diff --git a/d2l/mindspore.py b/d2l/mindspore.py
index 7b9523b..4515573 100644
--- a/d2l/mindspore.py
+++ b/d2l/mindspore.py
@@ -1291,7 +1291,7 @@ def __init__(self, key_size, query_size, num_hiddens, dropout, **kwargs):
         self.W_k = nn.Dense(key_size, num_hiddens, has_bias=False)
         self.W_q = nn.Dense(query_size, num_hiddens, has_bias=False)
         self.w_v = nn.Dense(num_hiddens, 1, has_bias=False)
-        self.dropout = nn.Dropout(1 - dropout)
+        self.dropout = nn.Dropout(p = dropout)
 
     def construct(self, queries, keys, values, valid_lens):
         queries, keys = self.W_q(queries), self.W_k(keys)
@@ -1307,7 +1307,7 @@ class DotProductAttention(nn.Cell):
 
     def __init__(self, dropout, **kwargs):
         super(DotProductAttention, self).__init__(**kwargs)
-        self.dropout = nn.Dropout(1 - dropout)
+        self.dropout = nn.Dropout(p = dropout)
 
     def construct(self, queries, keys, values, valid_lens=None):
         d = queries.shape[-1]
@@ -1376,7 +1376,7 @@ class PositionalEncoding(nn.Cell):
 
     def __init__(self, num_hiddens, dropout, max_len=1000):
         super(PositionalEncoding, self).__init__()
-        self.dropout = nn.Dropout(1 - dropout)
+        self.dropout = nn.Dropout(p = dropout)
         self.P = d2l.zeros((1, max_len, num_hiddens))
         X = d2l.arange(max_len, dtype=mindspore.float32).reshape(
             -1, 1) / d2l.pow(10000, d2l.arange(
@@ -1408,7 +1408,7 @@ class AddNorm(nn.Cell):
 
     def __init__(self, normalized_shape, dropout, **kwargs):
         super(AddNorm, self).__init__(**kwargs)
-        self.dropout = nn.Dropout(1 - dropout)
+        self.dropout = nn.Dropout(p = dropout)
         self.ln = d2l.LayerNorm(normalized_shape)
 
     def construct(self, X, Y):