Ovis icon indicating copy to clipboard operation
Ovis copied to clipboard

How to inference on V100 without Flash Attention?

Open SkywalkerSZW opened this issue 4 months ago • 1 comments

I only have V100S GPU, and I got "RuntimeError: FlashAttention only supports Ampere GPUs or newer.". How to inference on V100 without Flash Attention?

SkywalkerSZW avatar Aug 21 '25 02:08 SkywalkerSZW

The same problem, I hope ovis2.5 can use eager.

EdmunddzzZ avatar Aug 21 '25 18:08 EdmunddzzZ