!48293 [docs][api] Update GaussianNllLoss document

Merge pull request !48293 from shaojunsong/code_docs_gnl

docs/api/api_python/nn/mindspore.nn.GaussianNLLLoss.rst

@@ -12,7 +12,7 @@ mindspore.nn.GaussianNLLLoss
         \ \text{eps}\right)\right) + \frac{\left(\text{logits} - \text{labels}\right)^2}
         {\text{max}\left(\text{var}, \ \text{eps}\right)}\right) + \text{const.}
 
-    其中，:math:`eps` 用于 :math:`log` 的稳定性。在默认情况下，常数部分被忽略，除非 :math:`full=True`。如果 :math:`var` 和 :math:`logits` 的shape不一致（出于同方差性的假设），那么它必须最后一个维度是1，或者具有更少的维度（其他维度相同），来获得正确的广播。
+    其中，:math:`eps` 用于 :math:`log` 的稳定性。当 :math:`full=True` 时，一个常数会被添加到loss中。如果 :math:`var` 和 :math:`logits` 的shape不一致（出于同方差性的假设），那么它们必须能够正确地广播。
 
     参数：
         - **full** (bool) - 指定损失函数中的常数部分。如果为True，则常数为 :math:`const = 0.5*log(2*pi)`。默认值：False。

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

@@ -12,7 +12,7 @@ mindspore.ops.gaussian_nll_loss
         \ \text{eps}\right)\right) + \frac{\left(\text{x} - \text{target}\right)^2}
         {\text{max}\left(\text{var}, \ \text{eps}\right)}\right) + \text{const.}
 
-    其中，:math:`eps` 用于 :math:`log` 的稳定性。在默认情况下，常数部分被忽略，除非 :math:`full=True`。如果 :math:`var` 和 :math:`x` 的shape不一致（出于同方差性的假设），那么它必须最后一个维度是1，或者具有更少的维度（其他维度相同），来获得正确的广播。
+    其中，:math:`eps` 用于 :math:`log` 的稳定性。当 :math:`full=True` 时，一个常数会被添加到loss中。如果 :math:`var` 和 :math:`logits` 的shape不一致（出于同方差性的假设），那么它们必须能够正确地广播。
 
     参数：
         - **x** (Tensor) - shape为 :math:`(N, *)` 或 :math:`(*)`。`*` 代表着任意数量的额外维度。

mindspore/python/mindspore/nn/loss/loss.py

@@ -2376,25 +2376,25 @@ class CTCLoss(LossBase):
 
 
 class GaussianNLLLoss(LossBase):
-    r"""Gaussian negative log likelihood loss.
+    r"""
+    Gaussian negative log likelihood loss.
 
-    The targets are treated as samples from Gaussian distributions with expectations and variances predicted by the
-    neural network. For a `labels` tensor modelled as having Gaussian distribution with a tensor of expectations
-    `logits` and a tensor of positive variances `var` the loss is:
+    The target values are considered to be samples from a Gaussian distribution, where the expectation and variance are
+    predicted by a neural network. For `labels` modeled on a Gaussian distribution, `logits` to record expectations,
+    and the variance `var` (elements are all positive), the calculated loss is:
 
     .. math::
         \text{loss} = \frac{1}{2}\left(\log\left(\text{max}\left(\text{var},
         \ \text{eps}\right)\right) + \frac{\left(\text{logits} - \text{labels}\right)^2}
         {\text{max}\left(\text{var}, \ \text{eps}\right)}\right) + \text{const.}
 
-    where `eps` is used for stability of :math:`log`. By default, the constant term of the loss function is omitted
-    unless :math:`full=True`. If the shape of :math:`var` is not the same as `logits` (due to a
-    homoscedastic assumption), it must either have a final dimension of 1 or have one fewer dimension
-    (with all other sizes being the same) for correct broadcasting.
+    where `eps` is used for stability of :math:`log`. When :math:`full=True`, a constant will be added to the loss. If
+    the shape of :math:`var` and `logits` are not the same (due to a homoscedastic assumption), their shapes must allow
+    correct broadcasting.
 
     Args:
-        full (bool): Include the constant term in the loss calculation. When :math:`full=True`, the constant term
-            `const.` will be :math:`0.5 * log(2\pi)`. Default: False.
+        full (bool): Whether include the constant term in the loss calculation. When :math:`full=True`,
+             the constant term `const.` will be :math:`0.5 * log(2\pi)`. Default: False.
         eps (float): Used to improve the stability of log function. Default: 1e-6.
         reduction (str): Apply specific reduction method to the output: 'none', 'mean', or 'sum'. Default: 'mean'.
 

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

@@ -3975,19 +3975,18 @@ def gaussian_nll_loss(x, target, var, full=False, eps=1e-6, reduction='mean'):
     r"""
     Gaussian negative log likelihood loss.
 
-    The targets are treated as samples from Gaussian distributions with expectations and variances predicted by the
-    neural network. For a `target` tensor modelled as having Gaussian distribution with a tensor of expectations
-    `x` and a tensor of positive variances `var` the loss is:
+    The target values are considered to be samples from a Gaussian distribution, where the expectation and variance are
+    predicted by a neural network. For `labels` modeled on a Gaussian distribution, `logits` to record expectations,
+    and the variance `var` (elements are all positive), the calculated loss is:
 
     .. math::
         \text{loss} = \frac{1}{2}\left(\log\left(\text{max}\left(\text{var},
         \ \text{eps}\right)\right) + \frac{\left(\text{x} - \text{target}\right)^2}
         {\text{max}\left(\text{var}, \ \text{eps}\right)}\right) + \text{const.}
 
-    where `eps` is used for stability of :math:`log`. By default, the constant term of the loss function is omitted
-    unless :math:`full=True`. If the shape of :math:`var` is not the same as `x` (due to a
-    homoscedastic assumption), it must either have a final dimension of 1 or have one fewer dimension
-    (with all other sizes being the same) for correct broadcasting.
+    where `eps` is used for stability of :math:`log`. When :math:`full=True`, a constant will be added to the loss. If
+    the shape of :math:`var` and `logits` are not the same (due to a homoscedastic assumption), their shapes must allow
+    correct broadcasting.
 
     Args:
         x (Tensor): Tensor of shape :math:`(N, *)` or :math:`(*)` where :math:`*` means any number of