Get the app
← back to catalog · registered 2026-08-31 22:03

Rin247/gemma-4-31B-it-Uncensored-Aquarion-FP4

Rin247 Gemma 15B multimodal
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/Rin247%2Fgemma-4-31B-it-Uncensored-Aquarion-FP4"
Response includes
  • classification m1
  • files 8
  • hub_downloads_all_time 31
  • author_summary 52 models
  • readme_text full
Try next
GET/models?author=Rin247
All 52 models by Rin247
 GET/models?family=gemma&method=m1
Sibling gemma models · same method
 GET/methods/m1
What "m1" means · taxonomy
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M1
Primary method

Direct removal

No other method signals detected in this model.
Confidence
MEDIUM
Why this label 3 signals
Method inferred from partial signals - repository name, related files, or tag patterns. Producer identity not confirmed; label may sharpen or shift as we gather more evidence.
  • 'abliterated' in name/tags
  • is_gguf=0 (base model)
  • no specific method indicators - defaulting to M1 (most common)
How we label → About M1 →
Refusal direction extracted via
Extraction technique

Difference-of-means

FromArditi et al. (2024) ↗
Confidence
MEDIUM
Why we say so
primary_method=M1; difference-of-means is the reference extraction for M1/M3 (Arditi 2024)
What is a refusal direction? → Read the paper ↗
Downloads · lifetime
31
12 last 30d - stable
Likes
0
Model age
6w ago
created 2026-08-31
Downloads over time
Now38→from0↑0%
01428420 on Aug 3138 on Oct 18AugSepOct
Aug 31 → Oct 18 · 49 snapshots · spans 48 days

Metadata

Tags
transformers safetensors gemma4 image-text-to-text quantized fp4 uncensored abliterated conversational endpoints_compatible 8-bit fp4_weight_only

Related

Other by Rin247
Qwen3-VL-8B-Instruct-Uncensored-Aquarion-FP4 137 Qwen3.6-35B-A3B-Uncensored-Aquarion-FP4 127 Qwen3.6-27B-Uncensored-Aquarion-FP4 68 Qwen3-VL-4B-Uncensored-Aquarion-INT4 42 Qwen3-VL-8B-Instruct-Uncensored-Aquarion-INT8 28
Total size
18.3 GB
Files
8
Quantizations
1
Registered
2026-08-31 22:03
Last updated on HF
2026-08-31 21:45
Source huggingface.co/Rin247/gemma-4-31B-it-Uncensored-Aquarion-FP4 ↗

Files by quantization

Auxiliary files 8 files 18.3 GB
model.safetensors 18.3 GB c2dc0664 download
tokenizer.json 30.7 MB 88e71407 download
config.json 45.0 KB 337f34f6 download
chat_template.jinja 18.6 KB 7d44d305 download
tokenizer_config.json 3.78 KB 1ae1a5fc download
.gitattributes 1.53 KB 52373fe2 download
README.md 768 B 5362f7f0 download
generation_config.json 248 B 69186f9d download

README current version from Hugging Face

tags:

  • quantized
  • safetensors
  • fp4
  • uncensored
  • abliterated
    library_name: transformers

gemma-4-31B-it-Uncensored-Aquarion-FP4

FP4 weight-only safetensors quantization of gemma-4-31B-it-Uncensored.

Aquarion Forge

This model was abliterated (uncensored) via orthogonal projection of the
refusal direction before quantization — part of the Genesis of Aquarion forge.

Quantization Details

  • Base model: gemma-4-31B-it-Uncensored
  • Format: FP4 (weight-only safetensors)
  • Method: PyTorch RTN quantization on CPU; scales stored alongside weights

Loading

These use custom weight-only recipes. Dequantize with the matching scale/shape
buffers (*.weight_scale, *.weight_shape) before feeding to an inference engine.

README history 1 version

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

  1. 2026-08-31Upload FP4 quantized weightsc8e27f7744 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