🔬 官方開源與研究
Open-source projects and research backed by AI Singapore and the Singapore government.
ℹ️ This page covers AI Singapore and government-backed open-source projects and research. For broader academic and corporate contributions, see the Community Open Source page.
🦁 SEA-LION Model Ecosystem
56
Total models
27,988
Total downloads
278
Total likes
5,034
Top model downloads
| Version | Models | Downloads | Likes | Period |
|---|---|---|---|---|
| v4 | 13 | 15,267 | 50 | 2025-2026 |
| v3.5 | 6 | 2,853 | 15 | 2025 |
| v3 | 11 | 5,797 | 43 | 2024-2025 |
| v2 | 4 | 1,428 | 26 | 2024 |
| v1 | 6 | 2,157 | 104 | 2023-2024 |
💡 作為對比：Meta Llama 3 單模型下載量達千萬級，Mistral 和 Qwen 系列在百萬級。SEA-LION 的定位是東南亞多語言市場，使用者群體較小但增長趨勢明顯（v4 佔總下載量 55%）。
🤗 View all models on HuggingFace →🛡️ SEA-Guard Safety Models
4 models
371 downloads
2026 年 2 月釋出，尚處於早期階段
Gemma-SEA-Guard-12B-2602 Image-Text-to-Text
Qwen-SEA-Guard-8B-2602 Image-Text-to-Text
Qwen-SEA-Guard-4B-2602 Image-Text-to-Text
Llama-SEA-Guard-8B-2602 Text Generation
✅ AI Verify Governance Framework
AI Verify
全球首個 AI 治理測試框架 · Open-sourced 2023-06
- ✓ 測試 AI 系統是否符合 11 項國際公認治理原則
- ✓ 覆蓋傳統 AI + 生成式 AI（2025 年 5 月新增）
- ✓ 對齊 EU、OECD、US 治理框架
Partners
AWSDBS BankGoogleMetaMicrosoftSingapore AirlinesNCS/SingtelStandard Chartered
📦 Official Project Index
Each card opens a project profile that can keep growing over time.
AI Verify Foundation / IMDA
AI Verify
把 AI 治理原則轉成可執行測試的開源框架
TypeScript 開源運營中 GitHub 37 Stars
View profile
AI Singapore
SEA-LION
面向東南亞語言和文化語境的大模型家族
Python / Models 持續迭代 GitHub 400 Stars
View profile
AI Singapore
SEA-Guard
面向東南亞文化語境的安全分類與護欄模型
Models 早期釋出
View profile
AI Singapore
TagUI
免費 RPA 網頁/桌面自動化工具
JavaScript 社群維護 GitHub 6,282 Stars
View profile
AI Singapore
PeekingDuck
模組化計算機視覺推理框架
Python 維護放緩 GitHub 177 Stars
View profile
AI Singapore
SGNLP
新加坡 NLP 研究社群模型
Python 維護放緩 GitHub 37 Stars
View profile
AI Singapore / NUS / NTU
Speech Lab
Singlish 語音轉文字
Speech AI 產品化服務
View profile
AI Singapore
Synergos
隱私保護聯邦學習框架
Python 早期開源 GitHub 2 Stars
View profile
📄 Research Papers
SEA-Guard: Culturally Grounded Multilingual Safeguard for Southeast Asia
SEA-SafeguardBench: Evaluating AI Safety in SEA Languages and Cultures
SEA-LION: Southeast Asian Languages in One Network
SEA-HELM: Southeast Asian Holistic Evaluation of Language Models
本頁統計資料來源：HuggingFace API、GitHub API，由 新加坡 AI 觀察獨立採集，非官方資料。統計採集：2026-05-03；專案檔案更新：2026-05-04。