mradermacher's picture
auto-patch README.md
6dcbdc9 verified
|
raw
history blame
3.12 kB
metadata
exported_from: Undi95/MLewd-ReMM-L2-Chat-20B
language:
  - en
library_name: transformers
license: cc-by-nc-4.0
quantized_by: mradermacher
tags:
  - not-for-all-audiences
  - nsfw

About

weighted/imatrix quants of https://huggingface.co/Undi95/MLewd-ReMM-L2-Chat-20B

static quants are available at https://huggingface.co/mradermacher/MLewd-ReMM-L2-Chat-20B-GGUF

Usage

If you are unsure how to use GGUF files, refer to one of TheBloke's READMEs for more details, including on how to concatenate multi-part files.

Provided Quants

(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)

Link Type Size/GB Notes
GGUF i1-Q2_K 7.7 IQ3_XXS probably better
GGUF i1-IQ3_XXS 7.9 lower quality
GGUF i1-Q3_K_S 9.0 IQ3_XS probably better
GGUF i1-Q3_K_M 10.0 IQ3_S probably better
GGUF i1-Q3_K_L 10.9 IQ3_M probably better
GGUF i1-Q4_0 11.6
GGUF i1-Q4_K_S 11.7 optimal size/speed/quality
GGUF i1-Q4_K_M 12.3 fast, recommended
GGUF i1-Q5_K_S 14.1
GGUF i1-Q5_K_M 14.5
GGUF i1-Q6_K 16.7 practically like static Q6_K

Here is a handy graph by ikawrakow comparing some lower-quality quant types (lower is better):

image.png

And here are Artefact2's thoughts on the matter: https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9

Thanks

I thank my company, nethype GmbH, for letting me use its servers and providing upgrades to my workstation to enable this work in my free time.