cauchy7203

The same issue appeared. The source code calling qpth ``` python import torch from qpth.qp import QPFunction slv = QPFunction() ``` Then the error reported: ``` Traceback (most recent call...

> btw, the macbook is: macbook pro, m1, 13 inch (2020).

> [leovinus2001](/leovinus2001) Thanks for reply! I have read the issue you mentioned, but it is different to my obvervation. There is NO ERROR during conversion of the data from CPU...

I do not think the issue is caused by `torch.rand`, because from the above results we can confirm that the matrices A and B on CPU and CPU are the...

You're right! ! Within your code, it prints: ```python tensor([[ 46., 28., 22.], [118., 73., 58.], [190., 118., 94.]]) ``` Now it is excactly the same to that on GPU:...

> I can confirm your guess is correct now, because I tested it several times on random matrices. It is just the case all the operations `A @ B` are...

The same case: ```python import torch N = 3 A = torch.rand([N, N], dtype=torch.float32)*2 - 1 B = torch.rand([N, N], dtype=torch.float32)*5 - 2.5 # C = A @ B.T C...

This doc may help I guess: https://docs.python.org/3/library/operator.html#mapping-operators-to-functions It is actually the case that the Python itself maps `@` as the function `matmult`, so there must be a (private) method in...

> > 1.12.0 for macos (torch-nightly) > > @cauchy7203 thanks for the report. This issue should be already fixed in the latest pytorch nightly (**1.13.0**.dev20220924) - could you please give...

> Yeah I think we should close this issue , it’s part of test consistency . @cauchy7203 please reopen the issue if you have any concerns Yes. Thank you all...