dan_the_3rd

Hi, Why did you try to downgrade `gcc`? Also, what cuda version are you using ("nvcc --version")

Hi, I believe your batch size is too big. Can you try something smaller than `65536`?

The reason is that CUDA kernels are parallelized across 3 dimensions at most, and we chose to parallelize across batch size in the y or z dimension (I don't recall...

Hi Thanks for reporting this bug. At the moment, we only support python 3.8, 3.9 and 3.10. As you have python 3.11, you need to build from source (or change...

@1121301057 xFormers does not work on MacOS, as CUDA no longer supports MacOS

0.0.20 has just been released with python 3.11 support - let us know if you have issues :)

That is correct, reopening

Hi, Can you post the entire stacktrace? In the one you show there is nothing from xformers there. Can you also post the output of the following command? ``` python...

> It seems that the flop is incorrect when query's length is different from key-value' s length Yes indeed, you are right. I guess we also need to distinguish between...

What do you mean? What GPU are you building xFormers for? What are you trying to do?