Underdog-Woof-4B-1.1

Underdog-Woof-4B-1.1

Generates text as a compact 4.2B model, post-trained from Qwen 3 4B to work better with Underdog's own setup and quantized to 4-bit for MLX. For people running small local models for non-commercial use.

Made by ConwayResearch on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

About

What it is: Underdog-Woof-4B-1.1 is a small text generation model from ConwayResearch. The authors post-trained Qwen 3 4B so it works better with Underdog's own setup, then quantized it to 4-bit for MLX. Hugging Face lists 4.2B parameters, Safetensors files and English as the language.

Who it suits: People who want a compact model they can run in MLX tools and who will use it for non-commercial work.

Industry | Who uses it | How they use it | Example
Software development | Developers who work with MLX | Load the quantized files in an MLX based tool | Trying a small local model before choosing a larger one
Research | Researchers on non-commercial projects | Run the model as a small English text generator | Comparing a 4-bit model with other small models
Hobby projects | People who experiment at home | Download the files and test them on their own computer | Checking how a 4B model handles everyday text

Skip it if you need commercial use without a separate license, or a model for languages other than English.

Where to use it: It is meant for MLX tooling, and the files download on Windows, macOS or Linux.

• MLX: the library the files were prepared for
• Windows, macOS or Linux: where you can fetch the files with Python
• Underdog's own setup: the system the model was tuned to work with

Use case | Who | What they get back
Small local text generation | Developers | English text from a compact model
Setup testing | People using Underdog's own setup | A model tuned for that setup
Quantization study | Researchers | A 4-bit MLX example to compare

How to set it up: You need Python and enough disk space for the files. Hugging Face lists about 16.8 GB of weights stored as U32.

• Python with pip
• Disk space for the model files
• An MLX based tool to load them

1. Install the Hugging Face hub library.
2. Download the model files to your computer.
3. Load them in your MLX based tool.

pip install -U huggingface_hub
hf download ConwayResearch/Underdog-Woof-4B-1.1

How to use it: The card gives no run command of its own, so open the model page on Hugging Face for how to load it. The one command it lists downloads the files.

Tool | What it does
hf download | Fetches the model files to your computer

Example: You download the model with the Hugging Face command line tool. The model files are saved on your computer, ready for you to load in an MLX based tool.

For newcomers and experts: If you are new to this, download the files first and check that they arrive. Then load them in a tool that supports MLX and try a short English prompt.

If you know the field, this is Qwen 3 4B after post-training for Underdog's own setup, stored in 4-bit MLX form. The card does not describe that setup or the training data, so test it on your own tasks.

Qwen 3 4B base model
Post-trained for Underdog's own setup
Quantized to 4-bit for MLX
You download and load the files

Good to know: The card is short, so benchmarks and training details are not given.

• License: CC BY-NC 4.0, so non-commercial use with credit to Underdog is allowed. Commercial use needs a separate license from the authors.
• Cost to run: none named
• Data it reaches: the card names none
• Not for: commercial use without a separate license, and parts from upstream keep their own licenses

On Hugging Face, 13 November 2026

• Task: text generation
• Library: mlx
• Parameters: 4.2B
• Weights: about 16.8 GB in U32
• Formats: Safetensors
• Downloads in the last 30 days: 1,052
• Likes: 26
• License: cc-by-nc-4.0
• Languages: en
• Tags: safetensors, qwen3_5, 4bit, text-generation, conversational, 4-bit

Links

• Model: https://huggingface.co/ConwayResearch/Underdog-Woof-4B-1.1
• Files: https://huggingface.co/ConwayResearch/Underdog-Woof-4B-1.1/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download ConwayResearch/Underdog-Woof-4B-1.1

Then follow the usage in its model card. Nothing in this kit runs by itself.