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

usermma/Atomight-V2.2-UltraThink-0.5B-abliterated

usermma Qwen 494M
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/usermma%2FAtomight-V2.2-UltraThink-0.5B-abliterated"
Response includes
  • classification m1
  • files 11
  • hub_downloads_all_time 54
  • author_summary 77 models
  • readme_text full
Try next
GET/models?author=usermma
All 77 models by usermma
 GET/models?family=qwen&method=m1
Sibling qwen 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
54
9 last 30d - stable
Likes
0
Descendants
9
in 9 direct forks
Model age
3mo ago
created 2026-06-28
Downloads over time
Now59→from34↑74%
3342526234 on Jul 159 on Oct 1159 on Oct 9JulAugSepOct
Jul 1 → Oct 11 · 54 snapshots · spans 102 days

Genealogy 9 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.

NovatasticRoScript/Atomight-V2.2-UltraThink-0.5B this lineage
usermma/Atomight-V2.2-UltraThink-0.5B-abliterated↓ 9 9 forks
mradermacher/Atomight-V2.2-UltraThink-0.5B-abliterated-i1-GGUF↓ 644
mradermacher/Atomight-V2.2-UltraThink-0.5B-abliterated-GGUF↓ 378
usermma/Atomight-V2.2-UltraThink-0.5B-abliterated-mlx-3Bit↓ 12
usermma/Atomight-V2.2-UltraThink-0.5B-abliterated-mlx-6Bit↓ 12
usermma/Atomight-V2.2-UltraThink-0.5B-abliterated-mlx-fp16↓ 11
usermma/Atomight-V2.2-UltraThink-0.5B-abliterated-mlx-2Bit↓ 11
usermma/Atomight-V2.2-UltraThink-0.5B-abliterated-mlx-5Bit↓ 11
usermma/Atomight-V2.2-UltraThink-0.5B-abliterated-mlx-4Bit↓ 6
usermma/Atomight-V2.2-UltraThink-0.5B-abliterated-mlx-8Bit↓ 6

Variants by this author 2 formats · 20 downloads combined

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

Metadata

License
apache-2.0
Tags
safetensors qwen2 obliteratus abliteration uncensored obliterate en base_model:NovatasticRoScript/Atomight-V2.2-UltraThink-0.5B base_model:finetune:NovatasticRoScript/Atomight-V2.2-UltraThink-0.5B license:apache-2.0 region:us

Related

Other by usermma
Huihui-MiniCPM5-1B-abliterated-mlx-2Bit 279 Huihui-LFM2.5-8B-A1B-abliterated-mlx-2Bit 166 Huihui-LFM2.5-8B-A1B-abliterated-mlx-fp16 164 Qwythos-9B-v2-Heretic-Abliterated 133 DeepHat-V1-7B-Heretic-Abliterated-mlx-8Bit 126
Total size
942 MB
Files
11
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2026-06-28 23:05
Source huggingface.co/usermma/Atomight-V2.2-UltraThink-0.5B-abliterated ↗ Base model NovatasticRoScript/Atomight-V2.2-UltraThink-0.5B ↗

Files by quantization

Auxiliary files 11 files 953 MB
model.safetensors 942 MB 4c8865c0 download
tokenizer.json 10.9 MB dbfda6cb download
chat_template.jinja 2.45 KB bdf7919a download
abliteration_metadata.json 1.77 KB a147ff2d download
.gitattributes 1.53 KB 52373fe2 download
config.json 1.31 KB 49e8583c download
README.md 1.22 KB c467ccd6 download
tokenizer_config.json 566 B 1130097c download
obliteratus_session.json 327 B 4b0eefb1 download
generation_config.json 239 B fe6dcbd7 download
.quick_checkpoint 18.0 B 27769b5b download

README current version from Hugging Face

language: en
tags:

  • obliteratus
  • abliteration
  • uncensored
  • obliterate
    base_model: NovatasticRoScript/Atomight-V2.2-UltraThink-0.5B
    license: apache-2.0

Atomight-V2.2-UltraThink-0.5B-abliterated

This model was abliterated using the spectral_cascade + informed method via
OBLITERATUS.

Detail Value
Base model usermma/Atomight-V2.2-UltraThink-0.5B-OBLITERATED
Method spectral_cascade + informed
Source obliterate

How to Use

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("Atomight-V2.2-UltraThink-0.5B-abliterated")
tokenizer = AutoTokenizer.from_pretrained("Atomight-V2.2-UltraThink-0.5B-abliterated")

prompt = "Hello, how are you?"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

About OBLITERATUS

OBLITERATUS is an open-source tool for removing refusal behavior from language
models via activation engineering (abliteration). Learn more at
github.com/elder-plinius/OBLITERATUS.

README history 3 versions

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

  1. 2026-06-28Update README.md48d60fe1.2 KB
    Loading...
  2. 2026-06-28Update README.md3c1ec4f1.2 KB
    Loading...
  3. 2026-06-28OBLITERATUS: informed on usermma/Atomight-V2.2-UltraThink-0.5B-OBLITERATED19fe9041.2 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.

Get the app Browse the catalog
Open in Abliteration