[Zero-Dim] support input 0D Tensor for sundary api #47734
Conversation
|
你的PR提交成功，感谢你对开源项目的贡献! |
zhwesky2010
force-pushed
the
0d_sundary1
branch
from
0f6316e
to
b1c1c76
Compare
| ] | ||
|
|
||
| binary_api_list_without_grad = [ | ||
| {'func': paddle.pow, 'cls_method': '__pow__'}, | ||
| {'func': paddle.add, 'cls_method': '__add__'}, | ||
| {'func': paddle.subtract, 'cls_method': '__sub__'}, | ||
| {'func': paddle.multiply, 'cls_method': '__mul__'}, | ||
| {'func': paddle.divide, 'cls_method': '__div__'}, | ||
| {'func': paddle.subtract, 'cls_method': '__sub__'}, |
There was a problem hiding this comment.
why did paddle.subtract repeat 4 times, while paddle.add 2 times, what is the meaning of repeat in the list?
There was a problem hiding this comment.
shall we add paddle.maximum and paddle.minimum ?
There was a problem hiding this comment.
目前只支持了部分二元，就是这个list中包含的。这两个还未支持0D，后面会加。
There was a problem hiding this comment.
why did
paddle.subtractrepeat 4 times, whilepaddle.add2 times, what is the meaning of repeat in the list?
Some mistake, have fixed it.
| {'func': paddle.equal, 'cls_method': '__eq__'}, | ||
| {'func': paddle.not_equal, 'cls_method': '__ne__'}, | ||
| {'func': paddle.greater_equal, 'cls_method': '__ge__'}, | ||
| {'func': paddle.greater_than, 'cls_method': '__gt__'}, | ||
| {'func': paddle.less_equal, 'cls_method': '__le__'}, | ||
| {'func': paddle.less_than, 'cls_method': '__lt__'}, | ||
| {'func': paddle.remainder, 'cls_method': '__mod__'}, | ||
| paddle.mod, | ||
| paddle.floor_mod, | ||
| paddle.logical_and, | ||
| paddle.logical_or, | ||
| paddle.logical_xor, |
There was a problem hiding this comment.
shall we add paddle.bitwise_and/or/xor ?
PR types
New features
PR changes
APIs
Describe
为无需测反向的这类API支持输入0D Tensor，例如判断类、属性等API，如下：