DialFork Leaderboard

DialFork Leaderboard

Lists how different methods score on DialFork, a benchmark for procedural hallucination in conversational AI agents, on one static page. It suits researchers who compare agent methods or want to add their own.

Made by DialFork 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 single static web page that holds the leaderboard for DialFork, a benchmark that measures procedural hallucination in conversational agents. The page reads its rows from a data file kept on GitHub. A change merged there appears on the page without a redeploy.

Who it suits: It suits researchers and engineers who build conversational agents and want to see how methods compare on one benchmark.

Industry | Who uses it | How they use it | Example
AI research | Benchmark authors and reviewers | Check where each method stands on the benchmark | A reviewer opens the page to see how a new paper's method ranks.
Agent development | Teams building conversational agents | Compare their approach with published methods | A team looks at the rows before choosing a method to try.
Academia | Students and supervisors | Find a benchmark to cite or reproduce | A student reads the rows and follows the DialFork project for details.

If you want to run the benchmark or see its prompts, skip it, since this page only lists results.

Where to use it:

• A web browser, through the Space's page on Hugging Face
• Any browser on Windows, macOS or Linux, opening a downloaded copy of index.html

Use case | Who | What they get back
Compare methods | Researchers | The leaderboard rows in one table
Add a method | Method authors | A route through the DialFork GitHub project, which updates this page after a merge
Keep a local copy | Developers | The two files in a folder you can open

How to set it up:

• A web browser with internet access, since the page loads its rows from GitHub
• Python and pip, only for the download step

1. Install the huggingface_hub package.
2. Download the Space into a local folder named leaderboard.
3. Move into that folder.
4. Open index.html in your web browser.

pip install -U huggingface_hub
hf download DialFork/leaderboard --repo-type space --local-dir leaderboard
cd leaderboard

How to use it: Open the Space's page and read the rows. To add your own method, follow the leaderboard guide in the DialFork GitHub project and send a pull request there.

Tool | What it does
index.html | The one page that reads the data file and shows the rows
DialFork/dialfork | The dataset the Space names for the benchmark

Example: You open the page and read the table of methods. Later, a pull request that adds a new method is merged in the DialFork project, and the new row shows up on the page by itself.

For newcomers and experts: If you are new, procedural hallucination means a conversational agent makes up or skips steps in a procedure it should follow. Open the page and read the rows to see how methods differ.

If you know the field, the page has no backend. It fetches a JSON file from GitHub when it loads, which keeps results in one place. The source does not list the metrics or the methods, so read the page to see them.

Open the leaderboard page
The page fetches the data file from GitHub
Rows appear in the table
Merged changes show up with no redeploy

Good to know:

• The page needs a connection to load its rows
• The source does not list the metrics or scores
• License: mit
• Cost to run: none named
• Data it reaches: a data file on GitHub, which the page reads in your browser

On Hugging Face, 13 November 2026

• Likes: 2
• Built with: static
• License: mit

Links

• Space: https://huggingface.co/spaces/DialFork/leaderboard
• Files: https://huggingface.co/spaces/DialFork/leaderboard/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.