()
|
|
|
|
def check_api(self, axis=-1):
|
|
ref_out = np.apply_along_axis(ref_log_softmax, axis, self.x)
|
|
|
|
logsoftmax = paddle.nn.LogSoftmax(axis)
|
|
# test static api
|
|
with paddle.static.program_guard(paddle.static.Program()):
|
|
x = paddle.static.data(name='x', shape=self.x_shape)
|
|
y = logsoftmax(x)
|
|
exe = paddle.static.Executor(self.place)
|
|
out = exe.run(feed={'x': self.x}, fetch_list=[y])
|
|
np.testing.assert_allclose(out[0], ref_out, rtol=1e-05)
|
|
|
|
# test dygraph api
|
|
paddle.disable_static()
|
|
x = paddle.to_tensor(self.x)
|
|
y = logsoftmax(x)
|
|
np.testing.assert_allclose(y.numpy(), ref_out, rtol=1e-05)
|
|
paddle.enable_static()
|
|
|
|
def test_check_api(self):
|
|
for axis in [-1, 1]:
|
|
self.check_api(axis)
|
|
|
|
|
|
class TestNNFunctionalLogSoftmaxAPI(unittest.TestCase):
|
|
def setUp(self):
|
|
self.x_shape = [2, 3, 4, 5]
|
|
self.x = np.random.uniform(-1, 1, self.x_shape).astype(np.float32)
|
|
self.place = get_device_place()
|
|
|
|
def check_api(self, axis=-1, dtype=None):
|
|
x = self.x.copy()
|
|
if dtype is not None:
|
|
x = x.astype(dtype)
|
|
ref_out = np.apply_along_axis(ref_log_softmax, axis, x)
|
|
with paddle.static.program_guard(paddle.static.Program()):
|
|
x = paddle.static.data(name='x', shape=self.x_shape)
|
|
y = F.log_softmax(x, axis, dtype)
|
|
exe = paddle.static.Executor(self.place)
|
|
out = exe.run(feed={'x': self.x}, fetch_list=[y])
|
|
np.testing.assert_allclose(out[0], ref_out, rtol=1e-05)
|
|
|
|
paddle.disable_static()
|
|
x = paddle.to_tensor(self.x)
|
|
y = F.log_softmax(x, axis, dtype)
|
|
np.testing.assert_allclose(y.numpy(), ref_out, rtol=1e-05)
|
|
paddle.enable_static()
|
|
|
|
def test_check_api(self):
|
|
for axis in [-1, 1]:
|
|
self.check_api(axis)
|
|
self.check_api(-1, 'float64')
|
|
|
|
def test_errors(self):
|
|
with paddle.static.program_guard(paddle.static.Program()):
|
|
x = paddle.static.data(name='X1', shape=[100], dtype='int32')
|
|
self.assertRaises(TypeError, F.log_softmax, x)
|
|
|
|
x = paddle.static.data(name='X2', shape=[100], dtype='float32')
|
|
self.assertRaises(TypeError, F.log_softmax, x, dtype='int32')
|
|
|
|
|
|
def _check_cuda_memory_20GB():
|
|
if not hasattr(paddle.device.cuda, 'get_device_properties'):
|
|
return False
|
|
gpu_info = paddle.device.get_device_properties(get_devices()[0])
|
|
return gpu_info.total_memory >= 20 * (1024**3) # 20GB
|
|
|
|
|
|
@unittest.skipIf(
|
|
not (core.is_compiled_with_cuda() or is_custom_device())
|
|
or not _check_cuda_memory_20GB(),
|
|
"Need CUDA support and at least 20GB GPU memory",
|
|
)
|
|
class TestLogSoftmaxLargeOp(unittest.TestCase):
|
|
def test_check_run(self):
|
|
x = paddle.randn([4, 4096, 131072 + 2048]) # 8GB+4*4096*2048
|
|
paddle.nn.functional.log_softmax(x, axis=-1)
|
|
|
|
|
|
class TestLogSoftmaxOp_ZeroSize(OpTest):
|
|
def setUp(self):
|
|
self.op_type = 'log_softmax'
|
|
self.python_api = F.log_softmax
|
|
self.public_python_api = F.log_softmax
|
|
self.dtype = 'float64'
|
|
self.shape = [2, 0, 4, 5]
|
|
self.axis = -1
|
|
self.set_attrs()
|
|
|
|
x = np.random.uniform(0.1, 1.0, self.shape).astype(self.dtype)
