-
### 🐛 Describe the bug
fp32 static shape default wrapper
suite
name
thread
batch_size_new
speed_up_new
inductor_new
eager_new
compi…
-
### Your current environment
```text
PyTorch version: 2.3.1+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A
OS: Ubuntu 22.04.3 LTS (x86_64)
GCC ve…
-
### 🐛 Describe the bug
AMP static shape CPP wrapper
suite
name
thread
batch_size_new
speed_up_new
inductor_new
eager_new
compilatio…
-
### Bug description
Got
**RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:1 and cuda:0! (when checking argument for argument index in method wrapp…
-
Edited top comment for an informative overview (ignore the diagnostic comments below):
AmberTools is refusing to assign AM1 charges to methane on macOS-13. Notably this works on macOS-latest. The D…
-
### 🐛 Describe the bug
Hi, we have found that PyTorch 2.1.2 utilizes significantly higher CPU memory utilization from PyTorch 2.0.1 on AWS g5.xlarge instance type. Below are the results we observed…
-
Hi there,
I've been trying at to read a GeoParquet file using `pyogrio` and was wondering if:
1. It's actually possible with `pyogrio`, because looking at https://pyogrio.readthedocs.io/en/late…
-
### 🐛 Describe the bug
# Issue Message
```Bash
$python build_triton_wheel.py --device xpu
HEAD is now at c451d567 Update bfloat16 conversion property (#1945)
Traceback (most recent call la…
-
### 🐛 Describe the bug
I am using nn.Linear in my model. When I trained my model on different GPU , the model performance is different. I found the nn.Linear output is different when I used the same …
-
## 🐛 Bug
I was training **GRU Model** on **TPU V4-8** for the text classification task, but the **loss didn't converge** and the **validation accuracy remained the same**. Using **RNN** and **LSTM…