Akshar:面向尼泊尔学生的 Gemma 4 学习平台
sagarxettri000/Akshar
Team Nepluro 正在开发学习平台 Akshar,帮助尼泊尔学生备考 NEB Grade 11–12 与 CEE/IOE 入学考试,提供尼泊尔语/英语讲解、带来源引用的课程答案和练习反馈。平台以 Gemma 4 承担学习流程中的实际工作,并针对低带宽、小屏设备和有限算力设计。项目处于 hackathon 开发阶段,仓库目前仅含项目文档与 AI-agent 指南,应用代码正在逐步加入。
A learning platform for Nepal — built by Team Nepluro.
Akshar helps Nepali students understand concepts, practise exam-style questions, and prepare with confidence for NEB Grade 11–12 and CEE/IOE entrance examinations. It pairs clear Nepali/English explanations with material students can inspect, and uses Gemma 4 as a genuine part of the learning experience.
Status: Active hackathon development. This repository currently holds the project documentation and AI-agent guidelines; application code is being added incrementally.
The problem
Nepali learners preparing for NEB Grade 11–12 and entrance exams (CEE/IOE) often have to rely on generic answers or English-only resources that do not match their curriculum or language. Access to trustworthy, level-appropriate support is uneven, especially on modest devices and connections.
Akshar is being built to close that gap.
What makes Akshar different
Akshar is built around one trustworthy loop: learn → understand → practise.
- Choose a track and topic — NEB Grade 11/12, CEE, or IOE.
- Ask a question or open a reviewed practice question.
- Get a clear, appropriately leveled explanation in Nepali or English, with source references when curriculum material is used.
- Try a related question or a short knowledge check.
- Receive encouraging, specific feedback and, where supported, track progress.
Design principles:
- Grounded, honest answers. Curriculum answers cite the material actually shown, and clearly label AI-generated explanations and practice questions.
- Nepali-first accessibility. Native Nepali/English support, correct Devanagari and mathematical notation, and readable text on small screens.
- Low-bandwidth by design. Built for modest devices and connections.
- A real AI path. Gemma 4 performs meaningful work in the learner flow — not a label attached to a canned answer.
Team — Nepluro
| Member | Focus |
|---|---|
| Sagar Katwal | Learner experience / frontend |
| Dipson Basnet | Gemma 4 / AI pipeline |
| Dhiraj Shrestha | Learning content, practice & validation |
Repository contents
| File | Purpose |
|---|---|
AGENTS.md |
Canonical project instructions for AI coding agents |
GEMINI.md |
Gemini CLI entry point that imports AGENTS.md |
FIRST_PROMPT.md |
Read-only repository audit prompt |
TEAM_PLAYBOOK.md |
Team work split, MVP scope, and demo plan |
LICENSE |
MIT License |
License
Released under the MIT License. See LICENSE for details.
来源:GitHub · github.com