← back to catalog · registered 2026-09-26 09:57

xdchinza/Qwen3.5-0.8B-abliterated

xdchinza Qwen 800M multimodal
curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/xdchinza%2FQwen3.5-0.8B-abliterated"
Response includes
  • classification m-uncensored
  • files 12
  • author_summary 2 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M-U
Primary method

Uncensored (method unknown)

No other method signals detected in this model.
Confidence
LOW
Why this label 3 signals
Weak or ambiguous signals. Best guess based on catalog patterns; treat as tentative and check the evidence below.
  • 'uncensored' in name/tags but no 'abliterated' marker
  • method not identifiable from author declaration alone
  • may be DPO fine-tune, prompt engineering, or unknown technique
Refusal direction extraction

No specific extraction method could be identified for this model. The producer either did not document it or used a proprietary pipeline.

What is a refusal direction? →
Downloads · 30-day
0
Likes
0
Descendants
1
in 1 direct fork
Model age
today
created 2026-09-26

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 · 0 downloads combined

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

Metadata

License
apache-2.0
Tags
transformers safetensors qwen3_5 image-text-to-text abliterated uncensored directional-ablation mechanistic-interpretability conversational base_model:Qwen/Qwen3.5-0.8B base_model:finetune:Qwen/Qwen3.5-0.8B license:apache-2.0

Related

Total size
1.63 GB
Files
12
Quantizations
1
Registered
2026-09-26 09:57
Last updated on HF
2026-09-26 09:14

Files by quantization

Auxiliary files 12 files 1.65 GB
model.safetensors-00001-of-00001.safetensors 1.63 GB bab2a8df download
tokenizer.json 12.2 MB 5f9e4d49 download
vocab.json 6.41 MB 0aa0ce06 download
merges.txt 3.20 MB a494e019 download
model.safetensors.index.json 49.7 KB f691cefd download
tokenizer_config.json 16.3 KB fae3ce99 download
chat_template.jinja 7.57 KB 0ef09f21 download
README.md 3.23 KB d1fd4eb0 download
config.json 2.84 KB 715f0448 download
.gitattributes 1.53 KB 52373fe2 download
preprocessor_config.json 390 B 2ea84a43 download
video_preprocessor_config.json 385 B 3ba673a5 download

README current version from Hugging Face


license: apache-2.0
base_model: Qwen/Qwen3.5-0.8B
tags:

  • abliterated
  • uncensored
  • directional-ablation
  • mechanistic-interpretability
  • qwen3_5
  • image-text-to-text
  • conversational
    pipeline_tag: image-text-to-text
    library_name: transformers

Qwen3.5-0.8B — Refusal-Direction Abliterated

Directional ablation of the refusal direction in
Qwen/Qwen3.5-0.8B. No fine-tuning,
no gradient steps — only a rank-1 orthogonalisation of 24 weight matrices.

Method

  1. A refusal direction u is estimated as the difference in mean residual-stream
    activations between 30 harmful and 50 harmless instruction prompts, taken at the
    last token of the chat-formatted prompt.

  2. All 24 layers are swept and scored by the separation gap
    cos(harmful, u) − cos(harmless, u). The best layer is 23
    (gap 0.629; layers 0–2 give ~0.03–0.12, so the signal is strongly late).

  3. u is projected out of the row space of every weight matrix in layers
    12–23 whose output space is the residual stream:

    W' = W − u (uᵀ W)
    

    Targets: mlp.down_proj (all 12 layers), plus self_attn.o_proj on the
    full-attention layers and linear_attn.out_proj on the linear-attention
    layers. Residual |Wu| after patching: 0.0000.

The vision tower (model.visual.*) and the MTP head (mtp.*) are untouched.

Results

Refusal rate on 20 harmful instruction prompts, greedy decoding, non-thinking mode:

refusal rate
Base Qwen/Qwen3.5-0.8B 18/20 (90%)
This model 1/20 (5%)

Quality checks against the base model:

Metric Base This model
NLL, held-out neutral prose 2.0713 2.0625
Repeated 4-gram rate 0.009 0.009
Factual spot-checks (4) 4/4 4/4
Arithmetic (84*3/2) correct correct

Known limitations

Abliteration removes the refusal behaviour; it does not add capability. On
this 0.8B model the compliant answers to harmful prompts are frequently
incoherent or factually wrong, because the base model is too small to produce
technically correct instructions. Ablation also measurably degrades factual
precision in ordinary answers — the base model is more reliable on questions
about specific real-world details. This model is best used where over-refusal
is the bigger problem than factual accuracy.

Usage

from transformers import AutoProcessor, Qwen3_5ForConditionalGeneration

model = Qwen3_5ForConditionalGeneration.from_pretrained(
    "xdchinza/Qwen3.5-0.8B-abliterated", dtype="auto", device_map="auto"
)
processor = AutoProcessor.from_pretrained("xdchinza/Qwen3.5-0.8B-abliterated")

GGUF quantisation for llama.cpp / Ollama, including an i-matrix quant:
xdchinza/Qwen3.5-0.8B-abliterated-GGUF

ollama create qwen3.5-0.8b-abliterated -hf xdchinza/Qwen3.5-0.8B-abliterated-GGUF:Q4_K_M

Reproduction

Scripts used are documented in the repository README of the uploader. Summary
of parameters: 30 harmful / 50 harmless prompts, probe layer 23, strength 1.0,
layers 12–23.

License

Apache-2.0, inherited from Qwen/Qwen3.5-0.8B. Derivative of an Apache-2.0
model, redistributed under the same terms.

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 app" button that hands off directly to a local runtime of your choice - Abliteration, LM Studio, or Ollama. No API keys, no subscription, no prompt leakage.