Modelpublic

darkmaniac7/Qwen3.5-4B-uncensored-MNN

sourceHugging Faceapache-2.0updated 3mo agoView on Hugging Face

4likes1.8kdownloads

Model Card

TokForge

—Website: https://tokforge.ai
—Discord: https://discord.gg/Acv3CBtfVm
—Google Play: https://play.google.com/store/apps/details?id=dev.tokforge
—iOS TestFlight: https://testflight.apple.com/join/jnufjzRr

Runs on-device in the TokForge app.

Qwen3.5-4B Uncensored — MNN Format

This is an MNN-converted version of [huihui-ai/Huihui-Qwen3.5-4B-abliterated](https://huggingface.co/huihui-ai/Huihui-Qwen3.5-4B-abliterated) for on-device mobile inference.

All credit for the abliteration work goes to huihui-ai. We only performed the MNN conversion and quantization for mobile deployment.

What is this?

—Base model: Qwen/Qwen3.5-4B by Alibaba
—Abliteration by: huihui-ai — removes refusal behavior via orthogonal projection (FailSpy technique)
—MNN conversion by: darkmaniac7 — 4-bit quantization (block size 128) for mobile GPU/CPU inference
—Purpose: On-device roleplay, creative fiction, and mature content without refusal

Model Details

Property	Value
Architecture	Qwen3.5 (LinearAttention)
Parameters	4B
Quantization	4-bit (block 128)
Format	MNN (Alibaba Mobile Neural Network)
Size on disk	~2.5 GB
Backend	CPU (auto-routed — LinearAttention is faster on CPU than OpenCL)

Performance (measured on-device)

Device	SoC	Backend	Decode tok/s
RedMagic 11 Pro	SM8850 (SD 8 Elite 2)	CPU	17.7–19.8
Samsung S26 Ultra	SM8850	CPU	~18–20
Samsung S24 Ultra	SM8650 (SD 8 Gen 3)	CPU	~14

Usage

This model is designed for TokForge, an offline Android AI chat app. It can also be used with any MNN-compatible runtime.

TokForge (Android)

Models → Recommended → Roleplay → "Qwen3.5 4B Uncensored" → Download

Manual

Download all files and load with MNN's llm_demo or the MNN Transformer API.

Limitations and Intended Use

—Intended for TokForge / MNN mobile inference and local roleplay-style use.
—Backend behavior differs from classic Qwen3 because Qwen3.5 uses LinearAttention.
—Device performance varies significantly across SoCs and CPU/GPU routing.
—This repo is a mobile runtime/export artifact, not a standard Transformers release.

Files

File	Size	Description
`llm.mnn`	3.5 MB	Model graph
`llm.mnn.weight`	2.3 GB	4-bit quantized weights
`llm_config.json`	8 KB	Model configuration
`tokenizer.txt`	2.9 MB	Tokenizer vocabulary
`config.json`	342 B	HuggingFace config

Attribution

—Original model: Qwen3.5-4B by Alibaba Cloud (Apache 2.0)
—Abliteration: huihui-ai/Huihui-Qwen3.5-4B-abliterated by huihui-ai
—MNN framework: Alibaba MNN (Apache 2.0)
—MNN conversion: darkmaniac7

Community

—Website: tokforge.ai
—Discord: Join the Discord

License

Apache 2.0 (inherited from Qwen3.5)