-
**Describe the bug**
Getting zombie process exception as already reported for the [sagemaker-inference-toolkit](https://github.com/aws/sagemaker-inference-toolkit/pull/133)
**To reproduce**
Using…
-
https://pytorch.org/docs/master/tensor_attributes.html#torch-device
Now this function only return `cpu` or `cuda`. I failed to run some transformer based models on my MacOS due to this limitation.
!…
-
### System Info
### System Info
* CPU Architecture - x86_64
* CPU/Host memory size - 330gb from `/proc/meminfo`
* GPU name - NVIDIA H100 80GB HBM3
* GPU memory size - 81559MiB
* TensorRT-LLM…
-
### 🐛 Describe the bug
I am using PyTorch's execution trace observer to collect traces about a basic MNIST program. While the program **does** produce valid data (everything is there), it **does not*…
-
### 🐛 Describe the bug
When attempting to use `torch.nn.functional.rrelu` with `training=True` in a model that is being compiled with `torch.compile`, the compilation fails. Feel free to close issu…
-
### Search before asking
- [X] I have searched the YOLOv8 [issues](https://github.com/ultralytics/ultralytics/issues) and [discussions](https://github.com/ultralytics/ultralytics/discussions) and fou…
-
## 🐛 Bug
`torch.load(X, map_location='cpu')` fails when unserializing cuda tensors on a cpu-only device if `X` is, for example, a list of tensors serialized with `pickle.dump`.
## To Reproduce
…
-
## Description
https://docs.djl.ai/master/docs/development/profiler.html
I follow this page and set useCuda=true but I cannot see any cuda events in the profiler result
![image](https://user-imag…
-
### Describe the bug
I was trying to test different schedulers under DDPMPipeline. And an error occurred if I use PNDMScheduler beforehand I have found that PNDMScheduler should be compatible with DD…
-
### 🐛 Describe the bug
```
import torch
def fn(x):
result = torch.rrelu(x,0.2,0.8,training=True)
return result
x = torch.randn(4,4,dtype=torch.bfloat16,requires_grad=True)
compiled_fn =…