mirror of
https://github.com/ikawrakow/ik_llama.cpp.git
synced 2026-02-26 08:04:09 +00:00
PPL(LLaMA-3.1-8B-Instruct, 8192) is now 6.7689 after shrinking by 0.015 bpw by using iq4_k instead of q5_k for attn_v.
PPL(LLaMA-3.1-8B-Instruct, 8192) is now 6.7689 after shrinking by 0.015 bpw by using iq4_k instead of q5_k for attn_v.