Add Abs\AbsGrad\Sign\SmoothL1Loss\SmoothL1LossGrad and modify TopKV2->TopK for VM

This commit is contained in:
liuxiao 2020-04-15 17:50:34 +08:00
parent 0a9db34d5a
commit 5c9791a802
12 changed files with 327 additions and 10 deletions

View File

@ -42,7 +42,6 @@ static std::map<string, string> tbe_func_adapter_map = {
{"depthwise_conv2d_native", "depthwise_conv2d"},
{"depthwise_conv2d_native_backprop_filter", "depthwise_conv2d_backprop_filter_d"},
{"depthwise_conv2d_native_backprop_input", "depthwise_conv2d_backprop_input_d"},
{"top_kv2", "top_k"},
{"scatter_nd", "scatter_nd_d"},
{"tile", "tile_d"},
{"gather_v2", "gather_v2_d"},

View File

@ -14,6 +14,8 @@
# ============================================================================
"""tbe ops"""
from .abs import _abs_tbe
from .abs_grad import _abs_grad_tbe
from .adam_apply_one_with_decay import _adam_apply_one_with_decay_tbe
from .add import _add_tbe
from .add_n import _add_n_tbe
@ -49,7 +51,7 @@ from .sigmoid_cross_entropy_with_logits import _sigmoid_cross_entropy_with_logit
from .sigmoid_cross_entropy_with_logits_grad import _sigmoid_cross_entropy_with_logits_grad_tbe
from .tensor_add import _tensor_add_tbe
from .trans_data import _trans_data_tbe
from .topkv2 import _topk_v2_tbe
from .top_k import _top_k_tbe
from .matmul import _matmul_tbe
from .sub import _sub_tbe
from .reduce_mean_d import _reduce_mean_d_tbe
@ -107,6 +109,7 @@ from .minimum_grad import _minimum_grad_tbe
from .maximum_grad import _maximum_grad_tbe
from .concat import _concat_tbe
from .slice import _slice_tbe
from .sign import _sign_tbe
from .greater import _greater_tbe
from .clip_by_norm_no_div_sum import _clip_by_norm_no_div_sum_tbe
from .clip_by_value import _clip_by_value_tbe
@ -130,6 +133,8 @@ from .resize_nearest_neighbor_grad_d import _resize_nearest_neighbor_grad_d_tbe
from .pad_d import _pad_d_tbe
from .arg_max_with_value import _arg_max_with_value_tbe
from .arg_min_with_value import _arg_min_with_value_tbe
from .smooth_l1_loss import _smooth_l1_loss_tbe
from .smooth_l1_loss_grad import _smooth_l1_loss_grad_tbe
from .fused_mul_add import _fused_mul_add_tbe
from .fused_mul_add_n import _fused_mul_add_n_tbe
from .fused_mul_apply_momentum import _fused_mul_apply_momentum_tbe

View File

@ -0,0 +1,41 @@
# Copyright 2020 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.
# ============================================================================
"""Abs op"""
from mindspore.ops.op_info_register import op_info_register, TBERegOp, DataType
abs_op_info = TBERegOp("Abs") \
.fusion_type("ELEMWISE") \
.async_flag(False) \
.binfile_name("abs.so") \
.compute_cost(10) \
.kernel_name("abs") \
.partial_flag(True) \
.op_pattern("formatAgnostic") \
.input(0, "x", None, "required", None) \
.output(0, "y", True, "required", "all") \
.dtype_format(DataType.F16_Default, DataType.F16_Default) \
.dtype_format(DataType.F16_5HD, DataType.F16_5HD) \
.dtype_format(DataType.F32_Default, DataType.F32_Default) \
.dtype_format(DataType.F32_5HD, DataType.F32_5HD) \
.dtype_format(DataType.I32_Default, DataType.I32_Default) \
.dtype_format(DataType.I32_5HD, DataType.I32_5HD) \
.get_op_info()
@op_info_register(abs_op_info)
def _abs_tbe():
"""Abs TBE register"""
return

View File

@ -0,0 +1,44 @@
# Copyright 2020 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.
# ============================================================================
"""AbsGrad op"""
from mindspore.ops.op_info_register import op_info_register, TBERegOp, DataType
abs_grad_op_info = TBERegOp("AbsGrad") \
.fusion_type("ELEMWISE") \
.async_flag(False) \
.binfile_name("abs_grad.so") \
.compute_cost(10) \
.kernel_name("abs_grad") \
.partial_flag(True) \
.op_pattern("formatAgnostic") \
.input(0, "y", None, "required", None) \
.input(1, "dy", None, "required", None) \
.output(0, "z", False, "required", "all") \
.dtype_format(DataType.F16_5HD, DataType.F16_5HD, DataType.F16_5HD) \
.dtype_format(DataType.F16_FracZ, DataType.F16_FracZ, DataType.F16_FracZ) \
.dtype_format(DataType.F16_C1HWNCoC0, DataType.F16_C1HWNCoC0, DataType.F16_C1HWNCoC0) \
.dtype_format(DataType.F16_Default, DataType.F16_Default, DataType.F16_Default) \
.dtype_format(DataType.F32_5HD, DataType.F32_5HD, DataType.F32_5HD) \
.dtype_format(DataType.F32_FracZ, DataType.F32_FracZ, DataType.F32_FracZ) \
.dtype_format(DataType.F32_C1HWNCoC0, DataType.F32_C1HWNCoC0, DataType.F32_C1HWNCoC0) \
.dtype_format(DataType.F32_Default, DataType.F32_Default, DataType.F32_Default) \
.get_op_info()
@op_info_register(abs_grad_op_info)
def _abs_grad_tbe():
"""AbsGrad TBE register"""
return

