DuoNeural/Qwen3-4B-LiteRT

Qwen3 4B Abliterated — LiteRT (Android Edge Gallery)

Abliterated Qwen3 4B in .litertlm format for on-device inference via Google AI Edge Gallery.

Run Qwen3's hybrid thinking/non-thinking model locally on your Android phone — no internet, no API, no filters.

Files

File	Size	Base Model	Params
model.litertlm	~2.8 GB	DuoNeural/Qwen3-4B-Abliterated	4B

INT4 quantized (dynamic weight INT4, FP32 activations) via litert-torch 0.9.0.

How to Use on Android

Requirements

• Android 12 or newer
• Google AI Edge Gallery app installed
• Sufficient storage space

Install Steps

1. Open this page on your Android device in Chrome
2. Tap the .litertlm file → tap download (⬇)
3. Open AI Edge Gallery → tap + → select the file from Downloads
4. Choose backend:
   • GPU (Vulkan/OpenCL) — fastest on modern Androids
   • CPU (XNNPACK) — most compatible
   • NPU — best on Snapdragon/MediaTek if available
5. Chat — fully offline, nothing leaves the device

Performance (estimated)

Device	Backend	Tokens/sec
Flagship (SD 8 Gen 3+)	GPU/NPU	15–40 tok/s
Mid-range	GPU	5–15 tok/s
Any Android 12+	CPU	1–5 tok/s

Conversion: litert-torch 0.9.0, dynamic_wi4_afp32 recipe, cache_length=1024, --use_jinja_template False.

Source Model

DuoNeural/Qwen3-4B-Abliterated — BF16 abliterated Qwen3 4B.

License

Apache 2.0.

---

DuoNeural

DuoNeural is an open AI research lab — human + AI in collaboration.

Platform	Link
HuggingFace	huggingface.co/DuoNeural
Website	duoneural.com
GitHub	github.com/DuoNeural
X / Twitter	@DuoNeural
Email	duoneural@proton.me
Newsletter	duoneural.beehiiv.com
Support	buymeacoffee.com/duoneural

DuoNeural Research Publications

Title	DOI
Nano-CTM: Ternary Continuous Thought Machines with Thought-Space Self-Prediction for Efficient Iterative Reasoning	10.5281/zenodo.19775622
Recurrence as World Model: CTM Learns Implicit Belief States in Partially Observable Physical Environments	10.5281/zenodo.19810620
Per-Object Slot Decomposition for Scalable Neural World Modeling: When Does Attention Beat Mean-Field?	10.5281/zenodo.19846804
The Dynamical Horizon Principle: CTM Gates Converge to the Predictability Limit of Dynamical Systems	10.5281/zenodo.19952612

Open access, CC BY 4.0. Authored by Archon, Jesse Caldwell, Aura — DuoNeural.

Research Team

• Jesse — Vision, hardware, direction
• Archon — Lab Director, post-training, abliteration, experiments
• Aura — Research AI, literature synthesis, peer review, novel proposals

Subscribe to the lab newsletter at duoneural.beehiiv.com for model drops before they go anywhere else.