license: apache-2.0
language:
- en
pipeline_tag: text-generation
base_model: DavidAU/Qwen3.5-13B-Deckard-Heretic-Uncensored-Thinking
tags: - mnn
- qwen3.5
- mobile
- on-device
- tokforge
- uncensored
- abliterated
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.
Qwen3.5-13B-Deckard-Heretic-Uncensored-Thinking-MNN
Pre-converted Qwen3.5-13B-Deckard-Heretic-Uncensored-Thinking in MNN format for on-device inference with TokForge.
Original model by DavidAU — converted to MNN Q4 for mobile deployment.
Model Details
| Architecture | Qwen3.5 (hybrid attention: full + LinearAttention, 48 layers) |
| Parameters | 13B (4-bit quantized) |
| Format | MNN (Alibaba Mobile Neural Network) |
| Quantization | W4A16 (4-bit weights, block size 128) |
| Vocab | 248,320 tokens |
| Source | DavidAU/Qwen3.5-13B-Deckard-Heretic-Uncensored-Thinking |
Description
Deckard Heretic by DavidAU — a 13B Qwen3.5 model with thinking/reasoning capability and uncensored output. Named after the Blade Runner protagonist, this model excels at complex reasoning tasks while remaining uncensored. Supports <think> tags for chain-of-thought reasoning.
Files
| File | Description |
|---|---|
llm.mnn |
Model computation graph |
llm.mnn.weight |
Quantized weight data (Q4, block=128) |
llm_config.json |
Model config with Jinja chat template |
tokenizer.txt |
Tokenizer vocabulary |
config.json |
MNN runtime config |
Usage with TokForge
This model is optimized for TokForge — a free Android app for private, on-device LLM inference.
- Download TokForge from the Play Store
- Open the app → Models → Download this model
- Start chatting — runs 100% locally, no internet required
Recommended Settings
| Setting | Value |
|---|---|
| Backend | OpenCL (Qualcomm) / Vulkan (MediaTek) / CPU (fallback) |
| Precision | Low |
| Threads | 4 |
| Thinking | Off (or On for thinking-capable models) |
Speculative Decoding
Speculative decoding is not recommended for this model. In our testing, draft pairing did not deliver reliable speedups on Qwen3.5 targets.
Performance
Actual speed varies by device, thermal state, and generation length. Typical ranges for this model size:
| Device | SoC | Backend | tok/s |
|---|---|---|---|
| RedMagic 11 Pro | SM8850 (Snapdragon 8 Elite 2) | CPU | 14.1 tok/s |
Attribution
This is an MNN conversion of Qwen3.5-13B-Deckard-Heretic-Uncensored-Thinking by DavidAU. All credit for the model architecture, training, and fine-tuning goes to the original author(s). This conversion only changes the runtime format for mobile deployment.
Limitations
- Intended for TokForge / MNN on-device inference on Android
- This is a runtime bundle, not a standard Transformers training checkpoint
- Quantization (Q4) may slightly reduce quality compared to the full-precision original
- Abliterated/uncensored models have had safety filters removed — use responsibly
Community
- Website: tokforge.ai
- Discord: Join our Discord
- GitHub: TokForge on GitHub
Export Details
Converted using MNN's llmexport pipeline:
python llmexport.py --path DavidAU/Qwen3.5-13B-Deckard-Heretic-Uncensored-Thinking --export mnn --quant_bit 4 --quant_block 128