fix bug of function api multi_label_margin_loss

.jenkins/check/config/whitelizard.txt

@@ -266,6 +266,7 @@ mindspore/mindspore/ccsrc/pybind_api/ir/tensor_py.cc:mindspore::tensor::RegMetaT
 mindspore/mindspore/ccsrc/plugin/device/cpu/kernel/eltwise_grad_cpu_kernel.cc:mindspore::kernel::EltWiseGradCpuTypeFunc<T>::InitFunc
 mindspore/mindspore/lite/tools/converter/quantizer/weight_quantizer.cc:mindspore::lite::quant::WeightQuantizer::LinearQuant
 mindspore/mindspore/python/mindspore/ops/function/nn_func.py:conv3d
+mindspore/mindspore/python/mindspore/ops/function/nn_func.py:max_unpool3d
 mindspore/mindspore/ccsrc/plugin/device/cpu/kernel/nnacl/fp32/matmul_avx512_mask_fp32.c:GemmRowxColMaskKernelFp32
 mindspore/mindspore/ccsrc/plugin/device/cpu/kernel/crop_and_resize_cpu_kernel.cc:mindspore::kernel::CropAndResizeCpuKernelMod::LaunchKernel
 

docs/api/api_python/ops/mindspore.ops.func_multi_label_margin_loss.rst

@@ -32,7 +32,6 @@ mindspore.ops.multi_label_margin_loss
     返回：
         - **outputs** (Union[Tensor, Scalar]) - MultilabelMarginLoss损失。如果 `reduction` 的值为 "none"，
           那么返回shape为 :math:`(N)` 的Tensor类型数据。否则返回一个标量。
-        - **is_target** (Tensor) - 基于反向输入数据，返回Tensor类型的数据，shape和 `target` 相同，类型为int32。
 
     异常：
         - **TypeError** - 当 `inputs` 或者 `target` 数据不是Tensor时。

mindspore/python/mindspore/ops/function/nn_func.py

@@ -5115,8 +5115,6 @@ def multi_label_margin_loss(inputs, target, reduction='mean'):
     Returns:
         - **outputs** (Union[Tensor, Scalar]) - The loss of MultilabelMarginLoss. If `reduction` is "none", its shape
           is :math:`(N)`. Otherwise, a scalar value will be returned.
-        - **is_target** (Tensor) - Output tensor for backward input, with the same shape as `target`,
-          data type must be int32.
 
     Raises:
         TypeError: If `inputs` or `target` is not a Tensor.
@@ -5138,7 +5136,7 @@ def multi_label_margin_loss(inputs, target, reduction='mean'):
        [[1, 1, 1, 1], [0, 0, 1, 1]]))
     """
 
-    loss = _get_cache_prim(P.MultiLabelMarginLoss)(reduction)
+    loss = _get_cache_prim(P.MultilabelMarginLoss)(reduction)
     outputs, _ = loss(inputs, target)
     return outputs