-
### 🐛 Describe the bug
# Problem
When running compiled FlexAttention in a multi-GPU environment, if the device being used is not the first GPU (i.e., not `cuda` or `cuda:0`, but `cuda:1`, etc.), a…
w568w updated
6 minutes ago
-
# ❓ Questions and Help
Hi, I tested memory_efficient_attention with the pytorch equivalent implementation in the doc, and found they are not exactly the same. The code:
```
def attention_e(self…
-
### 🐛 Describe the bug
Hi, I was testing FlexAttention by comparing its output with that of `nn.MultiheadAttention` and `torch.nn.functional.scaled_dot_product_attention`. In the end, I tracked down …
-
### 🐛 Describe the bug
I attempted to train LLaVA (base LLM = LLaMA 3) using the Liger kernel. The loss graph was similar to when I trained LLaVA without the Liger kernel. However, the model traine…
-
### Your current environment
```sh
python env.py
Collecting environment information...
PyTorch version: 2.6.0.dev20241011+rocm6.2
Is debug build: False
CUDA used to build PyTorch: N/A
ROCM us…
-
I was able to build flash-attention ROCM for both my Mi100 and Mi50 cards, but only got flash attention working on the Mi100(very impressive performance I might add).
Trying to run flash attention …
-
### Information about bug
The consolidate financial Statement report shows the balances in company currency, there is a filed that lets you set the currency for the report so you may view certain acc…
-
## Description
CRL downloading from peer certificate.
## Background Information / Reproduction Steps
I have this use case:
1. UA Client and UA Server each have their own certificate, and e…
-
### System Info
- CPU architecture: x86_64
- GPU properties
- GPU name: NVIDIA A100
- GPU memory size: 40G
- Libraries
- TensorRT-LLM branch or tag: v0.9.0
-…
-
When im trying to use Videocrafter 2 - i get this error :
F:\Pinokio\api\videocrafter2.git\app\env\lib\site-packages\torch\nn\functional.py:5560: UserWarning: 1Torch was not compiled with flash att…