Slim attention: cut your context memory in half without loss of accuracy -- K-cache is all you need for MHA
Paper
•
2503.05840
•
Published
•
3
gräfics
graefics
·
AI & ML interests
None yet
Recent Activity
upvoted
a
paper
about 2 months ago
Slim attention: cut your context memory in half without loss of accuracy
-- K-cache is all you need for MHA
upvoted
a
paper
3 months ago
Flash normalization: fast RMSNorm for LLMs
Organizations
upvoted
a
paper
about 2 months ago
upvoted
an
article
3 months ago
Article
Topic 33: Slim Attention, KArAt, XAttention and Multi-Token Attention Explained – What’s Really Changing in Transformers?
By
and 1 other
•
•
14
upvoted
a
paper
3 months ago