Qwenputer Front Panel

Qwenputer Front Panel

Replays recorded runs of Qwenputer, a Qwen3.5-4B model with a RISC-V CPU built into extra layers, beside a front panel of layers, registers and RAM. It runs in your browser.

Made by jgalego 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: Qwenputer Front Panel is a static web page that jgalego published as a Hugging Face Space. It replays recorded conversations of Qwenputer, a Qwen3.5-4B model that has a 32-bit RISC-V CPU folded into 11 additional layers. Whenever the model requests a tool, the CPU executes it during the model's normal computation, and the result comes out as plain text.

Who it suits: People curious about how a language model can run a tool inside its own layers, and developers who want to read how the panel is built.

Industry | Who uses it | How they use it | Example
Machine learning | Model researchers | Replay a recorded conversation and watch the CPU work | Following which layers a token passes through
Computer science | Educators | Show the registers, RAM and program counter during a tool call | Teaching how a small processor executes a program
Web development | Front-end developers | Download the files and read app.js and index.html | Seeing how a data-driven replay page is built

Skip it if you want to chat with the model yourself, since this page replays recorded runs.

Where to use it: The panel opens in these places.

• A web browser: the Space runs there with nothing to install
• Windows, macOS or Linux: download the files with Python, then open index.html in a browser

Use case | Who | What they get back
Watch a tool call | Anyone curious | A recorded conversation beside the CPU's layers, registers, RAM and tool-call tokens
Check the arithmetic tool | Researchers | The calc tool's work on numbers of up to 62 digits
Play with Life | Anyone curious | Conway's Game of Life on a 32 by 32 grid in the CPU's RAM

How to set it up: Online use needs only a browser. A local copy needs a few more things first.

• A web browser
• Python on Windows, macOS or Linux, for a local copy

1. Open the Space's page on Hugging Face if you only want to use it online.
2. For a local copy, install the Hugging Face hub library with pip.
3. Download the Space's files into a folder called qwenputer-panel.
4. Move into that folder.
5. Open index.html in a web browser.

pip install -U huggingface_hub
hf download jgalego/qwenputer-panel --repo-type space --local-dir qwenputer-panel
cd qwenputer-panel

How to use it: Online, open the Space's page and pick a recorded conversation. The front panel shows the machine's state as each token is read. With a local copy, go into the qwenputer-panel folder and open index.html in your browser.

cd qwenputer-panel

Tool | What it does
calc | Precise arithmetic for numbers as long as 62 digits
letters | Spells a word and counts its letters, reading the bytes the model's tokens hide
weekday | Calendar arithmetic
days_between | Calendar arithmetic between two dates
life | Runs Conway's Game of Life on a 32 by 32 grid in the CPU's RAM

Example: You replay a recorded conversation in which Qwen asks for a tool. The panel shows the CPU reading the tool-call tokens, running the small program and writing the answer back as ordinary text.

For newcomers and experts: If you are new, open the Space's page and watch one recorded conversation from start to end. The panel shows what happens inside the model while it answers.

If you know the field, each recorded run used greedy decoding. A second, reference RISC-V machine ran alongside in step and had to reach the same output for every token the CPU was responsible for. The panel displays that machine's registers and RAM. A replay of the upstream IR then re-checked every position the CPU owned. The tools come from a small C program named toolbox.c.

Qwen calls a tool
The CPU in the extra layers runs it
The answer is written as plain text
The panel shows registers, RAM and layers

Good to know: It replays recorded runs, so it does not run the model for you.

• Models: divagr1925/Qwen3.5-4B-Qwenputer and its base, Qwen/Qwen3.5-4B
• Size: 18 files in the repository
• License: apache-2
• Cost to run: none named
• Data it reaches: the recorded conversations and data in the repository
• Not for: running the model live, since the page replays recorded conversations

On Hugging Face, 13 November 2026

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

Links

• Space: https://huggingface.co/spaces/jgalego/qwenputer-panel
• Files: https://huggingface.co/spaces/jgalego/qwenputer-panel/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.