mindspore/tests/vm_impl/nn_ops_vm_impl.py

377 lines
10 KiB
Python
Raw Normal View History

# Copyright 2021 Huawei Technologies Co., Ltd
#
# 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.
# ============================================================================
"""Generate vm_impl function for nn ops"""
import numpy as np
2020-05-13 11:30:27 +08:00
from mindspore.common.tensor import Tensor
from mindspore.ops import operations as P
from mindspore.ops.operations import _grad_ops as G
from mindspore.ops.vm_impl_registry import vm_impl_registry as vm_impl_getters
from .vm_interface import vm
2020-04-02 11:58:45 +08:00
# pylint: disable=unused-argument
@vm_impl_getters.register(P.ScalarSummary)
def vm_impl_scalar_summary(self):
"""Generate vm_impl function for ScalarSummary"""
2020-04-02 11:58:45 +08:00
def vm_impl(string_in, scalar):
"""Implement by vm mode."""
return scalar
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(P.ReLU)
def vm_impl_relu(self):
"""Generate vm_impl function for ReLU"""
2020-04-02 11:58:45 +08:00
def vm_impl(x):
x = x.asnumpy()
output = Tensor(vm.relu(x))
return output
2020-04-02 11:58:45 +08:00
return vm_impl
2020-04-02 11:58:45 +08:00
@vm_impl_getters.register(P.Flatten)
def vm_impl_flatten(self):
"""Generate vm_impl function for Flatten"""
2020-04-02 11:58:45 +08:00
def vm_impl(x):
x = x.asnumpy()
return Tensor(vm.flatten_batch(x))
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(P.Softmax)
def vm_impl_softmax(self):
"""Generate vm_impl function for Softmax"""
2020-04-02 11:58:45 +08:00
def vm_impl(x):
x = x.asnumpy()
return Tensor(vm.softmax(x))
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(P.LogSoftmax)
def vm_impl_log_softmax(self):
"""Generate vm_impl function for LogSoftmax"""
2020-04-02 11:58:45 +08:00
def vm_impl(x):
x = x.asnumpy()
return Tensor(vm.logsoftmax(x))
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(P.Tanh)
def vm_impl_tanh(self):
"""Generate vm_impl function for Tanh"""
2020-04-02 11:58:45 +08:00
def vm_impl(x):
x = x.asnumpy()
return Tensor(vm.tanh(x))
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(P.BatchNorm)
def vm_impl_batch_norm(self):
"""Generate vm_impl function for BatchNorm"""
2020-04-02 11:58:45 +08:00
def vm_impl(x, scale, b, mean, variance):
# pylint: disable=unused-argument
x = x.asnumpy()
scale = scale.asnumpy()
b = b.asnumpy()
mean = mean.asnumpy()
variance = variance.asnumpy()
out, x_mean, x_var, running_mean, running_var = vm.batch_norm(x, scale, b, mean, \
variance, \
eps=self.epsilon)
return Tensor(out), Tensor(x_mean), Tensor(x_var), \
Tensor(running_mean), Tensor(running_var)
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(P.Conv2D)
def vm_impl_conv2d(self):
"""Generate vm_impl function for Conv2D"""
2020-04-02 11:58:45 +08:00
def vm_impl(x, w):
x = x.asnumpy()
weight = w.asnumpy()
bias = None
out = vm.conv2d(x, weight, bias, self.stride, self.pad, self.dilation)
return Tensor(out)
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(G.MaxPoolGradWithArgmax)
def vm_impl_max_pool_grad_with_argmax(self):
"""Generate vm_impl function for MaxPoolGradWithArgmax"""
2020-04-02 11:58:45 +08:00
def vm_impl(x, dout, argmax):
x = x.asnumpy()
dout = dout.asnumpy()
arg_max = argmax.asnumpy()
2020-04-02 11:58:45 +08:00
dx = vm.max_pool_grad_with_argmax(x, dout, arg_max,
self.kernel_size[1], self.kernel_size[2], self.strides[1])
return Tensor(dx)
