[API compatibility] add dtype conversion method #74416
Merged
+386 −0
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters. Learn more about bidirectional Unicode characters
| 你的PR提交成功，感谢你对开源项目的贡献! |
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@ ## develop #74416 +/- ## =========================================== Coverage ? 100.00% =========================================== Files ? 2 Lines ? 46 Branches ? 0 =========================================== Hits ? 46 Misses ? 0 Partials ? 0 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
zhwesky2010 reviewed Aug 6, 2025
| return _C_ops.cast(self, dtype) | ||
| | ||
| def byte(self: Tensor) -> Tensor: | ||
| # since paddle don't support float to uint8, so we need to convert it to int8 first |
There was a problem hiding this comment.
这个在c++ cast kernel里能新增注册类型吗，c++里没注册上uint8的类型吧
There was a problem hiding this comment.
看到cast支持转换到uint8，默认会将小于0的float数据转换为0，直接使用cast转换float 到 uint8的结果与torch.Tensor.byte不一致，所以用了两次转换暂时保证paddle.Tensor.byte与Torch一致。
# 直接转换 >>> paddle.cast(paddle.to_tensor([-1, -2, -3, 12, 2, 3], dtype='float32'), 'uint8') Tensor(shape=[6], dtype=uint8, place=Place(gpu:0), stop_gradient=True, [0 , 0 , 0 , 12, 2 , 3 ]) # 先转成int, 再转成uint8 >>> paddle.cast(paddle.to_tensor([-1., -2., -3., 12., 2., 3.], dtype='int32'), 'uint8') Tensor(shape=[6], dtype=uint8, place=Place(gpu:0), stop_gradient=True, [255, 254, 253, 12 , 2 , 3 ]) # Torch >>> torch.Tensor([-1., -2., -3., 12., 2., 3.]).byte() tensor([255, 254, 253, 12, 2, 3], dtype=torch.uint8) | tensor = astype(self, 'int8') | ||
| return astype(tensor, 'uint8') | ||
| elif self.is_complex(): | ||
| real = astype(self.real(), 'int8') |
| | ||
| return _C_ops.cast(self, dtype) | ||
| | ||
| def byte(self): |
zhwesky2010 changed the title zhwesky2010 approved these changes Aug 7, 2025
| if self.is_floating_point(): | ||
| tensor = astype(self, 'int8') | ||
| return astype(tensor, 'uint8') | ||
| elif self.is_complex(): |
There was a problem hiding this comment.
复数情况下在PaConvert里和torch对比下，看torch是不是也是这么处理的
There was a problem hiding this comment.
已在PaConvert新增test: PaddlePaddle/PaConvert#617 ，并在本地测试通过
def test_complex64_to_byte(): # for complex --> uint8, we used the new test data pytorch_code = textwrap.dedent( """ import torch src = torch.tensor([0.+3.5j, -1+4.2j, 2.34-5.2j, -3.45+7.9j, -0.34-8.2j, 0.23+9.2j, 1.+1.j, 2.+0.5j, 3.-1.j,], dtype=torch.complex64) result = src.byte() """ ) obj.run(pytorch_code, ["result"]) def test_complex128_to_byte(): # for complex --> uint8, we used the new test data pytorch_code = textwrap.dedent( """ import torch src = torch.tensor([0.+3.5j, -1+4.2j, 2.34-5.2j, -3.45+7.9j, -0.34-8.2j, 0.23+9.2j, 1.+1.j, 2.+0.5j, 3.-1.j,], dtype=torch.complex128) result = src.byte() """ ) obj.run(pytorch_code, ["result"]) 
| /re-run all-failed |
| LGTM |
SigureMo approved these changes Aug 7, 2025
Enigmatisms pushed a commit to Enigmatisms/Paddle that referenced this pull request Aug 9, 2025
* add tensor.bool * add tensor.bool test * update tensor.bool test * update tensor.bool * delete useless test code * add dtype conversions method * skip the complex128 test on XPU * fix: Implement the byte function separately * fix: update unit testing
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit. This suggestion is invalid because no changes were made to the code. Suggestions cannot be applied while the pull request is closed. Suggestions cannot be applied while viewing a subset of changes. Only one suggestion per line can be applied in a batch. Add this suggestion to a batch that can be applied as a single commit. Applying suggestions on deleted lines is not supported. You must change the existing code in this line in order to create a valid suggestion. Outdated suggestions cannot be applied. This suggestion has been applied or marked resolved. Suggestions cannot be applied from pending reviews. Suggestions cannot be applied on multi-line comments. Suggestions cannot be applied while the pull request is queued to merge. Suggestion cannot be applied right now. Please check back later.
PR Category
User Experience
PR Types
New features
Description
add Tensor.bool, Tensor.float16, Tensor.half, Tensor.bfloat16, Tensor.float32, Tensor.float, Tensor.float64, Tensor.double, Tensor.int8, Tensor.char, Tensor.uint8, Tensor.byte, Tensor.int16, Tensor.short, Tensor.int32, Tensor.int, Tensor.int64, Tensor.long, Tensor.complex64, Tensor.complex128, Tensor.cfloat, Tensor.cdouble API
Pcard-75624