DeepSpeed
64bd29d6
- fix: materialize concat-sliced k/v for decode_attn contiguity (Loop 9b)
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Commit
View On
GitHub
Commit
6 days ago
fix: materialize concat-sliced k/v for decode_attn contiguity (Loop 9b) Signed-off-by: Guokai Ma <guokai.ma@intel.com>
Author
delock
Parents
e1baace7
Loading