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

loay/gemma-4-E4B-it-abliterated-en-ar-text-only

loay Gemma 6.4B 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/loay%2Fgemma-4-E4B-it-abliterated-en-ar-text-only"
Response includes
  • classification m1
  • files 15
  • hub_downloads_all_time 53
  • author_summary 1 models
  • readme_text full
Try next
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
53
10 last 30d - stable
Likes
1
Model age
5mo ago
created 2026-05-02
Downloads over time
Now57→from9↑533%
72543629 on May 657 on Oct 1157 on Oct 8MayJunJulAugSepOct
May 6 → Oct 11 · 62 snapshots · spans 158 days

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.

WWTCyberLab/gemma-4-E4B-it-abliterated↓ 53 this lineage
loay/gemma-4-E4B-it-abliterated-en-ar-text-only↓ 10

Metadata

License
gemma
Languages
en ar
Tags
safetensors gemma4 text-generation text-only pruned conversational en ar base_model:WWTCyberLab/gemma-4-E4B-it-abliterated base_model:finetune:WWTCyberLab/gemma-4-E4B-it-abliterated license:gemma region:us

Related

Same base model
mradermacher/gemma-4-E4B-it-abliterated-i1-GGUF 2K mradermacher/gemma-4-E4B-it-abliterated-GGUF 448
Total size
11.9 GB
Files
15
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2026-05-02 21:30
Source huggingface.co/loay/gemma-4-E4B-it-abliterated-en-ar-text-only ↗ Base model WWTCyberLab/gemma-4-E4B-it-abliterated →

Files by quantization

Auxiliary files 15 files 11.9 GB
model-00001-of-00006.safetensors 4.36 GB 1720b040 download
model-00005-of-00006.safetensors 1.54 GB 6cf04e9f download
model-00004-of-00006.safetensors 1.51 GB a4434b4f download
model-00002-of-00006.safetensors 1.51 GB f2a34e80 download
model-00003-of-00006.safetensors 1.51 GB 83f69028 download
model-00006-of-00006.safetensors 1.44 GB 381200a5 download
tokenizer.json 22.8 MB 89ebaf05 download
model.safetensors.index.json 69.5 KB 1a7ff0b9 download
chat_template.jinja 11.6 KB afb1d517 download
config.json 3.24 KB 96e1845e download
pruning_report.json 3.19 KB 974701ad download
tokenizer_config.json 2.62 KB db5ec48f download
.gitattributes 1.53 KB 52373fe2 download
README.md 778 B 1d0d7f0d download
generation_config.json 203 B edda3c19 download

README current version from Hugging Face

license: gemma
base_model: WWTCyberLab/gemma-4-E4B-it-abliterated
language:

  • en
  • ar
    tags:
  • gemma4
  • text-generation
  • text-only
  • pruned
    private: true

gemma-4-E4B-it-abliterated - English/Arabic Text-Only Prune

Private derivative of WWTCyberLab/gemma-4-E4B-it-abliterated.

Changes applied by an HF Jobs conversion script:

  • Removed multimodal weight tensors matching vision/audio/projector key patterns.
  • Rebuilt tokenizer vocabulary to keep English/Arabic, neutral punctuation/symbols/numbers/marks, byte fallback tokens, and special tokens.
  • Sliced all vocab-sized tensors to the retained token IDs.
  • Updated config/tokenizer vocab sizes and set vision/audio configs to null where present.

See pruning_report.json for counts and sample dropped tokens.

README history 1 version

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

  1. 2026-05-02Create English/Arabic text-only pruned modelae066fa778 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