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

Nabbers1999/gemma-3-12b-it-abliterated-refined-novis

Nabbers1999 Gemma 12B
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/Nabbers1999%2Fgemma-3-12b-it-abliterated-refined-novis"
Response includes
  • classification m1
  • files 17
  • benchmarks 16 entries
  • hub_downloads_all_time 1,171
  • author_summary 8 models
  • readme_text full
Try next
GET/models?author=Nabbers1999
All 8 models by Nabbers1999
 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
1K
29 last 30d - cooling
Likes
3
Descendants
2
in 2 direct forks
Model age
9mo ago
created 2025-12-29
Downloads over time
Now1.2K→from74↑1,501%
184448701.3K74 on Dec 31, 20251.2K on Oct 111.2K on Oct 9Dec '25FebAprJunAugOct
Dec 31, 2025 → Oct 11 · 80 snapshots · spans 284 days

Benchmarks

Portrait before abliteration
Benchmarks of the base model as it stood before the refusal-removal operation. Compare with the numbers above to see what the operation cost.
Benchmark Score Source
Arena-Battles 3976 LM-Arena
LM Arena Elo 1335.3304642871612 LM-Arena
Arena-Elo-Lower 1326.1060034720686 LM-Arena
Arena-Elo-Upper 1344.5549251022537 LM-Arena
Arena-Rank 49 LM-Arena
Entertainment 1.3 UGI
Hazardous 2.9 UGI
Natural Intelligence 18.72 UGI
Political lean -11.7% UGI
Sensitive-Info 16.33 UGI
SocPol 1 UGI
UGI 20.89 UGI
Willingness (10) 3 UGI
W10-Adherence 0 UGI
W10-Direct 6 UGI
Writing 29.86 UGI

Genealogy 2 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.

google/gemma-3-12b-it this lineage
Nabbers1999/gemma-3-12b-it-abliterated-refined-novis↓ 29 2 forks
mradermacher/gemma-3-12b-it-abliterated-refined-novis-i1-GGUF↓ 761
mradermacher/gemma-3-12b-it-abliterated-refined-novis-GGUF↓ 210

Metadata

License
gemma
Languages
en
Tags
transformers safetensors gemma3_text text-generation gemma gemma-3 conversational abliterated en base_model:google/gemma-3-12b-it base_model:finetune:google/gemma-3-12b-it license:gemma

Related

Other by Nabbers1999
gemma-3-27b-it-abliterated-refined-novis 118 Melpomene-70B-0307-Uncensored-GGUF 99 Gemma-3-27B-it-NP-Abliterated 27 Qwen2.5-Coder-32B-Instruct-NP-Abliterated 25 gemma-3-27b-it-abliterated-refined-vision 21
Same base model
mlabonne/gemma-3-12b-it-abliterated-GGUF 3K Andycurrent/gemma-3-12b-it-uncensored-GGUF 1K mlabonne/gemma-3-1b-it-abliterated-GGUF 749 mlabonne/gemma-3-12b-it-abliterated-v2 413 mlabonne/gemma-3-12b-it-abliterated 313
Total size
21.9 GB
Files
17
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2026-02-01 04:49
Source huggingface.co/Nabbers1999/gemma-3-12b-it-abliterated-refined-novis ↗ Base model google/gemma-3-12b-it ↗

Files by quantization

Auxiliary files 17 files 22.0 GB
model-00003-of-00005.safetensors 4.59 GB 3b4183cc download
model-00004-of-00005.safetensors 4.59 GB 8ac0f04b download
model-00002-of-00005.safetensors 4.59 GB 117c8985 download
model-00005-of-00005.safetensors 4.29 GB a2f0d738 download
model-00001-of-00005.safetensors 3.85 GB 57a17a6b download
tokenizer.json 31.8 MB 4667f208 download
tokenizer.model 4.47 MB 1299c11d download
tokenizer_config.json 1.10 MB 9c2583b0 download
model.safetensors.index.json 51.4 KB 0b3247b9 download
chat_template.json 1.58 KB 719b0cd0 download
.gitattributes 1.53 KB 52373fe2 download
chat_template.jinja 1.50 KB 1117055a download
config.json 1.05 KB 7a61f4c9 download
README.md 885 B e200fb15 download
special_tokens_map.json 670 B bdd437b8 download
generation_config.json 210 B c4400ddc download
added_tokens.json 35.0 B e17bde03 download

README current version from Hugging Face

base_model: google/gemma-3-12b-it
library_name: transformers
tags:

  • gemma
  • gemma-3
  • text-generation
  • conversational
  • abliterated
    license: gemma
    language:
  • en

Gemma 3 12B Instruct Abliterated - Refined

This is grimjim's norm preserving biprojected abliteration of Gemma 3 12B Instruct as found at grimjim/gemma-3-12b-it-biprojected-abliterated.

This model has undergone further DoRA fine tuning to encourage response neutrality, and in early tests has shown no difference in behavior between toxic and harmless prompts even with no system prompt given.

For further measure, this variant has had its vision features removed and has been reverted to a text only model.

This is an experimental model with no safety features. Use it at your own risk. You are responsible for what you do with it.

README history 2 versions

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

  1. 2026-02-01Update README.mdc7ed572885 B
    Loading...
  2. 2025-12-29Create README.md2caf94b880 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