Q8_0 model weights less than Q2_0

#1
by ivanstepanovftw - opened

Uploaded gemma-7b-it.Q8_0.gguf file have 3.41 GB size, which is less than any other quantized models.

@ivanstepanovftw thanks for the heads up!

I'm not sure how that happened to start with, but I've reconverted the q8_0 and it's uploading now, should be available within ~5 minutes. Sorry for any inconvenience!

I'm also converting the gemma 1.1 models soon, and adding imatrix quants to go with them!

Sign up or log in to comment