Downloads · 30 days
937
19% of all-time downloads
Qwen3.5-0.8B-uncensored-MNN
darkmaniac7/Qwen3.5-0.8B-uncensored-MNN
Qwen3.5-0.8B-uncensored-MNN is a image-text-to-text model from darkmaniac7. Use it for the image-text-to-text task on the model card, and read the license before you ship it in a product. It is set up for mnn. The card lists the license as apache-2.0.
huihui-ai's abliterated Qwen3.5 0.8B (fewer refusals, vision included) as a 4-bit MNN bundle for chatting with text and images on your phone in TokForge.
All-time downloads
5K
Public
Repo size
543 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.weight533 MB · 97%
At a glance
- Task
- Image-Text-to-Text
- Library
- mnn
- License
- apache-2.0
- Access
- Public
- Created
- Jun 12, 2026
- Updated
- Nov 11, 2026
- SHA
- 131c9ffa
Try a prompt
Base models
Model card
From the Hugging Face model README
Qwen3.5 0.8B Abliterated (MNN)
huihui-ai's abliterated Qwen3.5 0.8B (fewer refusals, vision included) as a 4-bit MNN bundle for chatting with text and images on your phone in TokForge.
Published by TokForge for on-device inference.
| What it is | Vision chat model (text + images) |
| Format | MNN bundle, 4-bit weights (block 64, HQQ), mix-mask export |
| Runs on | Android phones (CPU) |
| Phone class | No catalog class; needs more free RAM than its 0.47 GB main weights file |
| Download | 549 MB (9 files) |
| Base model | huihui-ai/Huihui-Qwen3.5-0.8B-abliterated |
| License | Apache-2.0 |
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.
Use it in TokForge
- Install TokForge from Google Play, or join the iOS TestFlight.
- On Android, open Models > Browse, search for
Qwen3.5-0.8B-uncensored-MNNand download it. - Load it and start a chat. The model runs on the phone; once it is downloaded, chatting needs no connection.
Collections: TokForge Qwen Models (MNN), TokForge Vision Models.
Files
| File | Size | What it is |
|---|---|---|
config.json | 693 B | MNN runtime settings |
export_args.json | 1.12 KB | Export settings used for the conversion |
llm.mnn | 2.16 MB | Model graph |
llm.mnn.json | 5.36 MB | Model graph as JSON (reference copy) |
llm.mnn.weight | 470 MB | Model weights |
llm_config.json | 8.74 KB | Model settings and chat template |
tokenizer.mtok | 7.34 MB | Tokenizer (new MNN binary format) |
visual.mnn | 252 KB | Vision encoder graph |
visual.mnn.weight | 63.0 MB | Vision encoder weights |
About the model
MNN conversion of huihui-ai/Huihui-Qwen3.5-0.8B-abliterated (vision included) for on-device inference. All credit for the abliteration goes to huihui-ai. Abliteration removes the model's refusal direction from its weights, so it declines fewer requests than the original Qwen3.5 0.8B. TokForge runs Qwen3.5 on the CPU by default.
| Architecture | Qwen3.5 (24 layers: 18 LinearAttention + 6 full attention), with vision encoder |
| Parameters | 0.8B (4-bit quantized) |
| Quantization | 4-bit, block 64, HQQ (matches the taobao-mnn 0.8B bundle parameters) |
| Tokenizer | tokenizer.mtok (new format) |
Runtime requirements
- Hybrid attention: 18 LinearAttention + 6 full-attention layers of 24: self-describing metadata (
attention_type:"mix",layer_nums:24,sliding_window:4), stacked mask graph[2,1,1,seq,kv] - Requires an MNN runtime with the mix-mask
gen_attention_maskpath (MNN 3.5.0 or later, as in TokForge's fork of MNN); legacy single-mask runtimes are not compatible
How it was made
- Exporter: llmexport from TokForge's fork of MNN, rebased on MNN 3.5.0 (with upstream MNN's attention-type detection)
- Settings: 4-bit weights, block 64, HQQ (see
export_args.json)
File checksums (md5)
3af7de552fb9cea05e67f615d96efb53 llm.mnn
10fe7c624159f62c69d93a0e836fc80f llm.mnn.weight
88bbc9851fc9e2aae364eaf59698326a llm_config.json
3c183b4e10bef80909cb6998fe5e4fa6 tokenizer.mtok
06b782392bdb4ca19a1045c3709fdbb6 visual.mnn
e48e427a48dc363278ecaacb7810db6a visual.mnn.weight
Limitations
- Intended for TokForge / MNN on-device inference
- This is a runtime bundle, not a standard Transformers checkpoint
- Quantization (Q4) may slightly reduce quality compared to the full-precision original
- Abliterated models have had their refusal behavior removed: use responsibly
License
Apache-2.0, as the original Qwen3.5 0.8B and huihui-ai's abliteration.
Credits
- Huihui-Qwen3.5-0.8B-abliterated by huihui-ai (Hugging Face). Used under the Apache-2.0 license.
- Qwen3.5-0.8B by the Qwen team (Alibaba Cloud) (Hugging Face). Used under the Apache-2.0 license.
- Converted with llmexport from MNN by Alibaba (Apache-2.0).
- Task
- Image-Text-to-Text
- Library
- mnn
- License
- apache-2.0
- Languages
- en
- Created
- Jun 12, 2026
- Updated
- Nov 11, 2026