A tiny $104-trained model that answers multiple-choice questions
Made by Supersonic Labs
View profile on Hugging Face (opens in a new tab)Julia-1 is a 144-million-parameter decision model, tiny beside the big chat models, that takes background context, a question and a list of possible answers, then picks the right one.
That single interface covers classifying text, routing requests, scoring options in order, and yes-or-no answers. It runs as PyTorch code on CPU or CUDA, or in your browser through a WebGPU playground the team ships. The Brazilian lab behind it reports about US$104 in cloud GPU spend to train it, and publishes its evaluation — 73% on a 2,000-case decision test, 94% on a news-topic pilot — while openly listing its failures: long option lists, outside knowledge, and multi-step reasoning.
Can I use this?
You'll need Python with PyTorch, or just a modern browser for the WebGPU demo
Worth knowing The creators call it a first test of their training system and publish its weak spots: long option lists, questions needing outside knowledge, and multi-step reasoning. A successor (Julia 2) is already in development. License: Apache-2.0. BuildTube has not run or verified this model.
Why it's here
- 327 people have liked it on Hugging Face.
- It was downloaded 2,556 times in the last 30 days.
Numbers from the snapshot taken 1 October 2026; not refreshed since.
Behind it
See the code on Hugging FaceBuildTube has not run or verified this project. Everything above is written from what the creator published.
Made this?