feat = flash_attn.flash_attn_varlen_qkvpacked_func( AttributeError: module 'flash_attn' has no attribute 'flash_attn_varlen_qkvpacked_func'
- the varlen version is slow at the moment, please use the non-varlen version if possible. ### Limits @@ -30,10 +32,8 @@ And also because we need to save extra fp32 buffer during computation, the memor - [x] Implement `ring_flash_attn_varlen_qkvpacked_func` - [x] Impleme...