View File

@ -0,0 +1,41 @@
# Copyright 2020 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.
# ============================================================================
"""Sign op"""
from mindspore.ops.op_info_register import op_info_register, TBERegOp, DataType
sign_op_info = TBERegOp("Sign") \
.fusion_type("ELEMWISE") \
.async_flag(False) \
.binfile_name("sign.so") \
.compute_cost(10) \
.kernel_name("sign") \
.partial_flag(True) \
.op_pattern("formatAgnostic") \
.input(0, "x", None, "required", None) \
.output(0, "y", True, "required", "all") \
.dtype_format(DataType.F16_Default, DataType.F16_Default) \
.dtype_format(DataType.F16_5HD, DataType.F16_5HD) \
.dtype_format(DataType.F32_Default, DataType.F32_Default) \
.dtype_format(DataType.F32_5HD, DataType.F32_5HD) \
.dtype_format(DataType.I32_Default, DataType.I32_Default) \
.dtype_format(DataType.I32_5HD, DataType.I32_5HD) \
.get_op_info()
@op_info_register(sign_op_info)
def _sign_tbe():
"""Sign TBE register"""
return

View File

@ -0,0 +1,44 @@
# Copyright 2020 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.
# ============================================================================
"""SmoothL1Loss op"""
from mindspore.ops.op_info_register import op_info_register, TBERegOp, DataType
smooth_l1_loss_op_info = TBERegOp("SmoothL1Loss") \
.fusion_type("OPAQUE") \
.async_flag(False) \
.binfile_name("smooth_l1_loss.so") \
.compute_cost(10) \
.kernel_name("smooth_l1_loss") \
.partial_flag(True) \
.attr("sigma", "required", "float", "all") \
.input(0, "predict", False, "required", "all") \
.input(1, "label", False, "required", "all") \
.output(0, "loss", False, "required", "all") \
.dtype_format(DataType.F16_Default, DataType.F16_Default, DataType.F16_Default) \
.dtype_format(DataType.F16_5HD, DataType.F16_5HD, DataType.F16_5HD) \
.dtype_format(DataType.F16_FracZ, DataType.F16_FracZ, DataType.F16_FracZ) \
.dtype_format(DataType.F16_C1HWNCoC0, DataType.F16_C1HWNCoC0, DataType.F16_C1HWNCoC0) \
.dtype_format(DataType.F32_Default, DataType.F32_Default, DataType.F32_Default) \
.dtype_format(DataType.F32_5HD, DataType.F32_5HD, DataType.F32_5HD) \
.dtype_format(DataType.F32_FracZ, DataType.F32_FracZ, DataType.F32_FracZ) \
.dtype_format(DataType.F32_C1HWNCoC0, DataType.F32_C1HWNCoC0, DataType.F32_C1HWNCoC0) \
.get_op_info()
@op_info_register(smooth_l1_loss_op_info)
def _smooth_l1_loss_tbe():
"""SmoothL1Loss TBE register"""
return

View File

@ -0,0 +1,45 @@
# Copyright 2020 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.
# ============================================================================
"""SmoothL1LossGrad op"""
from mindspore.ops.op_info_register import op_info_register, TBERegOp, DataType
smooth_l1_loss_grad_op_info = TBERegOp("SmoothL1LossGrad") \
.fusion_type("OPAQUE") \
.async_flag(False) \
.binfile_name("smooth_l1_loss_grad.so") \
.compute_cost(10) \
.kernel_name("smooth_l1_loss_grad") \
.partial_flag(True) \
.attr("sigma", "required", "float", "all") \
.input(0, "predict", False, "required", "all") \
.input(1, "label", False, "required", "all") \
.input(2, "dout", False, "required", "all") \
.output(0, "loss", False, "required", "all") \
.dtype_format(DataType.F16_Default, DataType.F16_Default, DataType.F16_Default, DataType.F16_Default) \
.dtype_format(DataType.F16_5HD, DataType.F16_5HD, DataType.F16_5HD, DataType.F16_5HD) \
.dtype_format(DataType.F16_FracZ, DataType.F16_FracZ, DataType.F16_FracZ, DataType.F16_FracZ) \
.dtype_format(DataType.F16_C1HWNCoC0, DataType.F16_C1HWNCoC0, DataType.F16_C1HWNCoC0, DataType.F16_C1HWNCoC0) \
.dtype_format(DataType.F32_Default, DataType.F32_Default, DataType.F32_Default, DataType.F32_Default) \
.dtype_format(DataType.F32_5HD, DataType.F32_5HD, DataType.F32_5HD, DataType.F32_5HD) \
.dtype_format(DataType.F32_FracZ, DataType.F32_FracZ, DataType.F32_FracZ, DataType.F32_FracZ) \
.dtype_format(DataType.F32_C1HWNCoC0, DataType.F32_C1HWNCoC0, DataType.F32_C1HWNCoC0, DataType.F32_C1HWNCoC0) \
.get_op_info()
@op_info_register(smooth_l1_loss_grad_op_info)
def _smooth_l1_loss_grad_tbe():
"""SmoothL1LossGrad TBE register"""
return

