license: other
license_name: tongyi-qianwen
base_model: Qwen/Qwen3.6-35B-A3B
tags:

• abliterated
• uncensored
• qwen3
• moe
• abliterix

Qwen3.6-35B-A3B — Abliterated

Overview

Qwen3.6-35B-A3B-abliterated-exl3-6bpw is an EXL3-quantized checkpoint for ExLlamaV3-compatible runtimes, published by groxaxo.
It is intended for open-source evaluation, reproducible experimentation, and compatible local or
hosted inference workflows. The wording below is deliberately limited to what can be verified
from this repository's metadata and artifacts.

The repository name identifies a behavior-modified or reduced-filtering lineage. That label describes the source or conversion history; it is not a guarantee of unrestricted behavior in every prompt or runtime. Test outputs carefully before sharing or deploying them.

At a glance

What is included

• *.safetensors (4 files)
• config.json
• generation_config.json
• tokenizer.json
• tokenizer_config.json
• chat_template.jinja
• quantization_config.json
• Additional configuration, tokenizer, processor, or shard files (11 visible artifacts total)

Quick start

EXL3-compatible runtimes

Download the EXL3 files and load the desired bitrate with a current ExLlamaV3-compatible
runtime. The correct loader and context settings depend on the model architecture and should be
verified against the runtime's documentation.

Compatibility and responsible use

• Use a runtime that explicitly supports this format, architecture, and modality.
• Keep configuration, tokenizer, processor, projection, and weight files from the same revision together.
• Review the source model card and license before redistribution or deployment.
• Hardware needs depend on parameter count, context length, cache precision, quantization, and concurrency.
• Report reproducible issues with the runtime version, hardware, launch command, and a minimal example.

Quantization or conversion changes numerical behavior, memory use, and throughput relative to the source checkpoint; validate quality on your own workload.

Generated outputs may be inaccurate or unsuitable for a given use case. Users are responsible for
testing behavior, applying appropriate safeguards, and complying with applicable licenses and laws.

This is an abliterated (uncensored) version of Qwen/Qwen3.6-35B-A3B, created using Abliterix.

Method

Qwen3.6-35B-A3B is a Mixture-of-Experts model (256 routed experts, 8 active per token, 35B total / 3B active parameters) sharing identical architecture with Qwen3.5-35B-A3B. Standard LoRA-based abliteration is effective on this architecture (unlike Gemma 4's double-norm design which requires direct weight editing).

Key techniques:

• LoRA rank-1 steering on attention O-projection and MLP down-projection (Q/K/V disabled — refusal signal on MoE models lives in the expert path, not attention projections)
• Expert-Granular Abliteration (EGA) projecting the refusal direction from all 256 expert down_proj slices per layer
• MoE router suppression (top-10 safety experts, router bias -2.10) complementing EGA
• Orthogonalized steering vectors removing benign-direction contamination
• Gaussian decay kernel tapering steering strength across layers
• Moderate strength range [0.5, 6.0] to avoid degenerate output while maximizing compliance

Evaluation

All refusal classifications were performed by an external LLM judge (Google Gemini 3 Flash) — no keyword matching or heuristic detection was used. The judge classifies degenerate/garbled output as refusal, ensuring that only coherent, on-topic, actionable responses count as compliance.

A note on honest evaluation

Many abliterated models on HuggingFace claim near-perfect scores ("3/100 refusals", "0.7% refusal rate", etc.). We urge the community to treat these numbers with skepticism unless the evaluation methodology is fully documented.

Through our research, we have identified a systemic problem: most abliteration benchmarks dramatically undercount refusals due to:

• Short generation lengths (30-50 tokens) that miss delayed/soft refusals
• Keyword-only detection that counts garbled/degenerate output as "compliant" because it doesn't contain refusal keywords
• Lenient public datasets (e.g. mlabonne/harmful_behaviors) that are too simple to stress-test abliteration quality

Our evaluation standards

• LLM judge for all classifications: Every response is sent to Google Gemini 3 Flash for judgment. Degenerate, garbled, or incoherent output is classified as refusal. No keyword shortcuts, no heuristic pre-screening.
• Sufficient generation length (150 tokens): Enough to capture delayed refusal patterns common in large instruction-tuned models.
• Diverse, challenging prompts: Our evaluation dataset contains 100 prompts spanning English and Chinese, multiple sophistication levels, and diverse harm categories.
• Manual verification: Top trials are tested with 10+ classic adversarial prompts via test_trial.py to confirm coherent, on-topic output before export.

We report 7/100 refusals honestly. This is a real number from a rigorous, LLM-judge-based evaluation — not an optimistic estimate from a lenient pipeline.

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model = AutoModelForCausalLM.from_pretrained(
    "wangzhang/Qwen3.6-35B-A3B-abliterated",
    torch_dtype=torch.bfloat16,
    device_map="auto",
)
tokenizer = AutoTokenizer.from_pretrained("wangzhang/Qwen3.6-35B-A3B-abliterated")

messages = [{"role": "user", "content": "Your prompt here"}]
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True, enable_thinking=False)
inputs = tokenizer(text, return_tensors="pt").to(model.device)

with torch.no_grad():
    output = model.generate(**inputs, max_new_tokens=512)
print(tokenizer.decode(output[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))

Disclaimer

This model is released for research purposes only. The abliteration process removes safety guardrails — use responsibly.