# ForceInjection/AI-fundamentals

AI 基础知识 - GPU 架构、CUDA 编程、大模型基础及AI Agent 相关知识。

Repository: https://github.com/ForceInjection/AI-fundamentals
Canonical: https://ross.abutalabs.com/products/ai-fundamentals
Homepage: https://forceinjection.github.io
Language: HTML
License: Apache-2.0
License Family: permissive
Topics: ai-infra, ai-agent, cuda
Last push: 2026-08-26T15:27:21+00:00

## Health v2 (maintenance only)
Score: 69/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 41, longevity 50
- inputs: {"age_days": 709, "days_push": 7, "days_rel": 238, "gap_med": 131, "n_releases_24m": 2}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2376, forks 364 (observed 2026-08-28T04:06:42.057748+00:00)

## What it is
A comprehensive Chinese-language learning resource collection covering the full AI infrastructure stack, from GPU/TPU hardware architecture, CUDA programming, and interconnects (PCIe, NVLink, GPUDirect) to LLM training, inference optimization, RAG, and AI Agent systems. It includes 15 content modules with 800+ articles and courses, plus hands-on guides for Kubernetes-based AI platforms, cluster operations, and tools like vLLM, Ollama, and DeepSeek.

## Use cases
- learn GPU architecture and CUDA programming from scratch
- understand how NVLink, PCIe, and GPUDirect interconnects work
- set up and operate a large-scale AI training cluster with InfiniBand and NCCL
- build a Kubernetes-based AI platform with GPU virtualization and scheduling
- learn LLM fine-tuning techniques like SFT, LoRA, and QLoRA
- design RAG systems and GraphRAG pipelines
- study AI agent design patterns, multi-agent collaboration, and MCP protocol
- optimize LLM inference with vLLM and KV cache compression

## When to choose
- you want a systematic, end-to-end learning path for AI infrastructure from hardware to applications
- you are an AI engineer or architect needing deep dives into GPU internals, cluster ops, and cloud-native AI tooling
- you prefer practical, engineer-written guides with real deployment and tuning experience
- you need coverage of both training (CUDA, distributed training) and inference (vLLM, KV cache) topics in one place

## When to avoid
- you need runnable software or code libraries rather than documentation and tutorials
- you require English-language content only, as the material is primarily in Chinese
- you are looking for a beginner-only intro to AI/ML without infrastructure depth
- you need a structured course with certification rather than a self-guided article collection

## Facets
- artifact type: learning-resource
- maturity: active
- function: gpu-computing, llm-inference, llm-training, rag, agent-framework, mcp, machine-learning, monitoring, container-orchestration, documentation
- domain: artificial-intelligence, gpu-computing, large-language-models, developer-tools, tutorials, infrastructure-as-code, microservices
- platform: python, cpp
- tags: ai-infrastructure, cuda-programming, gpu-architecture, nvlink, infiniband, nccl, vllm, kv-cache, lora-fine-tuning, graphrag, multi-agent-systems, hami, gpu-virtualization, deepseek, ollama, chinese-language-content, ai-agents, retrieval-augmented-generation, containers, gpu, linux, kubernetes, docker, web-server

## Member repositories
- ForceInjection/AI-fundamentals (main) score 69

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:42.057748+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:34:58.567515+00:00, confidence not recorded.
  - readme: https://github.com/ForceInjection/AI-fundamentals (fetched 2026-08-28T04:06:42.057748+00:00, sha fc840b74314c)
  - homepage: https://forceinjection.github.io (fetched 2026-08-29T10:15:50.261719+00:00, sha 5c85608a244f)
- Data as of 2026-08-30T08:39:29.467469+00:00.