-
### 🐛 Describe the bug
`aten::channel_shuffle` is not a core aten op and has no decomposition.
Repro
```python
import torch
m = torch.nn.ChannelShuffle(2)
inp = torch.arange(1, 17, dtype=t…
qihqi updated
3 months ago
-
### 🐛 Describe the bug
[log.txt](https://github.com/user-attachments/files/16172377/log.txt)
### Versions
Collecting environment information...
PyTorch version: N/A
Is debug build: N/A
CUDA us…
-
### 🐛 Describe the bug
Build is failing when Pytorch is build with BLIS. The following commands are used to built, as mentioned in https://github.com/pytorch/pytorch/pull/54953
$export BLIS_HOME=p…
-
### 🐛 Describe the bug
```
import torch
import torch.nn.functional as F
def test_log_softmax(device):
print(f"Testing on {device}")
input_tensor = torch.randn(3, 4, 5, device=device)…
-
### 🐛 Describe the bug
When converting `NaN` value to `int32`, the behavior of `torch.tensor` are different on different input type and different platform.
If the input is a python built-in list c…
dlibk updated
2 weeks ago
-
(largely copied from https://rt.cpan.org/Ticket/Display.html?id=144760)
When a `defer` block throws an exception due to stack unwind of an existing exception, we entirely lose one of them:
```
…
-
### 🐛 Describe the bug
Using MPS, torch.multinomial() always return [0], even when the probability is 0.
It works as expected on CPU:
```
import torch
In [1]: x = torch.tensor([0.5, 0.5], dty…
-
### 🐛 Describe the bug
`to(torch.int8)` will get different result on XPU.
```python
import torch
torch.tensor([[ 57.7637, 215.2612, 212.4291],[193.8332, 227.0923, 158.8016]], device='cpu').t…
-
When fit on large datasets, parametric UMAP throws an unknown shape error. To reproduce this issue, use the below code:
```
import numpy as np
from umap import ParametricUMAP
from sklearn.prep…
-
### 🐛 Describe the bug
In some cases certain ops are slow. The failure mode does not always trigger, but its relatively easy to trigger this behavior if run in rapid succession.
Take the followi…