Fishfishfishfishfish's picture
Update README.md
f3e69d1 verified
metadata
license: gemma
language:
  - en
base_model: google/gemma-2-2b-it

Gemma 2 2B quantized for wllama (under 2gb).

q4_0_4_8 is WAY faster when using llama.cpp, with wllama, it's about the same as q4_k.