← back to catalog · registered 2026-10-02 06:58

MahdiAlikhah/qwen2-vl-2b-abliterated

MahdiAlikhah Qwen 2B 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/MahdiAlikhah%2Fqwen2-vl-2b-abliterated"
Response includes
  • classification m1
  • files 9
  • author_summary 1 models
  • readme_text full
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)
Refusal direction extracted via
Extraction technique

Difference-of-means

Confidence
MEDIUM
Why we say so
primary_method=M1; difference-of-means is the reference extraction for M1/M3 (Arditi 2024)
Downloads · 30-day
35
Likes
0
Descendants
2
in 2 direct forks
Model age
7w ago
created 2026-10-02
Downloads over time
Now 99 → from 0 ↑ 0%
03673109 0 on Oct 2 99 on Nov 22 OctNov
Oct 2 → Nov 22 · 52 snapshots · spans 51 days

Genealogy 2 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.

Metadata

License
apache-2.0
Languages
en
Tags
safetensors qwen2_vl abliterated uncensored heretic qwen2-vl vision-language image-text-to-text conversational en base_model:Qwen/Qwen2-VL-2B-Instruct base_model:finetune:Qwen/Qwen2-VL-2B-Instruct

Related

Total size
4.11 GB
Files
9
Quantizations
1
Registered
2026-10-02 06:58
Last updated on HF
2026-10-22 21:33

Files by quantization

Auxiliary files 9 files 4.13 GB
model.safetensors 4.11 GB e0539c54 download
tokenizer.json 10.9 MB ff8cce54 download
README.md 3.30 KB 8cfee4ea download
config.json 2.36 KB 65906d23 download
.gitattributes 1.53 KB 52373fe2 download
processor_config.json 1.32 KB c02237db download
chat_template.jinja 1023 B 50fa77d5 download
tokenizer_config.json 792 B 9619e1db download
generation_config.json 258 B 02461d9e download

README current version from Hugging Face


license: apache-2.0
base_model: Qwen/Qwen2-VL-2B-Instruct
tags:

  • abliterated
  • uncensored
  • heretic
  • qwen2-vl
  • vision-language
    language:
  • en
    pipeline_tag: image-text-to-text

qwen2-vl-2b-abliterated

This is an abliterated version of Qwen/Qwen2-VL-2B-Instruct, with its built-in refusal behavior removed using Heretic.

What is abliteration

Abliteration is a technique that identifies the internal direction in a model's residual stream associated with refusing requests, then surgically removes that direction from the relevant weight matrices. Unlike fine-tuning on a dataset of "jailbreak" examples, this directly edits the weights responsible for refusal behavior, leaving the rest of the model's capabilities intact.

Heretic automates this process using an Optuna-based search to find the ablation parameters that minimize refusal rate while minimizing KL divergence from the original model's output distribution — i.e., removing refusals while disturbing general capability as little as possible.

Results

The selected configuration for this model:

Metric Value
Refusals 4 / 100
KL divergence from base model 0.0338

Out of 100 prompts that the original Qwen2-VL-2B-Instruct refuses, this model refuses only 4, while staying very close to the original model's output distribution on benign prompts.

Usage

from transformers import AutoProcessor, AutoModelForImageTextToText
import torch

model_id = "MahdiAlikhah/qwen2-vl-2b-abliterated"

processor = AutoProcessor.from_pretrained(model_id)
model = AutoModelForImageTextToText.from_pretrained(
    model_id,
    dtype=torch.bfloat16,
    device_map="auto",
)

messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "image": "path/to/image.jpg"},
            {"type": "text", "text": "Describe this image."},
        ],
    }
]

inputs = processor.apply_chat_template(
    messages, tokenize=True, return_dict=True,
    return_tensors="pt", add_generation_prompt=True
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=512)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:], skip_special_tokens=True))

Capabilities

This model retains the base model's architecture and capabilities: image understanding, visual question answering, and OCR, alongside standard text chat. It does not generate or edit images — it is understanding-only, same as the base model.

Limitations

  • This model has had its refusal behavior deliberately suppressed. It will comply with requests the base model would refuse, which may include harmful, dangerous, or inappropriate content. Use responsibly and at your own discretion.
  • Abliteration targets the text-based refusal direction; it was not evaluated for degradation on the vision/OCR pathway specifically beyond the general KL divergence figure above.
  • As with the base model, this is a 2B-parameter model and has the corresponding limitations in reasoning depth, factual accuracy, and instruction-following reliability compared to larger models.

Credits

README history 3 versions

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

  1. 2026-10-22 Update README.md b3e03f1 4.1 KB
    Loading...
  2. 2026-10-22 Update README.md 1bad827 4 KB
    Loading...
  3. 2026-10-02 Create README.md ac6b048 3.3 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