未验证 提交 e606b44a 编写于 作者: T Tomasz Socha 提交者: GitHub

Fix operator== for float16 (#39400)

上级 ece200b3
...@@ -73,7 +73,7 @@ void FakeDot(const paddle::CPUContext& dev_ctx, const paddle::Tensor& x, ...@@ -73,7 +73,7 @@ void FakeDot(const paddle::CPUContext& dev_ctx, const paddle::Tensor& x,
assert(fake_attr_float == 2); assert(fake_attr_float == 2);
assert(fake_attr_double == 3); assert(fake_attr_double == 3);
assert(fake_attr_int64 == 4); assert(fake_attr_int64 == 4);
assert(fake_attr_f16 == 5); assert(fake_attr_f16 == pten::dtype::float16(5));
assert(fake_attr_dtype == pten::DataType::UINT32); assert(fake_attr_dtype == pten::DataType::UINT32);
assert(fake_attr_int64_vec.size() == 0); assert(fake_attr_int64_vec.size() == 0);
assert(fake_attr_int_vec.size() == 0); assert(fake_attr_int_vec.size() == 0);
......
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