!5823 [MSLITE] Support exponent tensor broadcast for power op

Merge pull request !5823 from zhanyuan/dev

mindspore/lite/src/ops/power.cc

@@ -64,7 +64,9 @@ int Power::InferShape(std::vector<tensor::Tensor *> inputs, std::vector<tensor::
     return RET_OK;
   }
   if (exp_tensor != nullptr) {
-    if (exp_tensor->shape() != x_tensor->shape() || exp_tensor->data_type() != x_tensor->data_type()) {
+    if ((exp_tensor->shape().size() > 1 && exp_tensor->shape() != x_tensor->shape()) ||
+        (exp_tensor->shape().size() == 1 && exp_tensor->shape()[0] != 1) ||
+        exp_tensor->data_type() != x_tensor->data_type()) {
       MS_LOG(ERROR) << "Power inputs shape or type is not equal!";
       return RET_INPUT_TENSOR_ERROR;
     }

mindspore/lite/src/runtime/kernel/arm/fp32/power.cc

@@ -64,11 +64,11 @@ int PowerCPUKernel::RunImpl(int task_id) {
   bool broadcast = true;
   if (in_tensors_.size() == 2) {
     exp_addr = reinterpret_cast<float *>(in_tensors_[1]->Data());
-    broadcast = false;
+    broadcast = in_tensors_[0]->shape() == in_tensors_[1]->shape() ? false : true;
   }
   float *cur_exp = nullptr;
   if (broadcast) {
-    cur_exp = &power_;
+    cur_exp = in_tensors_.size() == 2 ? exp_addr : &power_;
   } else {
     cur_exp = exp_addr + stride * task_id;
   }