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(P.MaxPoolWithArgmax)
def vm_impl_max_pool_with_argmax(self):
"""Generate vm_impl function for MaxPoolWithArgmax"""
2020-04-02 11:58:45 +08:00
def vm_impl(x):
x = x.asnumpy()
out, out_argmax = vm.max_pool_with_argmax(x, self.kernel_size[1], self.kernel_size[2], self.strides[1])
return Tensor(out), Tensor(out_argmax)
2020-04-02 11:58:45 +08:00
return vm_impl
2020-04-02 11:58:45 +08:00
@vm_impl_getters.register(P.MaxPool)
def vm_impl_max_pool(self):
"""Generate vm_impl function for MaxPool"""
2020-04-02 11:58:45 +08:00
def vm_impl(x):
x = x.asnumpy()
out = vm.max_pooling(x, self.kernel_size[-2], self.kernel_size[-1], self.strides[-2])
return Tensor(out)
2020-04-02 11:58:45 +08:00
return vm_impl
2020-04-02 11:58:45 +08:00
@vm_impl_getters.register(G.MaxPoolGrad)
def vm_impl_max_pool_grad(self):
"""Generate vm_impl function for MaxPoolGrad"""
2020-04-02 11:58:45 +08:00
def vm_impl(x, out, dout):
x = x.asnumpy()
dout = dout.asnumpy()
out = vm.max_pool_grad(x, dout, self.kernel_size[-2], self.kernel_size[-1], self.strides[-2])
return Tensor(out)
2020-04-02 11:58:45 +08:00
return vm_impl
2020-04-02 11:58:45 +08:00
@vm_impl_getters.register(P.AvgPool)
2020-04-02 11:58:45 +08:00
def vm_impl_avg_pool(self):
"""Generate vm_impl function for AvgPool"""
2020-04-02 11:58:45 +08:00
def vm_impl(x):
x = x.asnumpy()
out = vm.avg_pooling(x, self.kernel_size[-2], self.kernel_size[-1], self.strides[-2])
return Tensor(out)
2020-04-02 11:58:45 +08:00
return vm_impl
2020-04-02 11:58:45 +08:00
@vm_impl_getters.register(G.AvgPoolGrad)
def vm_impl_avg_pool_grad(self):
"""Generate vm_impl function for AvgPoolGrad"""
2020-04-02 11:58:45 +08:00
def vm_impl(dout, origin_shape):
dout = dout.asnumpy()
out = vm.avg_pool_grad(dout, origin_shape, self.kernel_size[-2], self.kernel_size[-1], self.strides[-2])
return Tensor(out)
2020-04-02 11:58:45 +08:00
return vm_impl
2020-05-29 20:23:05 +08:00
# pylint: disable=function-redefined
@vm_impl_getters.register(G.BatchNormGrad)
def vm_impl_fused_batch_norm_grad(self):
"""Generate vm_impl function for BatchNormGrad"""
2020-04-02 11:58:45 +08:00
def vm_impl(dy, x, scale, save_mean, save_inv_variance):
dy = dy.asnumpy()
x = x.asnumpy()
scale = scale.asnumpy()
save_mean = save_mean.asnumpy()
save_inv_variance = save_inv_variance.asnumpy()
dx, dscale, dshift = vm.batch_norm_grad(dy, x, scale, save_mean, save_inv_variance)
return (Tensor(dx), Tensor(dscale), Tensor(dshift))
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(G.ReluGrad)
def vm_impl_relu_grad(self):
"""Generate vm_impl function for ReluGrad"""
2020-04-02 11:58:45 +08:00
def vm_impl(y_backprop, x):
x = x.asnumpy()
y_backprop = y_backprop.asnumpy()
2020-04-02 11:58:45 +08:00
y_backprop = vm.relu_grad(x.copy()) * y_backprop
return Tensor(y_backprop)
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(P.Conv2DBackpropInput)
def vm_impl_conv2d_backprop_input(self):
"""Generate vm_impl function for Conv2DBackpropInput"""
2020-04-02 11:58:45 +08:00
def vm_impl(dout, w, x_size):
dout = dout.asnumpy()
w = w.asnumpy()
dx = vm.conv2d_backprop_input(dout, x_size, w, self.stride, self.pad)
return Tensor(dx)
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(G.Conv2DBackpropFilter)
def vm_impl_conv2d_backprop_filter(self):
"""Generate vm_impl function for Conv2DBackpropFilter"""
2020-04-02 11:58:45 +08:00
def vm_impl(dout, x, w_size):
x = x.asnumpy()
