-
### Your current environment
```text
Collecting environment information...
PyTorch version: 2.3.0+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A
…
-
### 🐛 Describe the bug
**Introduction**:
I am developing an application using PyTorch and have noticed an unusual behavior related to memory management. Specifically, when I instantiate a batch …
-
### Your current environment
```text
The output of `python collect_env.py`
```
### How you are installing vllm
I install vLLM using Souce code.
```python
pip install -e .
```
but encounter…
-
Hey together,
I am having issues exporting a pytorch model to onnx via torch.onnx.dynamo_export.
I created a small code example for reproduction of the issue and have the following questions:
1. …
-
### 🐛 Describe the bug
When building pytorch main branch (978a2f2b276b51f615aa860d47fadd16a284b2f6) with:
```
python tools/amd_build/build_amd.py
export USE_ROCM=1
export BUILD_CAFFE2=1
pytho…
-
### Your current environment
```
Collecting environment information...
PyTorch version: 2.3.0+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A
OS…
-
### 🐛 Describe the bug
Since the pytorch 2.1.0, the forward propagation speed of ``nn.Linear`` against high-dimensional tensors, i.e. tensors with four, five, or even more dimensions, is noticeably…
-
### 🐛 Describe the bug
I have encountered a performance problem when executing a model that utilizes Flash Attention using torch.jit trace with C++ libtorch on Windows. The inference speed on Windo…
-
### 🐛 Describe the bug
The serialization of a treespec object is to losslessly convert it into a byte/char stream. Then the serialized representation can be losslessly recovered (deserialized) into…
-
### 🐛 Describe the bug
## RuntimeError: "scatter_gather_base_kernel_func" not implemented for 'Bool' :
"test_comprehensive_scatter_reduce_amax_xpu_bool",
"test_comprehensive_scatter_reduce_…