-
# 🐛 Bug
I am trying to use `memory_efficient_attention` with `torch.compile()`. But it seems that `memory_efficient_attention` leads to graph breaks.
`xformers.ops.unbind` also causes graph breaks…
-
This is in a conda environment on Python 3.9.16.
```
xformers\csrc\swiglu\swiglu_op.cpp(15): error C2039: 'optional': is not a member of 'std'
C:\apps\conda\envs\voicecraft\lib\site-packages\to…
-
```
PS D:\Work\Pipeline\LGM> python app.py big --resume pretrained/model_fp16.safetensors
Traceback (most recent call last):
File "D:\Work\Pipeline\LGM\app.py", line 6, in
import torch
F…
-
todo
-
First of all, thank you so much for your repository! It's very helpful in prototyping different models.
I had a question about the MOIRAI impelmemtantion
It seems that the [`TimerLayer`](https:…
-
I'm glad that I paid $5 for the 1 click destroy my pc instead of just doing via pip.
-
### System Info
GPU: `A10`
Base Image: `FROM nvidia/cuda:12.1.0-runtime-ubuntu22.04`
Tensorrt-llm:
- `0.12.0` : It's working, but I can't use it because of a version mismatch in TRT and trt-llm-back…
-
```
(venv) C:\svd-temporal-controlnet>python run_inference.py
WARNING[XFORMERS]: xFormers can't load C++/CUDA extensions. xFormers was built for:
PyTorch 2.1.2+cu121 with CUDA 1201 (you have 2.…
-
# 🚀 Feature
Build xformers with nvcc lto for a potential speed boost.
The motivation is simple... more speed!
## Additional context
Adding `NVCC_APPEND_FLAGS="--gpu-architecture=lto_75 --gp…
-
# 🐛 Bug
It seems CUDA arch 8.0+PTX is not properly detected by xformers.
It matches only on 8.0, but not 8.0+PTX and so produces the error:
requires device with capability > (8, 0) but your GP…