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

WeilJimmer/Qwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking-GGUF

WeilJimmer Qwen 4B GGUF multimodal 262K ctx
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/WeilJimmer%2FQwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking-GGUF"
Response includes
  • classification m3
  • files 11
  • hub_downloads_all_time 4,132
  • author_summary 1 models
  • readme_text full
Try next
GET/models?family=qwen&method=m3
Sibling qwen models · same method
 GET/methods/m3
What "m3" means · taxonomy
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M3
Primary method

Layer-wise ablation

Applied on top of direct removal inherited from the base model.
Confidence
HIGH
Inherited from base model
M1 Direct removal →
Why this label 2 signals
Producer identity confirmed by naming conventions, tags or the model card. This label is very unlikely to change.
  • 'heretic' in model name (Heretic-produced)
  • Heretic uses layer-wise optimization (M3) with underlying direction removal (M1)
How we label → About M3 →
Refusal direction extracted via
Extraction technique

Difference-of-means

FromArditi et al. (2024) ↗
Confidence
HIGH
Why we say so
name contains 'heretic'; Heretic default extraction is difference-of-means (Arditi 2024)
What is a refusal direction? → Read the paper ↗
Downloads · lifetime
4K
427 last 30d - stable
Likes
0
Model age
5mo ago
created 2026-05-14
Downloads over time
Now4.3K→from860↑405%
6862K3.4K4.7K860 on May 134.3K on Oct 11MayJunJulAugSepOct
May 13 → Oct 11 · 61 snapshots · spans 151 days

Metadata

Quantizations
BF16 Q2_K Q3_K Q4_K Q5_K Q6_K Q8_0
Tags
gguf qwen3_5 llama.cpp unsloth vision-language-model endpoints_compatible region:us conversational
Total size
24.7 GB
Files
11
Quantizations
9
Registered
2026-08-22 13:56
Last updated on HF
2026-05-15 00:02
Source huggingface.co/WeilJimmer/Qwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking-GGUF ↗

Files by quantization

BF16 1 file 7.85 GB
Qwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking.BF16.gguf 7.85 GB e443c48c download
Q8_0 1 file 4.17 GB
Qwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking.Q8_0.gguf 4.17 GB aee0f63c download
Q6_K 1 file 3.23 GB
Qwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking.Q6_K.gguf 3.23 GB af110f4b download
Q5_K 1 file 2.86 GB
Qwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking.Q5_K_M.gguf 2.86 GB bb4bb2dc download
Q4_K 1 file 2.52 GB
Qwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking.Q4_K_M.gguf 2.52 GB 638ec2b1 download
Q3_K 1 file 2.11 GB
Qwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking.Q3_K_M.gguf 2.11 GB 9bcff71a download
Q2_K 1 file 1.93 GB
Qwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking.Q2_K_L.gguf 1.93 GB 83e5210b download
mmproj 1 file 644 MB
Qwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking.BF16-mmproj.gguf 644 MB f7d87f5d download
Auxiliary files 3 files 6.06 KB
config.json 2.88 KB 069c7afa download
.gitattributes 2.22 KB 288ef72d download
README.md 974 B a034b7e3 download

README current version from Hugging Face

tags:

  • gguf
  • llama.cpp
  • unsloth
  • vision-language-model

based on https://huggingface.co/DavidAU/Qwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking

Qwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking-GGUF : GGUF

This model was finetuned and converted to GGUF format using Unsloth.

Example usage:

  • For text only LLMs: llama-cli -hf WeilJimmer/Qwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking-GGUF --jinja
  • For multimodal models: llama-mtmd-cli -hf WeilJimmer/Qwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking-GGUF --jinja

Available Model files:

  • Qwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking.BF16.gguf
  • Qwen3.5-4B-Deckard-HERETIC-UNCENSORED-Thinking.BF16-mmproj.gguf
    This was trained 2x faster with Unsloth

README history 11 versions

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

  1. 2026-05-15Update README.mdee71514974 B
    Loading...
  2. 2026-05-14Add READMEa329e32886 B
    Loading...
  3. 2026-05-14Add READMEd9b8e1b886 B
    Loading...
  4. 2026-05-14Add README5392a6b886 B
    Loading...
  5. 2026-05-14Add READMEd65ce08888 B
    Loading...
  6. 2026-05-14Add README542191f888 B
    Loading...
  7. 2026-05-14Add README94758a1888 B
    Loading...
  8. 2026-05-14Update README.mdecfba01975 B
    Loading...
  9. 2026-05-14Add READMEe970cb5888 B
    Loading...
  10. 2026-05-14Update README.mda85a796115 B
    Loading...
  11. 2026-05-14initial commit684ddaf28 B
    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