← back to catalog · registered 2026-09-24 12:57

OS-Software/CyberStag-Security-26B-A4B-V1-Uncensored-Heretic

OS-Software Gemma 25B MoE
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/OS-Software%2FCyberStag-Security-26B-A4B-V1-Uncensored-Heretic"
Response includes
  • classification m3
  • files 19
  • hub_downloads_all_time 74
  • author_summary 34 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M3
Primary method

Layer-wise ablation

Applied on top of direct removal inherited from the base model.
Confidence
HIGH
Inherited from base model
Why this label 2 signals
Producer identity confirmed by naming conventions, tags or the model card. This label is very unlikely to change.
  • 'heretic' in model name (Heretic-produced)
  • Heretic uses layer-wise optimization (M3) with underlying direction removal (M1)
Refusal direction extracted via
Extraction technique

Difference-of-means

Confidence
HIGH
Why we say so
name contains 'heretic'; Heretic default extraction is difference-of-means (Arditi 2024)
Downloads · lifetime
74
Likes
0
Descendants
1
in 1 direct fork
Model age
6w ago
created 2026-09-24
Downloads over time
Now175→from0↑0%
0641281930 on Sep 24175 on Nov 9175 on Nov 8SepOctNov
Sep 24 → Nov 9 · 47 snapshots · spans 46 days

Genealogy 1 direct fork

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

Variants by this author 2 formats · 4K downloads combined

The same weights this author released in different packaging. Pick the format that matches your runtime.

Metadata

License
apache-2.0
Languages
multilingual en ja
Tags
transformers safetensors gemma4_text text-generation heretic uncensored decensored abliterated cybersecurity security gemma reasoning

Related

Total size
47.0 GB
Files
19
Quantizations
1
Registered
2026-09-24 12:57
Last updated on HF
2026-09-24 12:46

Files by quantization

Auxiliary files 19 files 47.0 GB
model-00005-of-00011.safetensors 4.58 GB 1a2dc8e7 download
model-00007-of-00011.safetensors 4.58 GB ecb7aa00 download
model-00009-of-00011.safetensors 4.58 GB b31f0267 download
model-00003-of-00011.safetensors 4.58 GB 6ffeea1d download
model-00006-of-00011.safetensors 4.55 GB 0ab9f36e download
model-00008-of-00011.safetensors 4.55 GB b9aa54c4 download
model-00010-of-00011.safetensors 4.55 GB 29ed561a download
model-00004-of-00011.safetensors 4.55 GB 196bddbc download
model-00002-of-00011.safetensors 4.55 GB a9dadfcb download
model-00001-of-00011.safetensors 4.41 GB ab2ad7f5 download
model-00011-of-00011.safetensors 1.54 GB 10ce0fec download
tokenizer.json 30.7 MB cc8d3a0c download
model.safetensors.index.json 54.2 KB 83e9e117 download
chat_template.jinja 12.3 KB 5bc9bbf2 download
tokenizer_config.json 2.80 KB afb19b90 download
README.md 2.67 KB 2f2fab4a download
config.json 2.48 KB 71a7349c download
.gitattributes 1.53 KB 52373fe2 download
generation_config.json 218 B 2ae0dff2 download

README current version from Hugging Face


language:

  • multilingual
  • en
  • ja
    tags:
  • heretic
  • uncensored
  • decensored
  • abliterated
  • cybersecurity
  • security
  • gemma
  • reasoning
  • text-generation
  • fine-tuned
  • BF16
    base_model: google/gemma-4-26b-a4b
    license: apache-2.0
    pipeline_tag: text-generation
    library_name: transformers

This is a decensored version of deer-sec/CyberStag-Security-26B-A4B-V1-BF16-MLX, made using Heretic v2.0.0.dev0+custom

Abliteration parameters

Parameter Value
start_layer_index 12
end_layer_index 20
preserve_good_behavior_weight 1.0
steer_bad_behavior_weight 0.0001
overcorrect_relative_weight 1.0
neighbor_count 1
ridge_regularization 7e-07
transport_rank 4
entropy_regularization 0.1
transport gaussian
lora_rank 128
row_normalization none
target_components attn.o_proj, mlp.down_proj
covariance_regularization 0.01
max_weight_change 1.0

Performance

Metric This model Original model (a model)
Refusals 0/100 98/100
KL divergence 0.0135 0 (by definition)

⚠️ Important Notice

This model has undergone substantial reduction of its safety alignment. As a result, it is more likely than standard models to generate harmful, inaccurate, biased, offensive, or otherwise inappropriate content.

Intended Use

For research and experimentation only, including safety research, alignment studies, and red-teaming. Please avoid deploying it in public or end-user-facing services.

User Responsibility

All outputs should be treated as untrusted and independently verified before use. Users are solely responsible for:

  • Evaluating the accuracy and suitability of generated content
  • Implementing appropriate safeguards and human oversight
  • Complying with applicable laws, regulations, licenses, and ethical standards

Use of this model is entirely at your own risk.

Disclaimer

OS-Software provides this model without warranties of any kind and assumes no liability for any direct or indirect damages, losses, misuse, or legal consequences arising from its use.

Acknowledgements

Thanks to the base model developers, p-e-w for Heretic, and the wider open-source community.

This is a derivative work released under the base model’s applicable license. All rights to the base model remain with their respective owners.


README history 4 versions

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

  1. 2026-09-24Update README.md7f8b6882.6 KB
    Loading...
  2. 2026-09-24Update README.mdbc4690a2.6 KB
    Loading...
  3. 2026-09-24Upload README.md with huggingface_hub15796b22.3 KB
    Loading...
  4. 2026-09-24Upload Gemma4ForCausalLMd0a4e8d5.1 KB
    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.

Open in Abliteration