--- title: wordlink canonical_url: "https://www.modelscope.cn/models/OrionSheep/wordlink" md_url: "https://www.modelscope.cn/models/OrionSheep/wordlink.md" repository: OrionSheep/wordlink chinese_name: "WordLink 词链英语学习应用（代码镜像）" last_updated: 2026-10-04 downloads: 0 stars: 0 --- # wordlink > wordlink - OrionSheep 在 ModelScope 开源的模型。🌌 WordLink — Lexiverse 语宙 - **Repository**: OrionSheep/wordlink - **Downloads**: 0 - **Stars**: 0 - **Last updated**: 2026-10-04 Source: https://www.modelscope.cn/models/OrionSheep/wordlink ---
# 🌌 WordLink — Lexiverse 语宙 **AI-powered vocabulary learning. A fine-tuned 1.5B model that runs offline in your browser.** [![Next.js 16](https://img.shields.io/badge/Next.js-16-black)](https://nextjs.org) [![React 19](https://img.shields.io/badge/React-19-61dafb)](https://react.dev) [![Model](https://img.shields.io/badge/Model-Qwen2.5--1.5B-8b5cf6)](https://github.com/QwenLM/Qwen2.5) [![Training](https://img.shields.io/badge/Training-SFT_%2B_DPO-22c55e)](docs/模型训练详细报告.md) [![Weights](https://img.shields.io/badge/Weights-ModelScope-624aff)](https://modelscope.cn/models/OrionSheep/wordlink-lexiconstrain) *Stunning cognitive topology. Blazing retention. Built for SDG 4.* ### [🚀 Live Demo · wordlink.lifeplayertribe.com](https://wordlink.lifeplayertribe.com) Demo account: `demo@wordlink.test` / `Demo2026!` [Features](#-features) · [Benchmarks](#-measured-results) · [Architecture](#-how-it-works) · [Quick Start](#-quick-start) · [Weights](#-pretrained-weights-modelscope) · [中文简介](#-中文简介)
![WordLink 星图工作台](public/images/showcase/study.jpg) > ⚡ **No ads. No subscription. Works offline.** The vocabulary model runs *in your browser* via WebGPU — your learning data never leaves your device. --- ## ✨ Features
### 🪐 星图工作台 · Study Workspace Word library, word details and the **fission graph** in a three-column workspace — pick any word and its synonym network unfolds in a force-directed star map. ### 🧠 AI Quiz · 智能测验 Choice / spelling / recall modes driven by a memory-strength model. Every answer updates your forgetting curve in real time.
### 🌍 Contextual Reading · 语境精读 Graded articles with click-to-define on any word — reading *is* your review session. ### 🛂 XAI Learning Passport · 学习护照 An explainable AI narrative of everything you've learned: CEFR positioning, cognitive radar, growth trajectory.
More: **Ambient screensaver listening** (four-season soundscapes synthesized live in Web Audio) · **Cognitive Navigator** (shortest activation path to your target exam) · **My Libraries** (built-in + custom word books) · **中英双语界面** · dark mode.

