-
### 🐛 Describe the bug
```
import torch
cfn = torch.compile(torch.constant_pad_nd)
cfn(torch.zeros(5, dtype=torch.bool), [2, 3])
```
results in
```
BackendCompilerFailed: backend='inductor…
-
### 🐛 Describe the bug
When I calculate a hessian vector product of layernorm with `torch.func.jvp`, the returned output from `jvp` (i.e., the value of the function, or the average gradient of the la…
-
### 🐛 Describe the bug
Before https://github.com/pytorch/pytorch/pull/123403, when running the dynamo benchmark for AOTI, for inference on CPU, Conv and BN could be folded. However, after this PR whi…
-
With a recent Clang-18 snapshot (cd6e462d012f289cc4ec12927ca8198f9ed1469e), I've seen the following frontend crash:
```
PLEASE submit a bug report to https://github.com/llvm/llvm-project/issues/ a…
ms178 updated
1 month ago
-
Hi, and thanks for this amazing work!
I just tried to test your model on some files of mine but, I got an error when loading the model:
This is the command I use to run the code:
`CUDA_DEVICE…
-
@mthiboust Could you test with my/b-data's images?
---
`glcr.b-data.ch/jupyterlab/cuda/python/scipy:3.11.9`
* based on `nvidia/cuda:11.8.0-cudnn8-devel-ubuntu22.04`
Install TensorFlow: `pi…
-
I tried this with several different model options, same error.
InvalidArgu…
-
### 🐛 Describe the bug
After I set the `CUDNN_LIBRARY_PATH` and `CUDNN_INCLUDE_PATH`
I am still getting **CUDNN** **missing** error on building pytorch from source
```
echo $CUDNN_LIBRARY_PATH
/u…
-
### 🐛 Describe the bug
Work state in dump trace file will be `scheduled` if the watch dog thread has retired it.
#### To Reproduce
Set a sleep before dump trace in this test case:
https://github.…
-
### 🐛 Describe the bug
Hi, I'm trying to support `while_loop` with `DispatchKey.XLA`;
when I try linear and MNIST with torch, code would be dispatched to `DispatchKey.CompositeExplicitAutograd` …