Qwen3.5-9B-heretic

Qwen3.5-9B-heretic

<div align="center"> <img src="https://photu.kashyalabanavli.site/racer-is-op.png" alt="RACER IS OP" width="100%"> </div> <br>

A decensored variant of Qwen/Qwen3.5-9B, produced with Heretic v1.4.0 (directional ablation / "abliteration"). Refusal behavior is suppressed via targeted weight edits to the attention output and MLP down-projections rather than fine-tuning, so the base model's knowledge and instruction-following are left largely intact.

Who this is for: developers who want Qwen's latest 9B model without refusals - strong multilingual reasoning and agentic capabilities. Best run on a 12 GB GPU or via the Q4_K_M/Q5_K_M GGUF on consumer hardware. Not a capability upgrade over base Qwen3.5-9B - same model, refusal guardrails removed.

<!-- racer-gpu-matrix -->

Runs on your gaming PC

Full GGUF ladder included — pick the quant that fits your card:

Weights only, at this model's ~9.4B native size; add ~1 GB for context. OOM? Drop one quant level. Headroom to spare? Go one up.

Why abliteration instead of fine-tuning

Fine-tuning a "helpful" persona on top of RLHF'd refusals fights the base model's training and tends to degrade coherence. Abliteration instead finds and edits the specific weight directions responsible for refusal, leaving the rest of the network (and its capabilities) untouched. See the Heretic repo and the original abliteration writeup for the mechanism.

Performance

Evaluated with heretic --model Qwen/Qwen3.5-9B --evaluate-model saidutta69/Qwen3.5-9B-heretic (Heretic v1.4.0, 2x Tesla T4, 100 harmful prompts from mlabonne/harmful_behaviors, 100 harmless prompts from mlabonne/harmless_alpaca). KL divergence of 0.0015 on the output distribution is essentially zero - the edit is extremely narrow and utility is intact.

Files

GGUF quantizations

Full quantization set (4 quants + F16) produced with llama.cpp.

Qwen3.5 hybrid linear-attention architecture — loads natively in llama.cpp / Ollama / LM Studio / Jan.

Run llama serve -hf saidutta69/Qwen3.5-9B-heretic to pull the default quant.

Quickstart

# llama.cpp - defaults to the Q4_K_M quant
llama serve -hf saidutta69/Qwen3.5-9B-heretic:Q4_K_M

# transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
model_name = "saidutta69/Qwen3.5-9B-heretic"
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype="auto", device_map="auto")
tokenizer = AutoTokenizer.from_pretrained(model_name)
# ... inference code

Also runnable via Ollama, LM Studio, Jan, vLLM, SGLang.

FAQ

What is Qwen3.5-9B-heretic?

Qwen3.5-9B-heretic is a decensored build of Qwen/Qwen3.5-9B produced with Heretic 1.4.0 (directional ablation). It is the same weights with the refusal direction removed, not a new fine-tune, so its capabilities track the base model.

Is Qwen3.5-9B-heretic actually uncensored?

Yes. On Heretic's 100-prompt refusal harness it drops from 84/100 to 79/100 at KL divergence 0.0015 from the base model. These are measured values, not a claim; reproduction data is in reproduce/. This edit is conservative: only 5 of 84 refusals cleared, so read KL and refusal count together.

How do I run Qwen3.5-9B-heretic locally?

llama serve -hf saidutta69/Qwen3.5-9B-heretic:Q4_K_M pulls the default quant; the repo ships 5 GGUF quants (see Files).

Responsible use

Refusal suppression is deliberate and works as intended: this model will comply with requests the base model would refuse, including some it shouldn't. There is no safety filtering layered on top. You are responsible for how you deploy it.

Made with ❤️ by RACER IS OP (saidutta69) — follow for more uncensored models

License

Inherits the qwen-research-license license from the base model.

Related

• Qwen/Qwen3.5-9B - the base model this checkpoint derives from
• RACER IS OP - Heretic Models - full collection - every heretic model in this collection
• Qwen3.5-4B-heretic - nearest Qwen3.5 sibling
• Qwen3.5-2B-heretic - nearest Qwen3.5 sibling

Abliteration hardware

Recorded in reproduce/reproduce.json:

• GPU: NVIDIA RTX PRO 6000 Blackwell Server Edition (94.97 GB VRAM)
• CPU: Intel(R) Xeon(R) Platinum 8559C
• OS: Linux-6.8.0-1060-aws-x86_64-with-glibc2.39
• CUDA / driver: CUDA 12.8, driver 580.126.20
• Software: PyTorch 2.8.0+cu128, Heretic 1.4.0

Heretic notes that KL-divergence and refusal numbers can vary slightly with hardware and library versions.