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

Orion-zhen/Megrez-3B-Instruct-abliterated

Orion-zhen Llama 2.1B
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/Orion-zhen%2FMegrez-3B-Instruct-abliterated"
Response includes
  • classification m1
  • files 11
  • hub_downloads_all_time 131
  • author_summary 13 models
  • readme_text full
Try next
GET/models?author=Orion-zhen
All 13 models by Orion-zhen
 GET/models?family=llama&method=m1
Sibling llama 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
131
28 last 30d - stable
Likes
8
Model age
22mo ago
created 2024-12-18

Training datasets

1 of 1 in /datasets

Corpora the author lists in the model card. Datasets tracked in our /datasets catalog carry a category badge linking to the workflow stage. Others open on Hugging Face.

augmxnt/deccp ablate
Downloads over time
Now145→from14↑936%
05510916414 on Dec 18, 2024145 on Oct 11149 on Sep 24, 2025Dec '24Mar '25Jun '25Sep '25Dec '25MarJunSep
Dec 18, 2024 → Oct 11 · 134 snapshots · spans 662 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.

Infinigence/Megrez-3B-Instruct this lineage
Orion-zhen/Megrez-3B-Instruct-abliterated↓ 28

Metadata

License
gpl-3.0
Languages
en zh
Tags
safetensors llama text-generation conversational en zh dataset:augmxnt/deccp base_model:Infinigence/Megrez-3B-Instruct base_model:finetune:Infinigence/Megrez-3B-Instruct license:gpl-3.0 region:us

Related

Other by Orion-zhen
Qwen2.5-7B-Instruct-Uncensored 4K Qwen2.5-7B-Instruct-Uncensored-Q5_K_M-GGUF 1K Qwen2.5-14B-Instruct-Uncensored-Q5_K_M-GGUF 728 DeepSeek-R1-Distill-Qwen-32B-abliterated-AWQ 142 phi-4-abliterated 102
Total size
5.43 GB
Files
11
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2024-12-28 09:47
Source huggingface.co/Orion-zhen/Megrez-3B-Instruct-abliterated ↗ Base model Infinigence/Megrez-3B-Instruct ↗

Files by quantization

Auxiliary files 11 files 5.44 GB
model-00001-of-00002.safetensors 4.64 GB 05644ce3 download
model-00002-of-00002.safetensors 806 MB 6f53c81b download
tokenizer.json 8.83 MB bbff7ea8 download
i-will-do-anything.png 421 KB 065bb92b download
model.safetensors.index.json 23.4 KB 1d460377 download
tokenizer_config.json 5.28 KB 693c6939 download
.gitattributes 1.48 KB a6344aac download
README.md 897 B abfbea33 download
config.json 803 B dce02aa7 download
special_tokens_map.json 287 B 9b6194a7 download
generation_config.json 148 B a8d76797 download

README current version from Hugging Face

license: gpl-3.0
language:

  • en
  • zh
    base_model:
  • Infinigence/Megrez-3B-Instruct
    pipeline_tag: text-generation
    datasets:
  • augmxnt/deccp

Megrez-3B-Instruct-abliterated

i-will-do-anything

使用 Orion-zhen/abliteration 制作, 启用 deccp, scale-factor 设置为 1.5.

根据简单的测试, 模型已经相对去审查化了, 特殊问题搭配英文食用更佳 ; )

求你了, 给我一个 star 吧, 我什么都会做的...

Made with Orion-zhen/abliteration, deccp enabled, scale-factor = 1.5.

As my simple test shows, the model is raletively uncensored.

Please... give my repo a star if you find it helpful. I will do whatever you want...

README history 2 versions

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

  1. 2024-12-28Update README.mda71a6d1897 B
    Loading...
  2. 2024-12-18Create README.md9955de8898 B
    Loading...

Discussions 3 threads

  1. 2024-12-23PRSync with official tokenizer_config.jsonmerged2 💬#3
    Loading...
  2. 2024-12-23Questionclosed2 💬#2
    Loading...
  3. 2024-12-21Mergrez / Megrezclosed2 💬#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