Nice, thanks for the link! Yes, just wanted to be sure I need software update to make use of this. Looks like I may like IQ4_NL more in near future :P
AItomek ;P
altomek
AI & ML interests
MLOps, Polish LLMs
Recent Activity
new activity
about 11 hours ago
mergekit-community/QwQ-32B-Preview-Instruct-Coder:....... .....
Organizations
altomek's activity
....... .....
#1 opened about 11 hours ago
by
altomek
Huh interesting, however all inference engines need to adapt newer llama.cpp version correct? Q4_0 and IQ4_NL? Just scrolled throught the pull request. How do you know IQ4_NL should work this way also?