dout = dout.asnumpy()
dw = vm.conv2d_backprop_filter(dout, x, w_size, self.stride, self.pad)
return Tensor(dw)
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(G.FlattenGrad)
def vm_impl_flatten_grad(self):
"""Generate vm_impl function for FlattenGrad"""
2020-04-02 11:58:45 +08:00
def vm_impl(dout, x):
dout = dout.asnumpy()
dout = vm.flatten_grad(dout, x)
return Tensor(dout)
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(P.BiasAdd)
def vm_impl_bias_add(self):
"""Generate vm_impl function for BiasAdd"""
2020-04-02 11:58:45 +08:00
def vm_impl(wx, bias):
wx = wx.asnumpy()
bias = bias.asnumpy()
out = wx + bias
return Tensor(out)
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(G.BiasAddGrad)
def vm_impl_bias_add_grad(self):
"""Generate vm_impl function for BiasAddGrad"""
2020-04-02 11:58:45 +08:00
def vm_impl(dout):
dout = dout.asnumpy()
shape = np.shape(dout)
return Tensor(np.add.reduce(dout, axis=tuple(range(len(shape) - 1))))
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(P.SoftmaxCrossEntropyWithLogits)
def vm_impl_softmax_cross_entropy_with_logits(self):
"""Generate vm_impl function for SoftmaxCrossEntropyWithLogits"""
2020-04-02 11:58:45 +08:00
def vm_impl(logits, labels):
logits = logits.asnumpy()
labels = labels.asnumpy()
loss, dx = vm.softmax_cross_entropy_with_logits(logits, labels)
return (Tensor(np.array(loss)), Tensor(dx))
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(P.SparseSoftmaxCrossEntropyWithLogits)
def vm_impl_sparse_softmax_cross_entropy_with_logits(self):
"""Generate vm_impl function for SparseSoftmaxCrossEntropyWithLogits"""
2020-04-02 11:58:45 +08:00
def vm_impl(logits, labels):
logits = logits.asnumpy()
labels = labels.asnumpy()
n_class = labels.max() + 1
n_sample = labels.shape[0]
2020-04-02 11:58:45 +08:00
one_hot_label = np.zeros((n_sample, n_class)) # 3个样本4个类别
one_hot_label[:, labels] = 1 # 非零列赋值为1
loss, dx = vm.softmax_cross_entropy_with_logits(logits, one_hot_label)
if self.is_grad:
return (Tensor(dx),)
return (Tensor(np.array(loss)),)
2020-04-02 11:58:45 +08:00
return vm_impl
2020-04-02 11:58:45 +08:00
@vm_impl_getters.register(P.ApplyMomentum)
def vm_impl_momentum(self):
"""Generate vm_impl function for Momentum"""
2020-04-02 11:58:45 +08:00
def vm_impl(variable,
accumulation,
learning_rate,
gradient,
momentum,
use_nesterov=False):
gradient = gradient.asnumpy()
accumulation = accumulation.asnumpy()
variable = variable.asnumpy()
shape = accumulation.shape
2020-04-17 12:03:50 +08:00
learning_rate = np.full(shape, learning_rate.asnumpy())
momentum = np.full(shape, momentum.asnumpy())
accumulation = accumulation * momentum + gradient
if use_nesterov is True:
variable -= gradient * learning_rate + accumulation * momentum * learning_rate
else:
variable -= accumulation * learning_rate
return Tensor(variable)
return vm_impl
2020-04-02 11:58:45 +08:00
@vm_impl_getters.register(P.ResizeBilinear)
def vm_impl_resize_bilinear(self):
"""Generate vm_impl function for ResizeBilinear"""
2020-04-02 11:58:45 +08:00
def vm_impl(x):
out = vm.ResizeBilinear(x)
return Tensor(out)
2020-04-02 11:58:45 +08:00
return vm_impl
@vm_impl_getters.register(G.ResizeBilinearGrad)
def vm_impl_resize_bilinear_grad(self):
"""Generate vm_impl function for ResizeBilinearGrad"""
2020-04-02 11:58:45 +08:00
def vm_impl(dout, original_image):
out = vm.ResizeBilinearGrad(dout, original_image)
return Tensor(out)
2020-04-02 11:58:45 +08:00
return vm_impl