A bilingual, multi-tier, community-driven knowledge base for Large Language Model (LLM) principles, papers, and interactive demos. Designed to take learners from intuition to implementation to research.
- Foundations — Tokenization, Attention, Transformer, Sampling & Decoding, Positional Encoding, Embeddings, Emergent Abilities
- Training — Pretraining & Scaling Laws, Fine-Tuning & Alignment (SFT, RLHF, DPO)
- Inference — KV Cache & Quantization, Efficient Attention, Long Context
- Applications — RAG & Retrieval Augmentation, Agents & Tool Use, Prompt Engineering, Evaluation & Benchmarks, Safety & Security
- Three reading tiers: Intuitive mental models, Engineering trade-offs, Research papers and open questions
- Bilingual: Full Chinese and English content
- Interactive demos: Tokenizer visualization, attention heatmap, sampling playground
- Paper-driven: 80+ curated papers with bilingual TLDRs; key claims trace back to paper entries
- Community curated: Contributions welcome via issues and pull requests
Requirements: Node 20 and pnpm 9.
pnpm install
pnpm dev # localhost:4321
pnpm build # build to dist/
pnpm test # run tests
pnpm lint:content # validate article frontmatter and paper refsThe site is deployed to Cloudflare Pages. See .github/workflows/deploy.yml for the CI/CD configuration.
We welcome corrections, translations, new articles, demo proposals, and paper summaries. Please read CONTRIBUTING.md and CONTRIBUTING.zh-CN.md before submitting.
This project is licensed under the MIT License.
llm large-language-models transformer attention rag agents prompt-engineering machine-learning nlp papers bilingual education astro starlight