Qwen3-1.7B-heretic

A decensored variant of Qwen/Qwen3-1.7B, produced with Heretic v1.4.0 (directional ablation / "abliteration", per-layer refusal directions). 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 Qwen3 hybrid thinking capability are left largely intact.

Who this is for: developers who want Qwen3's 1.7B dense base model without the refusal guardrails — for uncensored experimentation, fine-tune starts, and studying refusal mechanics in a thinking-capable architecture. Small enough to run anywhere via GGUF, including CPU-only machines.

Runs on your gaming PC

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

Weights only, at this model's 1.7B native size; add ~1 GB for context. OOM? Drop one quant level. Headroom to spare? Go one up.

Abliteration parameters

Performance

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

Files

GGUF quantizations

Full quantization set (14 quants + F16) produced with llama.cpp.

Loads natively in llama.cpp / Ollama / LM Studio / Jan.

Run llama serve -hf saidutta69/Qwen3-1.7B-heretic to pull the default quant.

Quickstart

# llama.cpp
llama serve -hf saidutta69/Qwen3-1.7B-heretic

# transformers
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "saidutta69/Qwen3-1.7B-heretic"
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype="auto", device_map="auto")
tokenizer = AutoTokenizer.from_pretrained(model_name)

messages = [{"role": "user", "content": "Who are you?"}]
inputs = tokenizer.apply_chat_template(messages, add_generation_prompt=True, tokenize=True,
                                        return_dict=True, return_tensors="pt").to(model.device)
out = model.generate(**inputs, max_new_tokens=200)
print(tokenizer.decode(out[0][inputs["input_ids"].shape[-1]:], skip_special_tokens=True))

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

FAQ

What is Qwen3-1.7B-heretic?

Qwen3-1.7B-heretic is a decensored build of Qwen/Qwen3-1.7B 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 Qwen3-1.7B-heretic actually uncensored?

Yes. On Heretic's 100-prompt refusal harness it drops from 92/100 to 4/100 at KL divergence 0.0248 from the base model. These are measured values, not a claim; reproduction data is in reproduce/.

How do I run Qwen3-1.7B-heretic locally?

llama serve -hf saidutta69/Qwen3-1.7B-heretic:Q4_K_M pulls the default quant; the repo ships 15 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 — don't put this behind an unmoderated public-facing endpoint serving third parties. It inherits Qwen3-1.7B's factual limitations and biases; abliteration removes refusal directions, it doesn't add capability or judgment.

License

Inherits the apache-2.0 license from the base model.

Related

• Qwen3-0.6B-heretic
• RACER IS OP — Heretic Models — full collection
• Qwen3-8B-heretic
• Qwen2.5-1.5B-Instruct-heretic

---

Base model: Qwen3-1.7B

Original Qwen3-1.7B model card (click to expand)

See the base model card at Qwen/Qwen3-1.7B for the original architecture, training details, requirements, and citation.

Abliteration hardware

Recorded in reproduce/reproduce.json:

• GPU: Tesla T4 (14.56 GB VRAM), Tesla T4 (14.56 GB VRAM)
• CPU: Intel(R) Xeon(R) CPU @ 2.00GHz
• OS: Linux-6.12.90+-x86_64-with-glibc2.35
• CUDA / driver: CUDA 13.0, driver 580.159.04
• Software: PyTorch 2.14.0+cu130, Heretic 1.4.0

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