
Cloudflare/clef-flash 模型
9B 多模态决策模型一次前向输出各问题选项概率。
为什么重要
为需要结构化决策输出的应用提供可直接调用的多模态模型。
阅读原文huggingface.co
AI 解读(根据原文生成)
Cloudflare/clef-flash 是一个 9B 多模态模型,用于把状态和类型化问题 schema 转换为决策。它可读取文本、JSON、图像或视频形式的状态,并在单次前向中为每个问题的每个允许选项返回概率,不做自由文本生成,也无需输出解析。模型基于 Qwen/Qwen3.5-9B 后训练,包含 joint schema head,并在 Hugging Face 提供。
要点(AI 据原文整理,数字均已与原文核对)
- It reads the state as text, JSON, images, or video, and returns a probability for every allowed option of every question in a single forward pass.
- There is no free-form text generation and no output parsing.
- The Clef-Flash API is fully compatible with Jev and SystemOne.
- Backbone: Qwen/Qwen3.5-9B with its vision encoder, stored as standard sharded safetensors.
- Tested with `torch` 2.11 and `transformers` 5.10.2 on a single H200.
- Image and video inputs also need `pillow`.