mradermacher's picture
auto-patch README.md
f898949 verified
|
raw
history blame
3.46 kB
metadata
base_model:
  - CultriX/MonaTrix-v4
  - mlabonne/OmniTruthyBeagle-7B-v0
  - CultriX/MoNeuTrix-7B-v1
  - paulml/OmniBeagleSquaredMBX-v3-7B
exported_from: CultriX/NeuralMona_MoE-4x7B
language:
  - en
library_name: transformers
license: apache-2.0
quantized_by: mradermacher
tags:
  - moe
  - frankenmoe
  - merge
  - mergekit
  - lazymergekit
  - CultriX/MonaTrix-v4
  - mlabonne/OmniTruthyBeagle-7B-v0
  - CultriX/MoNeuTrix-7B-v1
  - paulml/OmniBeagleSquaredMBX-v3-7B

About

weighted/imatrix quants of https://huggingface.co/CultriX/NeuralMona_MoE-4x7B

static quants are available at https://huggingface.co/mradermacher/NeuralMona_MoE-4x7B-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 9.1 IQ3_XXS probably better
GGUF i1-IQ3_XXS 9.6 lower quality
GGUF i1-Q3_K_S 10.7 IQ3_XS probably better
GGUF i1-Q3_K_M 11.8 IQ3_S probably better
GGUF i1-Q3_K_L 12.8 IQ3_M probably better
GGUF i1-IQ4_XS 13.1
GGUF i1-Q4_0 13.9
GGUF i1-Q4_K_S 14.0 optimal size/speed/quality
GGUF i1-Q4_K_M 14.9 fast, recommended
GGUF i1-Q5_K_S 16.9
GGUF i1-Q5_K_M 17.4
GGUF i1-Q6_K 20.1 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.