提交 a5feb771 编写于 作者: K Kexin Zhao

address pr comment

上级 19e5c24f
此差异已折叠。
......@@ -55,10 +55,12 @@ TEST(float16, conversion_cpu) {
EXPECT_EQ(float16(false).x, 0x0000);
// Implicit conversion to and from Eigen::half
/*
Eigen::half tmp = float16(1.0f);
float16 v_conv = tmp;
EXPECT_EQ(tmp.x, 0x3c00);
EXPECT_EQ(v_conv.x, 0x3c00);
*/
// Default constructor
float16 v_def;
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册