!382 Add parallel operator for SigmoidCrossEntropyWithLogits

Merge pull request !382 from yangzhenzhang/sigmoidloss

mindspore/ccsrc/parallel/dynamic_creator.h

@@ -127,6 +127,7 @@ REGISTER(NegInfo);
 REGISTER(BatchMatMulInfo);
 REGISTER(ExpandDimsInfo);
 REGISTER(SqueezeInfo);
+REGISTER(SigmoidCrossEntropyWithLogitsInfo);
 }  // namespace parallel
 }  // namespace mindspore
 

mindspore/ccsrc/parallel/ops_info/arithmetic_info.h

@@ -120,6 +120,15 @@ class AssignSubInfo : public ArithmeticBase {
       : ArithmeticBase(name, inputs_shape, outputs_shape, attrs, std::make_shared<ArithmeticCost>(false)) {}
   ~AssignSubInfo() override = default;
 };
+
+// All dimensions can be split arbitrarily, but the split method of Logits should be the same as that of label.
+class SigmoidCrossEntropyWithLogitsInfo : public ArithmeticBase {
+ public:
+  SigmoidCrossEntropyWithLogitsInfo(const std::string& name, const Shapes& inputs_shape, const Shapes& outputs_shape,
+                                    const PrimitiveAttrs& attrs)
+      : ArithmeticBase(name, inputs_shape, outputs_shape, attrs, std::make_shared<ArithmeticCost>(false)) {}
+  ~SigmoidCrossEntropyWithLogitsInfo() override = default;
+};
 }  // namespace parallel
 }  // namespace mindspore
 

mindspore/ccsrc/parallel/ops_info/ops_utils.h

@@ -138,6 +138,7 @@ constexpr char ALL_GATHER[] = "AllGather";
 constexpr char REDUCE_SCATTER[] = "ReduceScatter";
 constexpr char CONCAT[] = "Concat";
 constexpr char SOFTMAX_CROSS_ENTROPY_WITH_LOGITS[] = "SoftmaxCrossEntropyWithLogits";
+constexpr char SIGMOID_CROSS_ENTROPY_WITH_LOGITS[] = "SigmoidCrossEntropyWithLogits";
 constexpr char MATMUL[] = "MatMul";
 constexpr char GELU[] = "Gelu";
 constexpr char TANH[] = "Tanh";

mindspore/ccsrc/parallel/step_auto_parallel.cc

@@ -78,6 +78,7 @@ std::vector<std::string> splittable_op_ = {MATMUL,
                                            FUSE_BATCH_NORM,
                                            POOLING,
                                            SOFTMAX_CROSS_ENTROPY_WITH_LOGITS,
+                                           SIGMOID_CROSS_ENTROPY_WITH_LOGITS,
                                            MAX_POOL_WITH_ARGMAX,
                                            SIMPLE_MEAN,
                                            FLATTEN,

tests/ut/python/parallel/test_sigmoid_cross_entropy_with_logits.py

@@ -0,0 +1,83 @@
+# 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 as ms
+from mindspore import context, Tensor, Parameter
+from mindspore.nn import Cell, TrainOneStepCell, Momentum
+from mindspore.ops import operations as P
+from mindspore.common.api import _executor
+
+
+class Net(Cell):
+    def __init__(self, mul_weight, strategy1=None, strategy2=None):
+        super().__init__()
+        self.mul = P.Mul().set_strategy(strategy1)
+        self.loss = P.SigmoidCrossEntropyWithLogits().set_strategy(strategy2)
+        self.mul_weight = Parameter(mul_weight, "w1")
+
+    def construct(self, x, b):
+        out = self.mul(x, self.mul_weight)
+        out = self.loss(out, b)
+        return out
+
+
+_x = Tensor(np.ones([128, 64]), dtype=ms.float32)
+_w1 = Tensor(np.ones([128, 64]), dtype=ms.float32)
+_b = Tensor(np.ones([128, 64]), dtype=ms.float32)
+
+
+def compile(net):
+    optimizer = Momentum(net.trainable_params(), learning_rate=0.1, momentum=0.9)
+    train_net = TrainOneStepCell(net, optimizer)
+    _executor.compile(train_net, _x,  _b)
+    context.reset_auto_parallel_context()
+
+
+def test_sigmoid_cross_entropy_with_logits_data_parallel():
+    context.set_auto_parallel_context(parallel_mode="semi_auto_parallel", device_num=16, global_rank=0)
+    strategy1 = ((16, 1), (16, 1))
+    strategy2 = ((16, 1), (16, 1))
+    net = Net(_w1, strategy1, strategy2)
+    compile(net)
+
+
+def test_sigmoid_cross_entropy_with_logits_model_parallel():
+    context.set_auto_parallel_context(parallel_mode="semi_auto_parallel", device_num=16, global_rank=0)
+    strategy1 = ((1, 16), (1, 16))
+    strategy2 = ((1, 16), (1, 16))
+    net = Net(_w1, strategy1, strategy2)
+    compile(net)
+
+
+def test_sigmoid_cross_entropy_with_logits_hybrid_parallel():
+    context.set_auto_parallel_context(parallel_mode="semi_auto_parallel", device_num=16, global_rank=0)
+    strategy1 = ((2, 8), (2, 8))
+    strategy2 = ((2, 8), (2, 8))
+    net = Net(_w1, strategy1, strategy2)
+    compile(net)
+
+
+def test_sigmoid_cross_entropy_with_logits_auto_parallel():
+    context.set_auto_parallel_context(parallel_mode="auto_parallel", device_num=16, global_rank=0)
+    net = Net(_w1)
+    compile(net)
+
+
+def test_sigmoid_cross_entropy_with_logits_repeat_calc():
+    context.set_auto_parallel_context(parallel_mode="semi_auto_parallel", device_num=16, global_rank=0)
+    strategy1 = ((2, 8), (2, 8))
+    strategy2 = ((2, 2), (2, 2))
+    net = Net(_w1, strategy1, strategy2)
+    compile(net)