View File

@ -13,15 +13,15 @@
# limitations under the License.
# ============================================================================
"""TopKV2 op"""
"""TopK op"""
from mindspore.ops.op_info_register import op_info_register, TBERegOp, DataType
top_k_v2_op_info = TBERegOp("TopKV2") \
top_k_op_info = TBERegOp("TopK") \
.fusion_type("OPAQUE") \
.async_flag(False) \
.binfile_name("top_k_v2.so") \
.binfile_name("top_k.so") \
.compute_cost(10) \
.kernel_name("top_k_v2") \
.kernel_name("top_k") \
.partial_flag(True) \
.attr("k", "required", "int", "all")\
.attr("sorted", "required", "bool", "all")\
@ -33,7 +33,7 @@ top_k_v2_op_info = TBERegOp("TopKV2") \
.get_op_info()
@op_info_register(top_k_v2_op_info)
def _topk_v2_tbe():
"""TopKV2 TBE register"""
@op_info_register(top_k_op_info)
def _top_k_tbe():
"""TopK TBE register"""
return

View File

@ -599,3 +599,4 @@ class DataType:
F32_NCHW = ("float32", "NCHW")
F32_NHWC = ("float32", "NHWC")
F32_HWCN = ("float32", "HWCN")

View File

@ -0,0 +1,42 @@
# Copyright 2020 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.
# ============================================================================
import numpy as np
import mindspore.nn as nn
import mindspore.context as context
from mindspore import Tensor
from mindspore.ops import operations as P
context.set_context(mode=context.GRAPH_MODE, device_target="Ascend")
class Net(nn.Cell):
def __init__(self, sigma=1.0):
super(Net, self).__init__()
self.SmoothL1Loss = P.SmoothL1Loss(sigma)
def construct(self, pred, gt):
return self.SmoothL1Loss(pred, gt)
def test_net():
pred = np.random.randn(2, 4).astype(np.float32)
gt = np.random.randn(2, 4).astype(np.float32)
smooth_l1_loss = Net()
loss = smooth_l1_loss(Tensor(pred), Tensor(gt))
print("------------- input ---------------")
print("predict:\n", pred)
print("grount truth:\n", gt)
print("------------- output ---------------")
print("loss:\n", loss.asnumpy())

View File

@ -0,0 +1,55 @@
# Copyright 2020 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.
# ============================================================================
import numpy as np
import mindspore.nn as nn
import mindspore.context as context
from mindspore.ops.composite import GradOperation
from mindspore import Tensor
from mindspore.ops import operations as P
context.set_context(mode=context.GRAPH_MODE, device_target="Ascend")
class Net(nn.Cell):
def __init__(self, sigma=1.0):
super(Net, self).__init__()
self.SmoothL1Loss = P.SmoothL1Loss(sigma)
def construct(self, pred, gt):
return self.SmoothL1Loss(pred, gt)
class Grad(nn.Cell):
def __init__(self, network):
super(Grad, self).__init__()
self.grad = GradOperation(name="get_all", get_all=True, sens_param=True)
self.network = network
def construct(self, pred, gt, dout):
return self.grad(self.network)(pred, gt, dout)
def test_net():
pred = np.random.randn(2, 4).astype(np.float32)
gt = np.random.randn(2, 4).astype(np.float32)
dout = np.random.randn(2, 4).astype(np.float32)
smooth_l1_loss_grad = Grad(Net())
output = smooth_l1_loss_grad(Tensor(pred), Tensor(gt), Tensor(dout))
print("------------- input ---------------")
print("predict:\n", pred)
print("grount truth:\n", gt)
print("dout:\n", dout)
print("------------- output ---------------")
print("predict grad:\n", output[0].asnumpy())

View File

@ -24,7 +24,7 @@ context.set_context(mode=context.GRAPH_MODE, device_target="Ascend")
class Net(nn.Cell):
def __init__(self, k):
super(Net, self).__init__()
self.topk = P.TopK()
self.topk = P.TopK(True)
self.k = k
def construct(self, x):