mradermacher's picture
auto-patch README.md
8349dd2 verified
metadata
base_model: stabilityai/stable-code-3b
datasets:
  - tiiuae/falcon-refinedweb
  - bigcode/the-stack-github-issues
  - bigcode/commitpackft
  - bigcode/starcoderdata
  - EleutherAI/proof-pile-2
  - meta-math/MetaMathQA
language:
  - en
library_name: transformers
license: other
quantized_by: mradermacher
tags:
  - causal-lm
  - code

About

weighted/imatrix quants of https://huggingface.co/stabilityai/stable-code-3b

static quants are available at https://huggingface.co/mradermacher/stable-code-3b-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-IQ1_S 0.8 for the desperate
GGUF i1-IQ1_M 0.8 mostly desperate
GGUF i1-IQ2_XXS 0.9
GGUF i1-IQ2_XS 1.0
GGUF i1-IQ2_S 1.0
GGUF i1-IQ2_M 1.1
GGUF i1-Q2_K 1.2 IQ3_XXS probably better
GGUF i1-IQ3_XXS 1.2 lower quality
GGUF i1-IQ3_XS 1.3
GGUF i1-IQ3_S 1.4 beats Q3_K*
GGUF i1-Q3_K_S 1.4 IQ3_XS probably better
GGUF i1-IQ3_M 1.4
GGUF i1-Q3_K_M 1.5 IQ3_S probably better
GGUF i1-Q3_K_L 1.6 IQ3_M probably better
GGUF i1-IQ4_XS 1.6
GGUF i1-Q4_0 1.7 fast, low quality
GGUF i1-Q4_K_S 1.7 optimal size/speed/quality
GGUF i1-Q4_K_M 1.8 fast, recommended
GGUF i1-Q5_K_S 2.0
GGUF i1-Q5_K_M 2.1
GGUF i1-Q6_K 2.4 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

FAQ / Model Request

See https://huggingface.co/mradermacher/model_requests for some answers to questions you might have and/or if you want some other model quantized.

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. Additional thanks to @nicoboss for giving me access to his private supercomputer, enabling me to provide many more imatrix quants, at much higher quality, than I would otherwise be able to.