StartLuxLabs/StartLux-Decision

选用指南

AI 根据 README 整理 · 2026/10/21 13:26:35

适用场景

适用于需要根据文本、JSON 或图像等状态，对一组类型化问题（如选择、是/否判断、评分）进行概率性决策的场景，例如客服工单分类、游戏策略决策等。

查看 README 依据

StartLux-Decision is a family of open decision models. Given a state (text, JSON or images) and a set of typed questions, a model returns a probability for every option of every question: a choice among listed options (`choice`), a yes or no judgement (`noul`), or a level on a scale (`score`). Instant models answer in a single forward pass with bidirectional attention over the prompt. AutoThink models answer in one pass when the top probability reaches a gate, 0.6 by default, and otherwise generate a short reasoning trace and read the answer again.

Download a model and start one vLLM server per GPU with the decision server in front of them: ```bash hf download startlux-models/StartLux-Decision-v1.1-4B-AutoThink --local-dir models/StartLux-Decision-v1.1-4B-AutoThink startlux-decide serve --model models/StartLux-Decision-v1.1-4B-AutoThink --launch-vllm --gpus 0,1,2,3 --port 8090 ``` Ask a question: ```bash curl -s localhost:8090/v1/systemone -H 'Content-Type: application/json' -d '{ "state": {"ticket": "I was charged twice for order #4411 and the app still shows it as unpaid."}, "questions": {

使用方式

通过命令行工具 `startlux-decide` 使用，可以启动 vLLM 服务器并运行决策服务，通过 HTTP API 发送请求。

查看 README 依据

Download a model and start one vLLM server per GPU with the decision server in front of them: ```bash hf download startlux-models/StartLux-Decision-v1.1-4B-AutoThink --local-dir models/StartLux-Decision-v1.1-4B-AutoThink startlux-decide serve --model models/StartLux-Decision-v1.1-4B-AutoThink --launch-vllm --gpus 0,1,2,3 --port 8090 ``` Ask a question: ```bash curl -s localhost:8090/v1/systemone -H 'Content-Type: application/json' -d '{ "state": {"ticket": "I was charged twice for order #4411 and the app still shows it as unpaid."}, "questions": {

部署要求

需要 Python 3.9 或更高版本，并安装到与 vLLM 相同的 Python 环境中。运行模型需要 GPU（例如 H200），但也可以使用免费的托管服务 bamboonode.dev（需要 Bamboo API 密钥）。

查看 README 依据

<p align="center"> <a href="https://arxiv.org/abs/2610.17309"><img src="https://img.shields.io/badge/arXiv-2610.17309-8A5CD6" alt="arXiv"></a> <a href="LICENSE"><img src="https://img.shields.io/badge/code-Apache--2.0-8A5CD6" alt="License"></a> <a href="pyproject.toml"><img src="https://img.shields.io/badge/python-3.9%2B-8A5CD6" alt="Python"></a> <a href="#engines"><img src="https://img.shields.io/badge/backends-vLLM%20%7C%20llama.cpp%20%7C%20MLX-8A5CD6" alt="Backends"></a> </p>

To try a model without a GPU, use [bamboonode.dev](https://bamboonode.dev/). It hosts the 27B AutoThink model with the same request format (`POST /v1/decisions`), free and without usage limits, and needs only a Bamboo API key. To run the models yourself, install the package into the same Python environment as vLLM. It carries the vLLM plugin that Instant models need. ```bash git clone https://github.com/StartLuxLabs/StartLux-Decision.git cd StartLux-Decision pip install -e '.[vllm]' # vLLM 0.31.0; the published numbers were measured on 0.19.1, which is also supported ```

已知限制

AutoThink 模型在不确定时会生成思考过程并再次读取答案，这比单次前向传播的 Instant 模型需要更长的时间。

查看 README 依据

StartLux-Decision is a family of open decision models. Given a state (text, JSON or images) and a set of typed questions, a model returns a probability for every option of every question: a choice among listed options (`choice`), a yes or no judgement (`noul`), or a level on a scale (`score`). Instant models answer in a single forward pass with bidirectional attention over the prompt. AutoThink models answer in one pass when the top probability reaches a gate, 0.6 by default, and otherwise generate a short reasoning trace and read the answer again.

contains the full evaluation. The engine also serves the [v1 models](https://huggingface.co/collections/startlux-models/startlux-decision-v1-6abba92b301b573fa154d493) (0.8B to 27B and 35B-A3B), which answer in one pass. On one H200 with vLLM and one request in flight, Instant models answer a request in 25 to 52 ms at the median, from 0.8B to 27B. Through this server, an AutoThink model needs one read-out of 38 to 59 ms when it is sure. When it is unsure it writes a thought and reads the answer a second time, and the thought takes longer the more it writes. ## Quick start