diff --git a/imperative/python/megengine/core/tensor/tensor_wrapper.py b/imperative/python/megengine/core/tensor/tensor_wrapper.py
index 011e7d1183db67dd129a82051e42a160082295ae..0bf0d7ee68d9e67db64dd352baeb420c93e3ff87 100644
--- a/imperative/python/megengine/core/tensor/tensor_wrapper.py
+++ b/imperative/python/megengine/core/tensor/tensor_wrapper.py
@@ -23,6 +23,15 @@ from .tensor import Tensor
 
 def _elwise(*args, mode):
     op = builtin.Elemwise(mode=mode)
+    if mode in ("TRUE_DIV", "POW"):
+        args = tuple(
+            map(
+                lambda x: x.astype("float32")
+                if hasattr(x, "dtype") and x.dtype != np.float32
+                else x,
+                args,
+            )
+        )
     args = utils.convert_inputs(*args)
     (result,) = apply(op, *args)
     return result
diff --git a/imperative/python/megengine/functional/elemwise.py b/imperative/python/megengine/functional/elemwise.py
index 3b8ac1f08f1230151d07be35bc1b14e270964386..f3a43733785c82a07e22a5ce976afbb46a341d60 100644
--- a/imperative/python/megengine/functional/elemwise.py
+++ b/imperative/python/megengine/functional/elemwise.py
@@ -76,6 +76,10 @@ __all__ = [
 
 def _elwise(*args, mode):
     op = builtin.Elemwise(mode=mode)
+    if mode in ("true_div", "exp", "pow", "log", "expm1", "log1p"):
+        args = tuple(
+            map(lambda x: x.astype("float32") if hasattr(x, "dtype") else x, args)
+        )
     args = utils.convert_inputs(*args)
     (result,) = apply(op, *args)
     return result