Modelpublic

alexokita/Darkhn-M3.2-36B-Animus-V12.0-Heretic-Uncensored-i1-GGUF

sourceHugging Faceapache-2.0updated 29d agoView on Hugging Face

1likes5.1kdownloads

Model Card

About

weighted/imatrix quants of https://huggingface.co/Silicone-Moss/Darkhn-M3.2-36B-Animus-V12.0-Heretic-Uncensored

Dense Mistral 36B (Heretic trial 177: 8/100 refusals, KL 0.0200). Text-only — no mmproj.

These are i1 (importance-matrix) GGUFs in the mradermacher naming style, converted with llama.cpp on an NVIDIA GB10. Calibration: froggeric/imatrix `groups_merged.txt` (llama-imatrix -c 512).

Original weights: `Darkhn/M3.2-36B-Animus-V12.0`. Abliteration: p-e-w/heretic.

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	imatrix	0.02	imatrix file (for creating your own quants)
GGUF	i1-IQ3_M	15.5	smallest in this pack
GGUF	i1-IQ4_XS	18.7
GGUF	i1-Q4KS	19.8	optimal size/speed/quality
GGUF	i1-Q4KM	21.0	fast, recommended
GGUF	i1-Q5KS	23.9
GGUF	i1-Q5KM	24.6
GGUF	i1-Q6_K	28.5	practically like static Q6_K
GGUF	i1-Q8_0	36.9	high quality; imatrix unused

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

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

Thanks

Source weights by Silicone-Moss / Darkhn. Heretic by p-e-w. Conversion with llama.cpp.