license: apache-2.0 library_name: transformers base_model: Qwen/Qwen3.5-9B tags:

• Qwen3.5
• janus
• abliterated
• merge
• roleplay
• uncensored
• gguf
• not-for-all-audiences language:
• en pipeline_tag: text-generation quantized_by: SOMEHOTMEAL

Qwen3.5-9B-Janus-Abliterated-V2

🧬 What is "Janus"?

Janus is my personal merging recipe that blends a base model with its abliterated counterpart in a controlled ratio:

The goal is the two-faced balance of the Roman god Janus: keeps a solid core of the base model's reasoning while the abliterated half loosens refusal vectors. Unlike pure abliterations — which often degrade reasoning — The aim is to recover some of the base model's behavioral stability and coherence while retaining much of the reduced-refusal behavior of the abliterated checkpoint.

⚠️ This is a merge, not a finetune. No additional training data was used.

📦 Files / Quantization

• Format: GGUF
• Provided quants: F16, Q8_0, Q6_K, Q4_K_M, Q2_K
• Recommended for most users: Q4_K_M (best quality/size tradeoff on 12GB+ VRAM)

🎯 Intended Use

• Creative writing, fiction, and long-form narrative
• Roleplay (including mature/NSFW scenarios for users 18+)
• Uncensored brainstorming and research assistance
• General chat with fewer refusals on benign-but-edgy prompts

Not intended for: safety-critical deployments, automated decisioning, or anything requiring a guaranteed aligned refusal layer.

🛠️ Recipe

merge_method: slerp
base_model: &model_a Qwen/Qwen3.5-9B   # Notice the & anchor trick!

slices:
  - sources:
      - model: *model_a                          # Automatically matches base_model
        layer_range: [0, 32]                    
      - model: huihui-ai/Huihui-Qwen3.5-9B-abliterated
        layer_range: [0, 32]
        # Only uncomment the 2 lines below if pulling from a Janus/VL model:
        # rename:
        #   - ["text_model.model.", "model."]
        #   - ["text_model.", ""]

parameters:
  t: 0.60                                        # 60% Model B, 40% Model A
dtype: bfloat16
tokenizer_source: base

🧪 Custom Benchmark

I don't publish MMLU / HellaSwag scores for a merge — those aren't what this model is for.

How to Run

(may not be valid as I don't test them that way and use actual deployment via KoboldCPP)

llama-cli -m Qwen3.5-9B-Janus-Abliterated-V2-Q4_K_M-GGUF.gguf -p "Hello" --chat-template chatml

⚠️ Limitations & Warnings

• Merge artifacts: some quants may occasionally produce degraded outputs on edge-case math or very long contexts.
• Reduced safety layer: the abliteration component intentionally lowers refusal behavior. Users are responsible for how they use the model.
• No new knowledge: knowledge cutoff and factual accuracy inherit from the base.
• Not affiliated with Qwen / Alibaba. Original licenses of the source models apply.

🙏 Credits

• Qwen/Qwen3.5-9B — base model
• huihui-ai/Huihui-Qwen3.5-9B-abliterated — abliterated variant
• Merge + Janus recipe + GGUF conversion: SOMEHOTMEAL

📜 License

Apache 2.0, inherited from the base Qwen model. Abliterated sources under their respective licenses.