llama.cpp
5440a127
- iq1_s: fix dequantize on the CPU
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Commit
View On
GitHub
Commit
1 year ago
iq1_s: fix dequantize on the CPU
References
ik/even_better_iq1s
#5999 - 1.5 bit: we can do even better
Author
Iwan Kawrakow
Parents
436c65e1
Loading