slvDev
精选
esp32-ai
A 28.9 million parameter language model runs on an $8 ESP32-S3 microcontroller entirely on-device, generating simple stories at about 9.5 tokens per second.
AI 仓库情报摘要
FR-AI / ANALYSIS为什么值得关注
它提供了一种结构化、可复现的方式来测量和比较LLM质量，并支持语言、评分尺度和权重的灵活配置。
适合谁使用
- LLM开发者和工程师
- 评估模型输出的AI研究人员
- 构建智能体评估系统的产品团队
- 创建评分量表的QA专业人员
典型使用场景
- 按自定义标准对LLM回答进行评分
- 在基准测试中比较模型性能
- 为智能体工作流设置评估量表
- 为自动化测试创建可重复的评分模板
项目优势
- 支持多语言量表（英语、中文、双语）
- 提供可配置的评分尺度（1-5、0-2、通过/失败）与权重
- 鼓励评估的可复现性和可比性
- 通过简单的CLI命令安装（npx skills add）
使用前须知
- 完整工作流详情在外部SKILL.md中，README内容有限
- 范围仅限于量表设计，不包含评估执行
- 安装需依赖'skills'生态系统
README 快速开始
Eval Rubric
Design evaluation rubrics for LLM/agent outputs: criteria, scores, failure modes, examples.
中文简介： 你做可复现、可对比的模型输出评测表。 — 详见 README.zh.md.
What this skill does
- You build rubrics that make model quality measurable and comparable
- Scope: See SKILL.md vertical lock.
- Deliverable: See SKILL.md deliverable.
- Triggers: eval rubric, LLM eval, grading rubric, agent eval, 评测量表.
Install
After this package is on GitHub:
npx skills add / --skill eval-rubric
Local monorepo: point your agent at skills/eval-rubric/ (this folder).
Options
| Option | Values | Default |
|---|---|---|
lang | en · zh · bilingual | en |
scale | 1-5 · 0-2 · pass-fail | 1-5 |
weighting | equal · custom | equal |
Runtime output language follows lang (default en). Skill docs stay EN-primary.
When to use
| Use | Don't |
|---|---|
| (see bullets) | (see bullets) |
Use
- When you need
eval-rubricworkflow
Don't
- Out-of-scope tasks listed in SKILL.md
Deliverable
See SKILL.md deliverable.
Full workflow, hard rules, and QA: see SKILL.md.
Example
Monorepo sample (if present): ../../output/samples/eval-rubric-example-01.md
License
MIT — see LICENSE.
相关仓库与替代方案
根据分类、Topic 和编程语言匹配的相似项目。
jamesob
精选
local-llm
A comprehensive guide for building and configuring a high-end local machine to run state-of-the-art LLMs, with detailed hardware choices, BIOS tuning, and Docker-based model serving.
makecindy
精选
cindy
Cindy is an open-source AI agent that runs locally on your machine, integrates multiple AI harnesses and models, and provides memory, skills, and automation to perform real work in your projects and apps.