-
### 🐛 Describe the bug
The op aten._weight_norm_interface is not a core Aten op, and it doesn't have a decomposition.
Repro:
```
import torch
args = (torch.randn(768, 48, 128), torch.randn(…
qihqi updated
9 months ago
-
The command:
```
python3.10 -muv pip install -U --prerelease allow torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu121
```
(Run in a clean python3.10 venv, wi…
-
### 🐛 Describe the bug
My repro:
```
python benchmarks/dynamo/torchbench.py --only Super_SloMo --backend onnxrt -dcuda --performance --inference
```
with this line `opset_version=17,` added f…
-
### Checklist
- [X] 1. I have searched related issues but cannot get the expected help.
- [ ] 2. The bug has not been fixed in the latest version.
- [ ] 3. Please note that if the bug-related issue y…
-
### 🐛 Describe the bug
I am currently testing the optimization features of torchdynamo and comparing them with the optimization results of TVM. I have already completed the optimization evaluation …
-
### 🐛 Describe the bug
```python
import torchvision
```
gives the warning:
```
/home/ray/anaconda3/lib/python3.9/site-packages/torchvision/io/image.py:13: UserWarning: Failed to load image Pyt…
-
### 🐛 Describe the bug
Compare with the 2023-03-08, there is a performance regression on huggingface model **MobileBertForQuestionAnswering** on [TorchInductor CPU Performance Dashboard](https://gith…
-
### 🐛 Describe the bug
```python
from functorch.experimental.control_flow import cond
import torch
def true_fn(x):
return x / 2.0
def false_fn(x):
return x*2.0
def fn(x, scale=4.):…
-
### Your current environment
```text
PyTorch version: 2.2.1+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A
OS: Ubuntu 22.04.4 LTS (x86_64)
GCC …
-
### 🐛 Describe the bug
This issue is for tracking cherry-picks to the release branch. Following is [release branch](https://github.com/pytorch/pytorch/tree/release/2.3) for the 2.3.1 release.
Ou…