Modelpublic

AdvancedDataIntelligence/adi-qwen3.5-4b-glm5.2-general-GGUF

sourceHugging Faceapache-2.0updated 3mo agoView on Hugging Face

7likes575downloads

Model Card

adi-qwen3.5-4b-glm5.2-general

Part of the ADI (Advanced Data Intelligence) model line — ADI Qwen3 series.

A small, fully local model that reasons and answers like a frontier teacher. Built by distilling glm-5.2 general-knowledge responses into a Qwen3.5-4B student with a bf16 LoRA fine-tune, then merged, converted, and quantized to GGUF. The student base retains native tool calling and a long context window.

Capabilities

Size	Context	Input	Output	Tools
2.7 GB	262K	🅣 Text	Text	✅


Base model	Qwen/Qwen3.5-4B
Teacher	glm-5.2 (responses distilled, thinking disabled)
Method	bf16 LoRA SFT (rank 16) → merge → GGUF
Quantization	Q4KM (~2.7 GB)
License	Apache-2.0 (inherited from Qwen3.5-4B)
Context	262K (inherited from base)
Tool calling	Supported (inherited from base)

Run it

Pull directly into Ollama:

bash

ollama run hf.co/AdvancedDataIntelligence/adi-qwen3.5-4b-glm5.2-general-GGUF:Q4_K_M

Or download the .gguf and point any llama.cpp-based runtime at it.

Try it live

A hosted demo is available as a Hugging Face Space — chat with the model directly in your browser, no install required.

▶ Launch the demo

Chat with the model directly in your browser — no install required.

What this model is

This is a knowledge distillation: a strong teacher (glm-5.2) generated high-quality answers across ~2,000 diverse general-knowledge prompts, and the Qwen3.5-4B student was fine-tuned to imitate them. The result reasons and responds noticeably more like its teacher on general topics, while staying small enough to run on a single consumer GPU.

What distillation does — and doesn't do. It transfers the teacher's reasoning style and answer quality, not net-new facts. A 4B model won't become an encyclopedia. For raw factual recall, retrieval-augmented generation (RAG) is the right tool, not fine-tuning. What you get here is a 4B that structures and explains like a much larger model on topics it already partly knows.

Training

Metric	Value
Training pairs	2,068
Teacher tokens generated	~1.36M
Epochs	3
Steps	777
Final train loss	0.9346
LoRA rank / alpha	16 / 16
Trainable params	21.2M (0.47% of 4.56B)
Precision	bf16 (not 4-bit — see note)
Hardware	single RTX 5060 Ti (16 GB)
Training time	2h 53m

The seed prompts were drawn from the human-written Databricks Dolly-15k dataset (filtered to remove items requiring an attached context passage, then deduplicated). The teacher was queried with thinking disabled so the student learns clean final answers rather than chain-of-thought it is too small to reproduce well.

Notes for re-builders

—Qwen3.5 trains in bf16 LoRA, not 4-bit QLoRA. Its gated-delta / Mamba-hybrid layers quantize poorly during training; 4-bit costs accuracy. bf16 LoRA uses ~10 GB on a 4B — comfortable on a 16 GB card.
—Version pins: Qwen3.5 requires transformers >= 5.2.0 to be recognized, while the Unsloth training stack caps at <= 5.5.0. The working version is transformers == 5.5.0 with numpy < 2.3.
—GGUF conversion was done with llama.cpp's convert_hf_to_gguf.py, which already understands the Qwen3.5 SSM/MTP architecture.

Intended use

General-purpose local assistant: explanations, reasoning, Q&A, and tool-calling workflows where a small, private, offline-capable model is preferred over a hosted API. Not intended as a source of authoritative facts without retrieval.

License

Apache-2.0, inherited from the Qwen3.5-4B base model. You are free to use, modify, and redistribute under the terms of that license. Distilled training data was generated using glm-5.2; users should review the teacher model's terms for their own use case.

Built at [theLAB](https://thelabsource.com) — Learning. Algorithms. Breakthroughs.