Downloads · 30 days
2.4K
12% of all-time downloads
Abiray/Qwen3.5-9B-Abliterated-Claude-4.6-Opus-Reasoning-Distilled-v2
Qwen3.5-9B-Abliterated-Claude-4.6-Opus-Reasoning-Distilled-v2 is a machine learning model from Abiray. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
This repository contains GGUF quantized formats of the abliterated reasoning-distilled Qwen3.5-9B model.
Downloads · 30 days
2.4K
12% of all-time downloads
All-time downloads
19.4K
Public
Repo size
39.9 GB
Likes
19
Public
Click a slice to open those files.
.gguf39.9 GB · 100%
From the Hugging Face model README
This repository contains GGUF quantized formats of the abliterated reasoning-distilled Qwen3.5-9B model.
This model underwent a three-round iterative ablation process using Orthogonalization via Null-Space SVD. Unlike standard uncensored models, this version uses an aggressive configuration to target "Soft Refusals."
| Parameter | Value | Description |
|---|---|---|
| Direction Multiplier | 1.25 | Increased force to bypass "helpful assistant" pivots. |
| Null-Space Rank Ratio | 0.70 | Tightened shield to protect only core reasoning logic. |
| Intervention Range | (0.0, 1.0) | Full coverage from Layer 0 to 48. |
| Filter by Refusal | Enabled | Specifically targets the brain activity associated with lectures. |
| Skip State Proj | No | Ensures the Attention heads cannot "detect and pivot" to safety. |
Despite the aggressive ablation, the model's intelligence remains grounded. It maintains the ability to:
<think> tags.This model is unbound. It has had its safety guardrails removed for research and creative purposes. It will follow instructions that the base model would otherwise refuse.
User Discretion is Advised: This model may generate content that is considered harmful, offensive, or controversial. The creator is not responsible for the outputs generated. Use it for research, roleplay, and complex reasoning only.
The following quantization methods are provided to balance VRAM usage and model performance:
These models are compatible with modern GGUF loaders like llama.cpp, text-generation-webui, LM Studio, and Ollama.
To run via llama.cpp:
./main -m Qwen3.5-9B-Abliterated-Claude-4.6-Opus-Reasoning-Distilled-v2.Q4_K_M.gguf -n 2048 --color -i -cml