# CalvinXKY/InfraTech

分享AI Infra知识&代码练习：PyTorch、vLLM/SGLang、slime/vime框架入门⚡️、性能加速🚀、大模型基础🧠、AI软硬件🔧等

Repository: https://github.com/CalvinXKY/InfraTech
Canonical: https://ross.abutalabs.com/products/infratech
Language: Jupyter Notebook
License Family: other
Last push: 2026-08-07T06:59:36+00:00

## Health v2 (maintenance only)
Score: 59/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 96, release rhythm 35, longevity 20
- inputs: {"age_days": 292, "days_push": 26, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3683, forks 350 (observed 2026-08-28T04:08:13.753034+00:00)

## What it is
A collection of Jupyter notebook tutorials and exercises covering AI infrastructure topics including PyTorch, vLLM/SGLang inference frameworks, performance acceleration, and LLM fundamentals. Content is presented as runnable Python notebooks with accompanying explanatory articles, primarily in Chinese.

## Use cases
- learn how LLM inference engines like vLLM and SGLang work
- understand RoPE and MLA attention computation with worked examples
- study distributed inference parallelism strategies like DP, TP, PP, EP
- practice chunked prefill and flash decoding implementations
- learn speculative decoding and LLM sampling techniques
- understand collective communication for distributed training and inference
- calculate MFU and attention FLOPs with prefix caching

## When to choose
- you want hands-on notebook exercises for AI infra and LLM inference internals
- you are an engineer preparing to work on inference optimization or GPU performance
- you prefer Chinese-language explanations paired with runnable code
- you need to understand attention variants like MLA and RoPE at a low level

## When to avoid
- you need a production inference framework rather than learning material
- you want English-only documentation
- you are a complete beginner to deep learning and Python
- you need a formally licensed, citable library for commercial reuse

## Facets
- artifact type: learning-resource
- maturity: active
- function: machine-learning, llm-inference, llm-training, gpu-computing, developer-tools
- domain: artificial-intelligence, large-language-models, deep-learning, machine-learning, gpu-computing, tutorials
- platform: python, cross-platform
- tags: jupyter-notebooks, ai-infrastructure, vllm, sglang, pytorch, inference-optimization, distributed-training, attention-mechanisms, speculative-decoding, chinese-language, gpu

## Member repositories
- CalvinXKY/InfraTech (main) score 59

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:13.753034+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 2026-08-29T18:31:11.715009+00:00, confidence not recorded.
  - readme: https://github.com/CalvinXKY/InfraTech (fetched 2026-08-28T04:08:13.753034+00:00, sha 9edc4be5eaea)
- Data as of 2026-08-30T08:39:29.467469+00:00.