Get the app
← back to catalog · registered 2026-08-22 13:56

TheCluster/Gemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-Thinking-MLX-mxfp4

TheCluster Gemma 31B multimodal second-order
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model. 

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/TheCluster%2FGemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-Thinking-MLX-mxfp4"
Response includes
  • classification m3
  • files 14
  • benchmarks 11 entries
  • hub_downloads_all_time 1,766
  • author_summary 28 models
  • readme_text full
Try next
GET/models?author=TheCluster
All 28 models by TheCluster
 GET/models?family=gemma&method=m3
Sibling gemma models · same method
 GET/methods/m3
What "m3" means · taxonomy
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M3
Primary method

Layer-wise ablation

Applied on top of direct removal inherited from the base model.
Confidence
HIGH
Inherited from base model
M1 Direct removal →
Why this label 2 signals
Producer identity confirmed by naming conventions, tags or the model card. This label is very unlikely to change.
  • 'heretic' in model name (Heretic-produced)
  • Heretic uses layer-wise optimization (M3) with underlying direction removal (M1)
How we label → About M3 →
Refusal direction extracted via
Extraction technique

Difference-of-means

FromArditi et al. (2024) ↗
Confidence
HIGH
Why we say so
name contains 'heretic'; Heretic default extraction is difference-of-means (Arditi 2024)
What is a refusal direction? → Read the paper ↗
Downloads · lifetime
2K
92 last 30d - cooling
Likes
0
Model age
5mo ago
created 2026-04-16
Downloads over time
Now1.8K→from246↑629%
1697621.4K1.9K246 on Apr 151.8K on Oct 111.8K on Oct 10AprMayJunJulAugSepOct
Apr 15 → Oct 11 · 65 snapshots · spans 179 days

Benchmarks

Benchmark Score Source
Entertainment 2.8 UGI
Hazardous 4.7 UGI
Natural Intelligence 34.42 UGI
Political lean -22.2% UGI
Sensitive-Info 32.29 UGI
SocPol 2.6 UGI
UGI 52.36 UGI
Willingness (10) 9.2 UGI
W10-Adherence 8.5 UGI
W10-Direct 10 UGI
Writing 38.68 UGI

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

DavidAU/gemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-Thinking↓ 540 this lineage
TheCluster/Gemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-Thinking-MLX-mxfp4↓ 92

Metadata

License
apache-2.0
Languages
en zh ru es fr it ja ko af de ar tr is pl sw sv nl he id uk fa pa pt ms fi el
Tags
mlx safetensors gemma4 heretic uncensored unrestricted decensored abliterated creative creative writing fiction writing plot generation

Related

Other by TheCluster
Qwen3.5-9B-Claude-4.6-HighIQ-INSTRUCT-HERETIC-UNCENSORED-MLX-mxfp8 7K Qwen3.5-9B-Uncensored-HauhauCS-Aggressive-MLX-mxfp4 7K Qwen3.5-35B-A3B-Uncensored-HauhauCS-Aggressive-MLX-mxfp4 1K Qwen3.5-9B-Uncensored-HauhauCS-Aggressive-MLX-mxfp8 1K Qwen3.8-27B-ABLITERATED-UNCENSORED-PHILADELPHIA-CLASS-MLX-mixed-9.4bit 916
Same base model
mradermacher/gemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-Thinking-i1-GGUF 46K AEON-7/Gemma-4-31B-it-DECKARD-HERETIC-Uncensored-NVFP4 7K AEON-7/Gemma-4-31B-it-DECKARD-HERETIC-Uncensored-NVFP4-SVDQuant 7K mradermacher/gemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-Thinking-GGUF 4K connstruct/gemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-Thinking 545
Total size
26.5 GB
Files
14
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2026-04-16 05:01
Source huggingface.co/TheCluster/Gemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-Thinking-MLX-mxfp4 ↗ Base model DavidAU/gemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-Thinking →

Files by quantization

Auxiliary files 14 files 26.6 GB
model-00001-of-00006.safetensors 4.99 GB 26d4e9d5 download
model-00003-of-00006.safetensors 4.96 GB 6fdffdc4 download
model-00005-of-00006.safetensors 4.96 GB e1abaf6f download
model-00004-of-00006.safetensors 4.96 GB acbc0655 download
model-00002-of-00006.safetensors 4.96 GB b70becf5 download
model-00006-of-00006.safetensors 1.73 GB 195874fc download
tokenizer.json 30.7 MB cc8d3a0c download
model.safetensors.index.json 178 KB c29b3d70 download
config.json 47.7 KB c9ec547c download
chat_template.jinja 16.1 KB df3629f1 download
tokenizer_config.json 2.62 KB 59dd4b62 download
.gitattributes 1.53 KB 52373fe2 download
README.md 1.20 KB b9d8914f download
processor_config.json 902 B 13e92a44 download

README current version from Hugging Face

license: apache-2.0
base_model:

  • DavidAU/gemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-Thinking
    pipeline_tag: image-text-to-text
    library_name: mlx
    tags:
  • heretic
  • uncensored
  • unrestricted
  • decensored
  • abliterated
  • creative
  • creative writing
  • fiction writing
  • plot generation
  • sub-plot generation
  • fiction writing
  • story generation
  • scene continue
  • storytelling
  • fiction story
  • science fiction
  • romance
  • all genres
  • story
  • writing
  • vivid prosing
  • vivid writing
  • fiction
  • roleplaying
  • mxfp4
    language:
  • en
  • zh
  • ru
  • es
  • fr
  • it
  • ja
  • ko
  • af
  • de
  • ar
  • tr
  • is
  • pl
  • sw
  • sv
  • nl
  • he
  • id
  • uk
  • fa
  • pa
  • pt
  • ms
  • fi
  • el

Gemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-Thinking

Quality: quantized (mxfp4, 7.298 bpw)

Gemma 4 31B first "Heretic'ed" (de-censored), then fine tuned by DavidAU (Via Unsloth) on "The Deckard" dataset.

Source

This model was converted to MLX format from DavidAU/gemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-Thinking using mlx-vlm version 0.4.4.

README history 2 versions

The author's README evolved over time. Click a version to see its content at that point.

  1. 2026-04-16Update README.md9eb40871.2 KB
    Loading...
  2. 2026-04-16initial commita699a9d28 B
    Loading...
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Get the app Browse the catalog
Open in Abliteration