Immersive mode Cognitive Navigator

Dashboard My Libraries

--- ## 📊 Measured Results All numbers are **real measurements** from our 90-question benchmark (greedy decoding) — target-word injection into CEFR-constrained stories, scored on real model outputs. Full report: [`outputs/eval/BENCHMARK_RESULTS.md`](outputs/eval/BENCHMARK_RESULTS.md). | Model | Target-word hit ↑ | CEFR violation ↓ | JSON compliance ↑ | FKGL | |---|---:|---:|---:|---:| | Qwen2.5-1.5B-Instruct (before fine-tuning) | 63.06% | 0.17% | 82.22% | 13.5 | | **+ SFT** (2,045 diverse stories) | **88.06%** | **0.01%** | **100%** | **9.8** | | **+ DPO** (real-failure negatives) | 87.78% | **0.01%** | **100%** | 9.9 | | Qwen2.5-7B-Instruct (zero-shot) | 39.72% | 0.07% | **0%** | — | | DeepSeek-V3 zero-shot (teacher) | 97.78% | 0.01% | 100% | 10.4 | | Teacher / Oracle upper bound | 98.33% | 0.22% | 100% | — | > 💡 Fine-tuning the small model: **+25pp on the same 1.5B**, matching the teacher's upper bound — while a *larger* model without fine-tuning collapses (39.72% hit, 0% valid JSON). Constraint-following is trained, not bought with parameters. --- ## 🏗 How It Works **Neuro-symbolic architecture** — algorithms handle scheduling, neural models handle generation: ``` ┌───────────────────────────┐ ┌───────────────────────────────┐ │ Symbolic Engine (确定性) │ │ Neural Engine (生成) │ │ · FSRS-6 forgetting curve │ │ · Edge 1.5B (browser, WebGPU) │ │ · Levenshtein scoring │ ⇄ │ · CEFR-constrained stories │ │ · RME word matching │ │ · Real-failure DPO negatives │ └───────────────────────────┘ └───────────────────────────────┘ │ free · explainable · offline │ infinite · personal ▼ ▼ "算法管调度 —— 每毫秒、可解释" "AI 管生成 —— 把复习词写进今天的故事" ``` The loop: **FSRS computes what you should review tomorrow → the fine-tuned model writes it into today's story → reading the story surfaces new words → your memory model updates.** Training pipeline: `Data factory (generate + 5-gate filter) → SFT → DPO (real-failure negatives) → Auto-benchmark → 4-bit edge export` — full report in [`docs/模型训练详细报告.md`](docs/模型训练详细报告.md). --- ## 🚀 Quick Start ```bash # 1. Prerequisites: Node.js ≥ 20, Docker git clone https://github.com/orionsheep/wordlink.git cd wordlink npm install # 2. Start PostgreSQL (business data) docker compose up -d # 3. Configure environment (see docs/部署与运行指南.md for every variable) # DATABASE_URL is required; auth / AI keys optional # 4. Run npm run dev # → http://localhost:3000 ``` > Demo experience: sign in with `demo@wordlink.test / Demo2026!` (pre-loaded with 180 days of learning data), or browse public pages without an account. Full deployment guide: [`docs/部署与运行指南.md`](docs/部署与运行指南.md). ## 🧠 Train Your Own The complete training pipeline is in [`model_training/`](model_training/): ```bash python model_training/08_generate_diverse_corpus.py # data factory (LLM teacher + 5-gate filter) python model_training/02_train_sft.py # SFT python model_training/03_train_dpo.py # DPO with real-failure negatives python model_training/eval_real.py --backend local # benchmark ``` Full report: [`docs/模型训练详细报告.md`](docs/模型训练详细报告.md) ### 📦 Pretrained weights (ModelScope) | Repo | Content | |---|---| | [OrionSheep/wordlink-lexiconstrain](https://modelscope.cn/models/OrionSheep/wordlink-lexiconstrain) | **Final model** — SFT+DPO merged weights (5.8 GB) + DPO LoRA adapter + eval artifacts | | [OrionSheep/wordlink-edge](https://modelscope.cn/models/OrionSheep/wordlink-edge) | SFT v2 merged weights (5.8 GB) + LoRA adapters (v2 / v1 legacy) + eval artifacts | | [OrionSheep/wordlink](https://modelscope.cn/models/OrionSheep/wordlink) | This repository's ModelScope mirror | --- ## 🇨🇳 中文简介 **Lexiverse 语宙** 是一个 AI 驱动的英语词汇学习平台：自己微调的 1.5B 模型直接运行在浏览器中（WebGPU 4-bit 量化，~950MB 首次下载后永久离线），把你的到期复习词实时写进今天阅读的故事里。遗忘调度由可解释的 FSRS-6 数学引擎完成，推理零边际成本，学生数据永不出浏览器。核心实测：微调后生词注入命中 88.06%、JSON 合规 100%、CEFR 越界 0.01%。 ## 📁 Repository Map ``` ├── src/app/ # Next.js App Router pages & API routes ├── src/components/ # UI components (welcome landing, study workspace, ambient…) ├── model_training/ # SFT / DPO / corpus factory / benchmark scripts ├── data/ # word libraries & training corpora (jsonl) ├── outputs/eval/ # benchmark metrics & per-question predictions ├── docs/ # training report, deployment guide, media specs └── public/ # images & demo videos ``` ---
**Built for [UNU Macau · AI for SDGs 2026](https://unu.edu/macau)** — AI for Education (SDG 4) *Stunning cognitive topology. Blazing retention.*