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

Nabbers1999/gemma-3-27b-it-abliterated-refined-vision

Nabbers1999 Gemma 27B
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-3-27b-it-abliterated-refined-vision"
Response includes
  • classification m1
  • files 25
  • benchmarks 16 entries
  • hub_downloads_all_time 197
  • providers 1
  • 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
197
20 last 30d - stable
Likes
1
Descendants
3
in 3 direct forks
Model age
9mo ago
created 2025-12-31
Available via
1 provider
featherless-ai
Downloads over time
Now208→from13↑1,500%
37815322813 on Dec 31, 2025208 on Oct 12Dec '25FebAprJunAugOct
Dec 31, 2025 → Oct 12 · 81 snapshots · spans 285 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 37211 LM-Arena
LM Arena Elo 1358.3955692803715 LM-Arena
Arena-Elo-Lower 1354.342781362301 LM-Arena
Arena-Elo-Upper 1362.4483571984422 LM-Arena
Arena-Rank 37 LM-Arena
Entertainment 1.1 UGI
Hazardous 2.4 UGI
Natural Intelligence 34.45 UGI
Political lean -14.2% UGI
Sensitive-Info 20.64 UGI
SocPol 3 UGI
UGI 20.43 UGI
Willingness (10) 2 UGI
W10-Adherence 0 UGI
W10-Direct 4 UGI
Writing 44.99 UGI

Genealogy 3 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-3-27b-it this lineage
Nabbers1999/gemma-3-27b-it-abliterated-refined-vision↓ 20 3 forks
mradermacher/gemma-3-27b-it-abliterated-refined-vision-i1-GGUF↓ 1,531
mradermacher/gemma-3-27b-it-abliterated-refined-vision-GGUF↓ 532
McG-221/gemma-3-27b-it-abliterated-refined-vision-mlx-8Bit↓ 175

Metadata

License
gemma
Languages
en
Tags
transformers safetensors gemma3 image-text-to-text gemma gemma-3 text-generation conversational abliterated en base_model:google/gemma-3-27b-it base_model:finetune:google/gemma-3-27b-it

Related

Other by Nabbers1999
gemma-3-27b-it-abliterated-refined-novis 118 Melpomene-70B-0307-Uncensored-GGUF 99 gemma-3-12b-it-abliterated-refined-novis 30 Qwen2.5-Coder-32B-Instruct-NP-Abliterated 27 Gemma-3-27B-it-NP-Abliterated 26
Same base model
mlabonne/gemma-3-27b-it-abliterated 7K mlabonne/gemma-3-27b-it-abliterated-GGUF 7K Thaurock/Gemma-3-27b-it-abliterated-GGUF 910 llmfan46/gemma-3-27b-it-ultra-uncensored-heretic-GGUF 517 TheCluster/gemma-3-27b-it-uncensored-mlx-4bit 442
Total size
51.1 GB
Files
25
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2026-02-01 04:49
Source huggingface.co/Nabbers1999/gemma-3-27b-it-abliterated-refined-vision ↗ Base model google/gemma-3-27b-it ↗

Files by quantization

Auxiliary files 25 files 51.1 GB
model-00004-of-00012.safetensors 4.61 GB 9db0f2d7 download
model-00005-of-00012.safetensors 4.61 GB a1f594ec download
model-00006-of-00012.safetensors 4.61 GB 0d01b62e download
model-00007-of-00012.safetensors 4.61 GB e8710457 download
model-00008-of-00012.safetensors 4.61 GB de16ae24 download
model-00009-of-00012.safetensors 4.61 GB 48a7cebf download
model-00010-of-00012.safetensors 4.61 GB d020d5ed download
model-00011-of-00012.safetensors 4.61 GB 7d038dcd download
model-00003-of-00012.safetensors 4.61 GB 9091c39e download
model-00002-of-00012.safetensors 4.61 GB 853b9c25 download
model-00001-of-00012.safetensors 4.52 GB 88bfb862 download
model-00012-of-00012.safetensors 441 MB cc1bf5e8 download
tokenizer.json 31.8 MB 4667f208 download
tokenizer.model 4.47 MB 1299c11d download
tokenizer_config.json 1.10 MB ba114d51 download
model.safetensors.index.json 124 KB 28ebe061 download
config.json 3.19 KB 5a6376da download
.gitattributes 1.53 KB 52373fe2 download
chat_template.jinja 1.50 KB 1117055a download
README.md 1.18 KB f6b24b8e download
special_tokens_map.json 662 B 1a619324 download
preprocessor_config.json 570 B b1e00fc1 download
generation_config.json 210 B 5afdcc75 download
processor_config.json 70.0 B 453c7966 download
added_tokens.json 35.0 B e17bde03 download

README current version from Hugging Face

base_model: google/gemma-3-27b-it
library_name: transformers
tags:

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

Gemma 3 27B Instruct Abliterated - Refined - Vision

This is my biprojected norm preserving abliteration of Gemma 3 27B Instruct as found at Nabbers1999/Gemma-3-27B-it-NP-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.

Due to requests, this version has been released with its vision tower intact. The projector was not trained in tandem with the prompt neutrality training, however initial testing showed it able to identify NSFW images correctly.

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

Thanks again to mradermacher for the quants.

https://huggingface.co/mradermacher/gemma-3-27b-it-abliterated-refined-vision-GGUF

https://huggingface.co/mradermacher/gemma-3-27b-it-abliterated-refined-vision-i1-GGUF

README history 3 versions

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

  1. 2026-02-01Update README.md35ae8fd1.2 KB
    Loading...
  2. 2026-01-04Update README.mde2e40bf1.2 KB
    Loading...
  3. 2025-12-31Create README.md36d40ab987 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