|
|
# shape is same as x, size is 0.
|
|
out = np.random.random(self.shape).astype(self.dtype)
|
|
|
|
self.inputs = {'X': x}
|
|
self.outputs = {'Out': out}
|
|
self.attrs = {'axis': self.axis}
|
|
|
|
def set_attrs(self):
|
|
pass
|
|
|
|
def test_check_output(self):
|
|
self.check_output(check_pir=True)
|
|
|
|
def test_check_grad(self):
|
|
self.check_grad(['X'], ['Out'], check_pir=True)
|
|
|
|
|
|
class TestLogSoftmaxParamAlias(unittest.TestCase):
|
|
"""Test parameter aliases: input=x, dim=axis."""
|
|
|
|
def setUp(self):
|
|
paddle.disable_static()
|
|
self.x_np = np.random.uniform(0.1, 1.0, [3, 4]).astype('float32')
|
|
self.x3d_np = np.random.uniform(0.1, 1.0, [2, 3, 4]).astype('float32')
|
|
|
|
def tearDown(self):
|
|
paddle.enable_static()
|
|
|
|
def _ref(self, x_np, axis):
|
|
return np.apply_along_axis(ref_log_softmax, axis, x_np)
|
|
|
|
# --- `input` alias for `x` ---
|
|
|
|
def test_input_alias_keyword(self):
|
|
x = paddle.to_tensor(self.x_np)
|
|
expected = F.log_softmax(x, axis=-1).numpy()
|
|
result = F.log_softmax(input=x, axis=-1).numpy()
|
|
np.testing.assert_allclose(result, expected, rtol=1e-6)
|
|
|
|
def test_input_alias_with_axis(self):
|
|
x = paddle.to_tensor(self.x_np)
|
|
expected = F.log_softmax(x, axis=0).numpy()
|
|
result = F.log_softmax(input=x, axis=0).numpy()
|
|
np.testing.assert_allclose(result, expected, rtol=1e-6)
|
|
|
|
# --- `dim` alias for `axis` ---
|
|
|
|
def test_dim_alias_keyword(self):
|
|
x = paddle.to_tensor(self.x_np)
|
|
expected = F.log_softmax(x, axis=1).numpy()
|
|
result = F.log_softmax(x, dim=1).numpy()
|
|
np.testing.assert_allclose(result, expected, rtol=1e-6)
|
|
|
|
def test_dim_alias_negative(self):
|
|
x = paddle.to_tensor(self.x3d_np)
|
|
expected = F.log_softmax(x, axis=-2).numpy()
|
|
result = F.log_softmax(x, dim=-2).numpy()
|
|
np.testing.assert_allclose(result, expected, rtol=1e-6)
|
|
|
|
# --- Both aliases together ---
|
|
|
|
def test_both_aliases(self):
|
|
x = paddle.to_tensor(self.x_np)
|
|
expected = F.log_softmax(x, axis=1).numpy()
|
|
result = F.log_softmax(input=x, dim=1).numpy()
|
|
np.testing.assert_allclose(result, expected, rtol=1e-6)
|
|
|
|
def test_both_aliases_with_dtype(self):
|
|
x = paddle.to_tensor(self.x_np)
|
|
expected = F.log_softmax(x, axis=0, dtype='float64').numpy()
|
|
result = F.log_softmax(input=x, dim=0, dtype='float64').numpy()
|
|
np.testing.assert_allclose(result, expected, rtol=1e-10)
|
|
self.assertEqual(result.dtype, np.float64)
|
|
|
|
# --- 3D inputs ---
|
|
|
|
def test_3d_input_alias_dim0(self):
|
|
