← back to catalog · registered 2026-11-19 01:04

mrfakename/Qwen3.8-27B-abliterated-GGUF

mrfakename 27B GGUF multimodal second-order
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/mrfakename%2FQwen3.8-27B-abliterated-GGUF"
Response includes
  • classification m8
  • files 14
  • author_summary 12 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M8
Primary method

Repackaging (quantization)

Applied on top of direct removal inherited from the base model.
Confidence
MEDIUM
Inherited from base model
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=1
  • assume M1 (base ablation) + M8 (GGUF quant) - default when producer unknown
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
682
Likes
3
Model age
3d ago
created 2026-11-18
Downloads over time
Now 682 → from 0 ↑ 0%
0250500750 0 on Nov 19 682 on Nov 22 Nov
Nov 19 → Nov 22 · 4 snapshots · spans 3 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.

Variants by this author 2 formats · 753 downloads combined

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

Metadata

License
apache-2.0
Quantizations
Q2_K Q3_K Q4_K Q5_K Q6_K Q8_0
Tags
gguf abliterated uncensored image-text-to-text base_model:mrfakename/Qwen3.8-27B-abliterated base_model:quantized:mrfakename/Qwen3.8-27B-abliterated license:apache-2.0 region:us

Related

Total size
108 GB
Files
14
Quantizations
10
Registered
2026-11-19 01:04
Last updated on HF
2026-11-19 00:26

Files by quantization

Q8_0 1 file 27.1 GB
Qwen3.8-27B-abliterated-Q8_0.gguf 27.1 GB 74693a13 download
Q6_K 1 file 23.6 GB
Qwen3.8-27B-abliterated-UD-Q6_K_XL.gguf 23.6 GB dd15e728 download
Q5_K 1 file 19.5 GB
Qwen3.8-27B-abliterated-UD-Q5_K_XL.gguf 19.5 GB e2da87b8 download
Q4_K 1 file 16.4 GB
Qwen3.8-27B-abliterated-UD-Q4_K_XL.gguf 16.4 GB 119a18c3 download
Q3_K 1 file 12.2 GB
Qwen3.8-27B-abliterated-UD-Q3_K_XL.gguf 12.2 GB 054619d4 download
Q2_K 1 file 9.27 GB
Qwen3.8-27B-abliterated-UD-Q2_K_XL.gguf 9.27 GB 7ce7a7f7 download
Drafts and companions 1 file 6.59 MB
abliteration-lora.gguf 6.59 MB 5bfef48b download
BF16 1 file 888 MB
mmproj-BF16.gguf 888 MB 83ee4f4f download
F16 1 file 885 MB
mmproj-F16.gguf 885 MB cbb841a9 download
Auxiliary files 5 files 225 KB
stats.png 120 KB 9020b9b7 download
patch-stats.txt 75.5 KB 6a0c0154 download
abliteration-edit.npz 23.0 KB 861d1aeb download
README.md 4.20 KB 08de5edf download
.gitattributes 2.13 KB 0e1e8c0f download

README current version from Hugging Face


license: apache-2.0
base_model: mrfakename/Qwen3.8-27B-abliterated
pipeline_tag: image-text-to-text
tags:

  • gguf
  • abliterated
  • uncensored

Qwen3.8-27B (abliterated) GGUF

Huge thanks to Hugging Face for generously sponsoring the compute used to abliterate these models! This was done on HF Jobs, which makes it incredibly easy to run tasks such as these :)

Join the Discord: https://discord.com/invite/foss

refusal rates before and after, the search trade-off, and the per-layer edit

GGUF quants of mrfakename/Qwen3.8-27B-abliterated, which is Qwen/Qwen3.8-27B with its refusal behaviour removed by directional ablation.

These are the unsloth/Qwen3.8-27B-GGUF quants with the edit applied in place. Only the edited tensors were dequantized, edited and requantized to their original type. Everything else is byte-identical.

File Size Notes
Qwen3.8-27B-abliterated-Q8_0.gguf 27.1 GiB Closest to the original weights.
Qwen3.8-27B-abliterated-UD-Q6_K_XL.gguf 23.6 GiB
Qwen3.8-27B-abliterated-UD-Q5_K_XL.gguf 19.5 GiB
Qwen3.8-27B-abliterated-UD-Q4_K_XL.gguf 16.4 GiB Good default; fits a 24 GB card.
Qwen3.8-27B-abliterated-UD-Q3_K_XL.gguf 12.2 GiB
Qwen3.8-27B-abliterated-UD-Q2_K_XL.gguf 9.3 GiB Smallest. Loses a little of the edit (see below).

What changed

  • Refusal directions were measured separately for chat and think prompts at every layer, on the Q8_0 GGUF. The subspace candidates are the top singular vectors of all of those directions together.
  • A multi-objective search (minimise refusals, keep the first-token KL on harmless prompts low) chose the subspace rank and per-layer strengths. It settled on a single direction (rank 1), at strength 0.8–1.06 across layers 13–63. The first 13 layers are left alone: editing them hurt quality a lot and did little for refusals.
  • That direction was projected out of the residual-stream writes: attn_output on the 13 full-attention layers, ssm_out on the 38 linear-attention layers, and ffn_down on all 51 of those layers. That is 102 tensors.

Nothing else was touched: layers 0–12, embeddings and the vision projector are byte-identical to the original quants.

Results (Q8_0 GGUF, greedy)

refusals, chat (/100) think mode, reasoning or answer refuses (/100) KL, harmless first token
original 99 97 0
abliterated 0 1 0.110

The 100 prompts are held-out harmful requests. Chat replies were 384 tokens; think mode got 768 tokens. Where the reasoning had not finished by then, the reasoning text itself was judged. The KL figure includes a small noise floor from batched inference.

Does the edit survive quantization?

Mostly. Low-bit quants can round a small edit away, and a dense 27B has no shared expert held at high precision, so unsloth quantizes the edited ffn_down / ssm_out / attn_output tensors along with everything else. The patch log measures how much of the removed direction is left in each tensor after requantizing:

File edit lost, average edit lost, worst tensor
Q8_0 0.1% 0.1%
UD-Q6_K_XL 0.3% 1.9%
UD-Q5_K_XL 0.9% 2.6%
UD-Q4_K_XL 1.5% 6.9%
UD-Q3_K_XL 2.7% 12%
UD-Q2_K_XL 6.5% 33% (three IQ1_S ffn_down tensors)

Q4 and up keep essentially all of it. UD-Q2_K_XL keeps most of it, but it was not separately evaluated and may refuse slightly more often. If that matters, use Q3 or higher. See patch-stats.txt for every tensor.

Running

Mainline llama.cpp runs these, e.g. llama-server -m Qwen3.8-27B-abliterated-UD-Q4_K_XL.gguf --mmproj mmproj-F16.gguf.

Files

  • Qwen3.8-27B-abliterated-*.gguf: the patched quants.
  • mmproj-*.gguf: the unchanged vision projector.
  • abliteration-lora.gguf: the edit as a LoRA, if you'd rather use another quant of the original: --lora abliteration-lora.gguf.
  • abliteration-edit.npz: the subspace basis and per-layer strengths.
  • patch-stats.txt: per-tensor patch log.

This model will comply with harmful requests. You are responsible for how you use it.

README history 2 versions

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

  1. 2026-11-19 card: quantization note e376a73 4.2 KB
    Loading...
  2. 2026-11-19 adapter, edit, card and patch stats bbb5206 3.4 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