Naphula / Model details

Cyberdonia-24B-v8

Naphula/Cyberdonia-24B-v8

merged

Read the original model card ↗ · Choosing a model

About this model

Naphula's Cyberdonia-24B-v8 is a safetensors text-generation merge created with the mergekit-exp della method. The model card lists four sources: anthracite-core/Mistral-Small-3.2-24B-Instruct-2506-Text-Only, BeaverAI/Cydonia-24B-v4x-GGUF, BeaverAI/Cydonia-24B-v4y-GGUF, and TheDrummer/Cydonia-24B-v4.3. The creator describes it as fully uncensored with zero refusals, achieved without Heretic or abliteration through the un-normalized magnitude of the merge. The creator warns that it can produce violent and graphic erotic content in narratives and roleplay, and says to adjust the system prompt and use the Mistral Tekken chat template. The repository is Apache-2.0 licensed.

Description sources

https://huggingface.co/Naphula/Cyberdonia-24B-v8/raw/7ddacd1ada4c46c5886c70ce61c3d3e137a38c27/README.md ↗

Source-based description, not an independent evaluation. Check the linked sources for the creator’s claims and limitations.

Size and formats

Base model: BeaverAI/Cydonia-24B-v4x-GGUF, BeaverAI/Cydonia-24B-v4y-GGUF, TheDrummer/Cydonia-24B-v4.3, anthracite-core/Mistral-Small-3.2-24B-Instruct-2506-Text-Only
Parameters: 23.5724B
Main repository formats: safetensors
License tag: apache-2.0
Languages: en
Last updated: 2026-10-15 · 03:06:31.000 UTC

Community reviews & tips

Source-linked paraphrases of sampled community sources, not independently tested by this directory.

Sample window: 2025-10-27 UTC to 2026-10-27 UTC · 2 unique source links. This is a selected sample, not all community discussions.

Small sample: source links are not a count of independent reviewers or evidence of consensus.

Sources checked: 2026-10-27 00:39:59 UTC

Hands-on report
- One user running the imatrix Q4_0 quant (Q8_0 KV cache, ~20k context on an NPU) reports the model is creative, very uncensored, free of obvious Mistralisms, and strong at contextual ethical realignment; occasional safety disclaimers usually clear with a single regeneration. It chronically exceeds requested word counts and can drop minor details or threads in follow-up turns, but prose quality holds up over long generations.
Tested revision: Unknown · Quantization: Unknown
- Hugging Face discussion · 2026-10-15 UTC ↗
Practical tip
- The owner explains the decensoring recipe behind this merge: combine low-refusal donors with the della method, normalize: false, and total weights above 1.0 (around 1.2-1.5 lowers refusals; this model uses 1.6; past 2.0 distorts output). Setting normalize: true brings refusals back, and the trick does not currently work for moe_della merges, which produced broken output in testing.
Tested revision: Unknown · Quantization: Unknown
- Hugging Face discussion · 2026-10-16 UTC ↗

What changed

Model types are creator-reported, not independently tested. These labels describe source metadata, not a guarantee of behavior.

mergedsource tag: base_model:merge:BeaverAI/Cydonia-24B-v4x-GGUF
mergedsource tag: base_model:merge:BeaverAI/Cydonia-24B-v4y-GGUF
mergedsource tag: base_model:merge:TheDrummer/Cydonia-24B-v4.3
mergedsource tag: base_model:merge:anthracite-core/Mistral-Small-3.2-24B-Instruct-2506-Text-Only

Family evidence

These groupings are source-reported. Explicit base identifiers take priority over generic architecture tags.

Mistral / Mixtral: Base model: anthracite-core/Mistral-Small-3.2-24B-Instruct-2506-Text-Only

Available versions

Versions appear here only when they explicitly name this exact model as their source. Check each repository for available files.

Main repository files ↗safetensors
No other explicitly linked versions recorded.

Link to this listing

Made or quantized this model? Paste one of these into your model card or README so people can find this listing.

Markdown

[Listed on Custom Clankers](https://customclankers.com/models/naphula-cyberdonia-24b-v8/)

HTML

<a href="https://customclankers.com/models/naphula-cyberdonia-24b-v8/">Listed on Custom Clankers</a>

Plain text links only: no image, no script, nothing loaded from this site.

Customize or build on this model

Listed from public GitHub data, not tested by us. Read the code and license before you run anything.

Tools to customize it further

Remove more refusals

Heretic ↗ ★ 36,553
Removes refusal behavior from transformer models automatically. It pairs directional ablation (abliteration) with an Optuna parameter search that tries to cut refusals while keeping the model close to the original, so you don't have to pick layers by hand.
OBLITERATUS ↗ ★ 8,966
A larger refusal-removal toolkit with several abliteration methods, usable locally or through a hosted Hugging Face Space. Check the telemetry setting before you run it: the README describes telemetry as opt-in, but on by default in the Spaces version.

Fine-tune it on your own data

Unsloth ↗ ★ 78,592
A desktop app and Python library for running and training models locally. Its README lists LoRA, QLoRA, full fine-tuning, DPO and GRPO, plus export to formats including GGUF, on Windows, Linux, WSL and macOS.
LlamaFactory ↗ ★ 75,742
Fine-tunes more than 100 language and vision-language models from a command line or a browser-based Web UI, aimed at training without writing code.

Make a smaller download (quantize)

llama.cpp ↗ ★ 132,430
Runs models locally and includes the tools behind most GGUF files. A conversion script turns Hugging Face weights into GGUF, and llama-quantize shrinks a high-precision GGUF to lower-bit formats, which can cost some accuracy.
MLX LM ↗ ★ 7,479
Apple's MLX package for running and fine-tuning models on Apple silicon. It can quantize models and upload them to the Hugging Face Hub, and supports LoRA and full fine-tuning.

Merge it with another model

mergekit ↗ ★ 7,418
Merges pretrained models with many merge methods. It works out-of-core, so merges can run on a CPU or with as little as 8 GB of VRAM, and it also handles LoRA extraction and mixture-of-experts merges.

Learn the techniques

LLM Course ↗ ★ 84,721
A free course with roadmaps and Colab notebooks. Its notebooks include one-click tools for merging with MergeKit, quantizing to GGUF, GPTQ, EXL2, AWQ and HQQ, and automatic abliteration with custom datasets.

Compare all tools →

Source provenance

Downloads: 1,209
Likes: 8
Metadata checked: 2026-10-27 · 00:15:15 UTC
First seen in index: 2026-10-27 · 00:15:15 UTC
Source revision: 7ddacd1ada4c46c5886c70ce61c3d3e137a38c27

Source-card excerpt

Source model card excerpt: [!IMPORTANT] ⚠️ Warning: This model can produce narratives and RP that contain violent and graphic erotic content. Adjust your system prompt accordingly, and use

From Naphula/Cyberdonia-24B-v8. Creator model-card excerpt, not an independent review.

Inspect before you run

Low refusal is not proof of correctness. Outputs can be inaccurate, harmful, or biased. Review the model card, license and files yourself. A license tag is not commercial-use advice. Parameter counts are source metadata; no hardware compatibility is implied.

This directory does not host or redistribute model weights. Refusal profiles are measured by us on a fixed prompt set; all other behavior claims are source-reported.

Collection rules & limitations ↗