x = paddle.to_tensor(self.x3d_np)
|
|
expected = F.log_softmax(x, axis=0).numpy()
|
|
result = F.log_softmax(input=x, dim=0).numpy()
|
|
np.testing.assert_allclose(result, expected, rtol=1e-6)
|
|
|
|
def test_3d_input_alias_dim1(self):
|
|
x = paddle.to_tensor(self.x3d_np)
|
|
expected = F.log_softmax(x, axis=1).numpy()
|
|
result = F.log_softmax(input=x, dim=1).numpy()
|
|
np.testing.assert_allclose(result, expected, rtol=1e-6)
|
|
|
|
def test_3d_input_alias_dim_neg1(self):
|
|
x = paddle.to_tensor(self.x3d_np)
|
|
expected = F.log_softmax(x, axis=-1).numpy()
|
|
result = F.log_softmax(input=x, dim=-1).numpy()
|
|
np.testing.assert_allclose(result, expected, rtol=1e-6)
|
|
|
|
# --- float64 input ---
|
|
|
|
def test_float64_input_alias(self):
|
|
x_np = self.x_np.astype('float64')
|
|
x = paddle.to_tensor(x_np)
|
|
expected = F.log_softmax(x, axis=1).numpy()
|
|
result = F.log_softmax(input=x, dim=1).numpy()
|
|
np.testing.assert_allclose(result, expected, rtol=1e-10)
|
|
|
|
# --- Conflict error ---
|
|
|
|
def test_conflict_x_and_input_raises(self):
|
|
x = paddle.to_tensor(self.x_np)
|
|
with self.assertRaises(ValueError):
|
|
F.log_softmax(x=x, input=x)
|
|
|
|
def test_conflict_axis_and_dim_raises(self):
|
|
x = paddle.to_tensor(self.x_np)
|
|
with self.assertRaises(ValueError):
|
|
F.log_softmax(x, axis=0, dim=1)
|
|
|
|
|
|
class TestLogSoftmaxOutParam(unittest.TestCase):
|
|
"""Test out parameter for F.log_softmax."""
|
|
|
|
def setUp(self):
|
|
paddle.disable_static()
|
|
self.x_np = np.random.uniform(0.1, 1.0, [3, 4]).astype('float32')
|
|
|
|
def tearDown(self):
|
|
paddle.enable_static()
|
|
|
|
def test_out_param(self):
|
|
x = paddle.to_tensor(self.x_np)
|
|
expected = F.log_softmax(x, axis=-1)
|
|
out = paddle.empty_like(x)
|
|
result = F.log_softmax(x, axis=-1, out=out)
|
|
np.testing.assert_allclose(out.numpy(), expected.numpy(), rtol=1e-6)
|
|
np.testing.assert_allclose(result.numpy(), expected.numpy(), rtol=1e-6)
|
|
|
|
def test_out_param_with_dim_alias(self):
|
|
x = paddle.to_tensor(self.x_np)
|
|
expected = F.log_softmax(x, axis=0)
|
|
out = paddle.empty_like(x)
|
|
F.log_softmax(x, dim=0, out=out)
|
|
np.testing.assert_allclose(out.numpy(), expected.numpy(), rtol=1e-6)
|
|
|
|
def test_out_param_with_dtype(self):
|
|
x = paddle.to_tensor(self.x_np)
|
|
out = paddle.empty([3, 4], dtype='float64')
|
|
F.log_softmax(x, axis=-1, dtype='float64', out=out)
|
|
self.assertEqual(out.dtype, paddle.float64)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
paddle.enable_static()
|
|
unittest.main()
|