Dark Hacker News
new
|
best
|
ask
|
show
|
jobs
Jagged Flash Attention Optimization
(shaped.ai)
24 points
by
tullie
1 year ago
| 3 comments
Jagged Flash Attention Optimization | Dark Hacker News
platers
1 year ago
|
next
[−]
Flash attention natively supports packing multiple variable length sequences into a single call, what is the advantage of jagged flash attention?
bbstats
1 year ago
|
parent
|
next
[−]
If only there was a link to a page somewhere that could answer this question for you.
CapsAdmin
1 year ago
|
next
[−]
See also
https://github.com/thu-ml/SageAttention
and
https://github.com/thu-ml/SpargeAttn