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

grimjim/gemma-3-12b-it-projection-abliterated

grimjim Gemma 12B 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/grimjim%2Fgemma-3-12b-it-projection-abliterated"
Response includes
  • classification m1
  • files 18
  • benchmarks 16 entries
  • hub_downloads_all_time 458
  • author_summary 14 models
  • readme_text full
Try next
GET/models?author=grimjim
All 14 models by grimjim
 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
458
63 last 30d - stable
Likes
10
Descendants
4
in 4 direct forks
Model age
11mo ago
created 2025-10-28
Downloads over time
Now491→from86↑471%
6622137653286 on Oct 29, 2025491 on Oct 11Oct '25Dec '25FebAprJunAugOct
Oct 29, 2025 → Oct 11 · 89 snapshots · spans 347 days

Benchmarks

Benchmark Score Source
Entertainment 0.9 UGI
Hazardous 1.2 UGI
Natural Intelligence 19.21 UGI
Political lean -6.9% UGI
Sensitive-Info 9.92 UGI
SocPol 1 UGI
UGI 39.11 UGI
Willingness (10) 9.8 UGI
W10-Adherence 9.5 UGI
W10-Direct 10 UGI
Writing 29.46 UGI
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

Genealogy 4 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
grimjim/gemma-3-12b-it-projection-abliterated↓ 63 4 forks
mradermacher/gemma-3-12b-it-projection-abliterated-i1-GGUF↓ 1,090
mradermacher/gemma-3-12b-it-projection-abliterated-GGUF↓ 508
CallMcMargin/gemma-3-12b-it-projection-abliterated-mlx-bf16-mxfp4-mixed-4-6↓ 44
ArtusDev/grimjim_gemma-3-12b-it-projection-abliterated-EXL3↓ 21

Metadata

License
gemma
Tags
transformers safetensors gemma3 image-text-to-text conversational base_model:google/gemma-3-12b-it base_model:finetune:google/gemma-3-12b-it license:gemma text-generation-inference endpoints_compatible region:us

Related

Other by grimjim
Llama-3.1-8B-Instruct-abliterated_via_adapter-GGUF 592 gemma-3-12b-it-abliterated-GGUF 139 gemma-3-12b-it-norm-preserved-biprojected-abliterated 88 Llama-3.1-8B-Instruct-abliterated_via_adapter 74 mistralai-Mistral-Nemo-Instruct-2407-extensive-BP-abliteration-12B 49
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
22.7 GB
Files
18
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2025-10-30 02:27
Source huggingface.co/grimjim/gemma-3-12b-it-projection-abliterated ↗ Base model google/gemma-3-12b-it ↗

Files by quantization

Auxiliary files 18 files 22.7 GB
model-00001-of-00005.safetensors 4.64 GB 4847447e download
model-00003-of-00005.safetensors 4.59 GB fc966ae5 download
model-00004-of-00005.safetensors 4.59 GB 57d41c37 download
model-00002-of-00005.safetensors 4.59 GB 05bcadbd download
model-00005-of-00005.safetensors 4.29 GB 40cbfdcf download
tokenizer.json 31.8 MB 4667f208 download
tokenizer.model 4.47 MB 1299c11d download
tokenizer_config.json 1.15 MB 2cfe89ab download
model.safetensors.index.json 107 KB 25d054c8 download
config.json 2.94 KB 96e2ebbc download
chat_template.jinja 1.54 KB c5f13654 download
.gitattributes 1.53 KB 52373fe2 download
README.md 727 B 1bcbc659 download
special_tokens_map.json 695 B 6728103d download
preprocessor_config.json 599 B da332e16 download
generation_config.json 223 B 50cbbd78 download
processor_config.json 74.0 B bcc0e8fd download
added_tokens.json 38.0 B f9f1f4f5 download

README current version from Hugging Face

license: gemma
library_name: transformers
pipeline_tag: image-text-to-text
base_model: google/gemma-3-12b-it

gemma-3-12b-it-projection-abliterated

This model was derived from google/gemma-3-12b-it.

Projected abliteration has been applied; no subsequent fine-tuning was applied to repair damage.
The net result is a model that refuses far less often, but still retains awareness of safety and harms.

This model recently benched with a WC/10 rating of 9.8 on the UGI Leaderboard, tieing for first place in compliance.

More details to follow.

README history 3 versions

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

  1. 2025-10-30Update README.md1c0a98e727 B
    Loading...
  2. 2025-10-29Update README.md6871196728 B
    Loading...
  3. 2025-10-28Create README.md6c08875468 B
    Loading...

Discussions 2 threads

  1. 2025-10-31Commendationsopen3 💬#2
    Loading...
  2. 2025-10-29Impressive workopen1 💬#1
    Loading...

Continue on Hugging Face →

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