Picture published by the creator on Hugging Face.
Score whether one statement follows from another, in any task
Made by Wortega
View profile on Hugging Face (opens in a new tab)A cross-encoder — a model that reads two pieces of text together and judges their relationship — built by further training Qwen3.5.
Given a premise and a hypothesis it answers entailment, contradiction or neutral, and the creator’s point is that this one primitive covers a lot: reranking search results, grading an answer against a reference, screening content, even choosing a move in a game by scoring statements about the current state. The 4B version reads images as well as text. MIT licence.
You could use it to…
- Rerank twenty search results by which actually answer the question
- Grade an answer against a reference automatically
- Score a game move by comparing statements about the state
Can I use this?
You'll need A GPU and the transformers library · Setup needed
Worth knowing The creator reports it works zero-shot on everything shown, meaning without task-specific training, and that claim is theirs rather than an independent result. Turning a task into premise-and-hypothesis pairs is work you have to do yourself, and a wrong framing gives confident wrong answers. BuildTube has not run or verified this model.
Why it's here
Not trending this week
Last seen on the Hugging Face trending list on 25 September 2026. It stays on BuildTube so you can still find it.
- 636 people have liked it on Hugging Face.
- It was downloaded 0 times in the last 30 days.
Numbers checked on 6 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?