diff --git a/python/_dynet.pyx b/python/_dynet.pyx
index 79ab8f8ba..8e004bb0e 100644
--- a/python/_dynet.pyx
+++ b/python/_dynet.pyx
@@ -2774,7 +2774,7 @@ cpdef Expression nobackprop(Expression x):
         dynet.Expression: An output expression containing the same as input (only effects on backprop process)
     """
     return Expression.from_cexpr(x.cg_version, c_nobackprop(x.c()))
-cpdef Expression flip_gradient(Expression x, float lambd = 1.0):
+cpdef Expression flip_gradient(Expression x):
     """Flip gradient
     
     This node has no effect on the forward pass, but takes negative on backprop process. This operation is widely used in adversarial networks.
diff --git a/python/dynet_viz.py b/python/dynet_viz.py
index 58847880a..130316215 100644
--- a/python/dynet_viz.py
+++ b/python/dynet_viz.py
@@ -254,7 +254,7 @@ def max_dim(a, d=0): return GVExpr('max_dim', [a, d], make_dim(1, inferred=True)
 def min_dim(a, d=0): return GVExpr('min_dim', [a, d], make_dim(1, inferred=True))
 
 def nobackprop(x): return GVExpr('nobackprop', [x], copy_dim(x))
-def flip_gradient(x, lambd=1.0): return GVExpr('flip_gradient', [x, lambd], copy_dim(x))
+def flip_gradient(x): return GVExpr('flip_gradient', [x], copy_dim(x))
 
 # binary-exp
 def cdiv(x, y): return GVExpr('cdiv', [x,y], ensure_same_dim(x,y))