THX-02

THX-02

Gives a probability for every option of a question you ask about a described situation, using the THX-02 decision model, with an optional thinking mode. A Gradio demo for researchers and builders testing calibrated decisions.

Made by HAL-X-AI on Hugging Face. These notes are Mindket's, written from its Space page and README; the app's own files and license are on Hugging Face.

About

What it is: This is a Gradio demo of THX-02, a decision model that answers with a probability for each option instead of one pick. You describe a state and add typed questions of three kinds: choice, a yes/no kind called noul, and score. An optional thinking mode can be switched on. It runs on zero-a10g hardware at Hugging Face.

Who it suits: Researchers who study how well a model's probabilities match reality, and developers who want a decision step that reports its uncertainty.

Industry | Who uses it | How they use it | Example
Research | ML researchers | Look at calibrated probabilities from a decision model | Compare the options of one choice question
Software | Agent builders | Test a decision step that returns uncertainty | Try a yes/no question on a described state
Analytics | Analysts | Describe a situation and ask typed questions | Read the probability for each option
Education | Students | Learn what calibrated probabilities look like | Switch thinking on and off and compare

Skip it if you need one plain answer, since the model returns a probability for every option.

Where to use it:

• A web browser, on the Space's page at Hugging Face
• Your own computer, as a local copy, once Python 3.12 is installed
• The listing carries an MCP tag, but the page does not explain how to use it that way

Use case | Who | What they get back
Rank the options of a choice | Researchers and analysts | A calibrated probability for every option
Answer a yes/no question | Builders | A probability for yes and for no
Score something | Builders | A probability-based score for the item

How to set it up:

• Nothing at all if you only use the online Space
• Python 3.12 and pip, for a local copy
• The Python packages listed in requirements.txt

1. Open the Space's page at Hugging Face to use it online.
2. For a local copy, install the Hugging Face hub package and download the Space.
3. Move into the new folder and install Gradio at the pinned version.
4. Install the packages from requirements.txt.
5. Start the app with Python.

pip install -U huggingface_hub
hf download HAL-X-AI/thx-02 --repo-type space --local-dir thx-02
cd thx-02
pip install gradio==6.31.0
pip install -r requirements.txt
python app.py

The model runs inside the app process with transformers, not through a separate vLLM server. The page does not say what hardware a local run needs.

How to use it: Online, open the Space, describe a state and add your typed questions. You can load the bundled examples from the examples folder. On your own machine, run the app file and open the page Gradio serves in your browser.

Tool | What it does
choice | A question with several options, answered with a probability for each
noul | A yes/no question
score | A scoring question
python app.py | Starts the Gradio app on your computer

Example: You describe a short situation in a few sentences and add one choice question with three options. The app returns a probability for each option, so you can see how sure the model is rather than only which option leads.

For newcomers and experts: If you are new, a calibrated probability is a number meant to match how often an answer turns out right. Start with the bundled examples, then change one detail and watch the numbers move.

If you know the field, the checkpoint is the authors' merge of the decision weights with a think LoRA, built by their own function. Thinking is optional and gates itself.

You describe a state
You add typed questions
THX-02 runs, thinking if enabled
A probability returns for every option

Good to know:

• License: apache-2
• Cost to run: none named, though the Space runs on zero-a10g hardware at Hugging Face
• Data it reaches: the state and questions you type, and the model files from Hugging Face

On Hugging Face, 13 November 2026

• Likes: 1
• Built with: gradio
• License: apache-2.0

Links

• Space: https://huggingface.co/spaces/HAL-X-AI/thx-02
• Files: https://huggingface.co/spaces/HAL-X-AI/thx-02/tree/main

Try it

Open the Space in your browser, or copy it to your own Hugging Face account from its page. Nothing in this kit runs by itself.