Clinassist Train

Clinassist Train

Holds the training scripts and Gradio app used to tune a Qwen model for the author's Clinassist project with GRPO. Suits machine learning developers who want to read or rerun that training.

Made by Gs2004 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 Space holds the training side of the author's Clinassist project. It is used to tune a small Qwen model, and it names a LoRA adapter called clinassist-grpo-lora and the 4-bit Qwen2.5 1.5B Instruct model from Unsloth as the base. The files include a GRPO training script and a patient data file.

Who it suits: It suits machine learning developers who tune small language models, and anyone who wants to see how a clinical assistant project organises its training code.

Industry | Who uses it | How they use it | Example
Machine learning | Developers | Read a working GRPO training setup for a small Qwen model | Open train_grpo_script.py and follow the steps
Health technology | Prototype builders | See how a clinical assistant project arranges its training files | Look at the patient data file's layout
Research | Students | Study reinforcement learning tuning on a small model | Compare the adapter with its named base model

Skip it if you want a finished medical tool, since this is training code for the author's own project.

Where to use it:

• A browser: open the Space's page on Hugging Face, with nothing to install
• A copy on Windows, macOS or Linux, using Python

Use case | Who | What they get back
Reading the training code | Developers | A GRPO script and an install-and-train helper
Rerunning a tune | Practitioners | A training run on the data file in the repository
Studying the adapter | Researchers | A LoRA adapter tied to a named base model

How to set it up: The online copy needs nothing. For a copy on your own computer, you need the following.

• Python with pip on Windows, macOS or Linux
• The Hugging Face hub client, which the first step installs
• The packages listed in requirements.txt

1. Install the Hugging Face hub client.
2. Download the Space's files into a local folder.
3. Move into that folder.
4. Install the Gradio version the Space pins.
5. Install the remaining packages from requirements.txt.
6. Start app.py.

pip install -U huggingface_hub
hf download Gs2004/clinassist-train --repo-type space --local-dir clinassist-train
cd clinassist-train
pip install gradio==6.13.0
pip install -r requirements.txt
python app.py

How to use it: Online, open the Space's page and use the controls it shows. On your own computer, start the app with the command below. The source does not say how the training is started, so read the app and the two training scripts before running anything.

python app.py

Example: You download the Space and open train_grpo_script.py next to train_patients.json to see how the training reads its data. The source does not give a run command or any results.

For newcomers and experts: If you are new, this is a project's workshop, not a finished product. Open the Space and read the files before you start anything, since training can take a long time.

If you know the field, the setup tunes a 4-bit Qwen2.5 1.5B Instruct base with GRPO and saves the result as a LoRA adapter. The Space sits on cpu-basic hardware, so heavy training probably happens elsewhere. The source does not say where.

You open the Space or run app.py
The app and scripts load the patient data
GRPO tunes the small Qwen base model
A LoRA adapter comes out

Good to know: The source is short, so results and limits are not described.

• License: mit, as listed on Hugging Face
• Cost to run: no API key or paid service is named, and the online copy runs on cpu-basic hardware
• Data it reaches: the patient data file in the repository and the models it names on Hugging Face
• Not for: the source lists no exclusions, but do not use it for real medical decisions without expert review

On Hugging Face, 13 November 2026

• Likes: 1
• Built with: gradio
• License: mit

Links

• Space: https://huggingface.co/spaces/Gs2004/clinassist-train
• Files: https://huggingface.co/spaces/Gs2004/clinassist-train/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.