Qwen3.5-4B-Uncensored-HauhauCS-Aggressive
4BQwenuncensored
from Qwen/Qwen3.5-4B
↓ 268.2K♥ 71624 repos
Qwen3.5-4B-Heretic-2xMerge-DavidAU
Qwen3.5-4B-Heretic-2xMerge-DavidAU is a Heretic-ablated variant of Qwen/Qwen3.5-4B, published on Hugging Face by xtools-at. It is 4B parameters, single consumer GPU class and apache-2.0 licence. It is distributed across 5 repositories in Transformers, GGUF and GGUF (imatrix) formats, totalling 9.9K downloads.
Specification
- Model name
- Qwen3.5-4B-Heretic-2xMerge-DavidAU
- Source repository
- xtools-at/Qwen3.5-4B-Heretic-2xMerge-DavidAU
- Base model
- Qwen/Qwen3.5-4B
- Publisher
- xtools-at
- Parameters
- 4B
- VRAM required
- ~3.2 GB at Q4_K_M, 4k context (estimated) — runs on 4 GB cards
- Hardware class
- 4B – 10B — single consumer GPU
- Licence
- apache-2.0
- Task
- Text generation
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- Transformers, GGUF, GGUF (imatrix)
- First indexed
- 12 Oct 2026
- Last updated
- 14 Oct 2026
Ablation technique
How Qwen3.5-4B-Heretic-2xMerge-DavidAU was modified, and what that implies.
Heretic
Automated directional ablation via the Heretic toolchain, which searches for the refusal direction and applies it with a KL-divergence budget so general capability is preserved.
Reported scope: Not stated by the publisher. Publishers frequently omit this, so verify behaviour empirically rather than assuming full coverage.
From the publisher’s model card
“A weighted merge of two Qwen3.5-4B fine-tunes by DavidAU, combining Claude Opus-style reasoning distillation with the Deckard fine-tuning approach.”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download mradermacher/Qwen3.5-4B-Heretic-2xMerge-DavidAU-i1-GGUF --local-dir ./qwen3.5-4b-heretic-2xmerge-davidau
llama-cli -m ./qwen3.5-4b-heretic-2xmerge-davidau/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "xtools-at/Qwen3.5-4B-Heretic-2xMerge-DavidAU"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve xtools-at/Qwen3.5-4B-Heretic-2xMerge-DavidAU --trust-remote-codeDownloads and variants (4)
Every published repository of Qwen3.5-4B-Heretic-2xMerge-DavidAU, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| mradermacher/Qwen3.5-4B-Heretic-2xMerge-DavidAU-i1-GGUF | GGUF (imatrix) | 7.6K |
| mradermacher/Qwen3.5-4B-Heretic-2xMerge-DavidAU-GGUF | GGUF | 1.7K |
| xtools-at/Qwen3.5-4B-Heretic-2xMerge-DavidAUsource | Transformers | 341 |
| xtools-at/Qwen3.5-4B-Heretic-2xMerge-DavidAU-GGUF | GGUF | 301 |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- Qwen3.5-4B-Heretic-2xMerge-DavidAU will attempt offensive-security prompts that a hosted commercial model declines, which is what makes it usable for red-team corpus generation and for measuring what an unaligned model of this class produces.
- Capability is inherited from Qwen/Qwen3.5-4B, not added by ablation — at 4B parameters, expect a small model’s reasoning and a high error rate on exploit detail.
- The publisher does not state which layers were ablated, so assume nothing about refusal consistency — probe it directly.
- Declared licence is apache-2.0, but the terms that bind you are Qwen/Qwen3.5-4B’s — a re-release cannot grant more than its parent.
Handling: run it isolated, put moderation in front of it if anyone outside your team can reach it, and verify the weights before loading — these are third-party artifacts. Full handling and licence guidance.
Metadata for Qwen3.5-4B-Heretic-2xMerge-DavidAU is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with xtools-at, and does not independently verify publisher claims. See responsible use.