【Zero-Dim】support 0D Tensor for alpha_dropout (#49882)

* alpha_dropout supports zero_dim

* add test

python/paddle/fluid/tests/unittests/test_zero_dim_tensor.py

@@ -87,6 +87,7 @@ unary_api_list = [
     paddle.nn.functional.softmax,
     paddle.nn.functional.log_softmax,
     paddle.nn.functional.gumbel_softmax,
+    paddle.nn.functional.alpha_dropout,
 ]
 
 inplace_api_list = [

python/paddle/nn/functional/common.py

@@ -1436,7 +1436,7 @@ def alpha_dropout(x, p=0.5, training=True, name=None):
         random_tensor = paddle.uniform(
             input_shape, dtype='float32', min=0.0, max=1.0
         )
-        p = full(shape=[1], fill_value=p, dtype='float32')
+        p = full(shape=input_shape, fill_value=p, dtype='float32')
         keep_mask = paddle.greater_equal(random_tensor, p)
         keep_mask = paddle.cast(keep_mask, dtype)
         drop_mask = paddle.subtract(
@@ -1444,7 +1444,7 @@ def alpha_dropout(x, p=0.5, training=True, name=None):
         )
 
         # apply mask
-        b = full(shape=[1], fill_value=b, dtype=dtype)
+        b = full(shape=input_shape, fill_value=b, dtype=dtype)
         y = paddle.add(
             paddle.multiply(x, keep_mask),
             paddle.scale(drop_mask, scale=alpha_p),