# bojieli/ai-infra-book

> Open book "Understanding AI Infra" by Bojie Li: quantitative derivation of LLM inference and training system design from hardware constraints and model architecture. Includes full text, PDF, calculation tools and experiments.

- Magnitude: 2.5 out of 10 — Steady
- Stars: 7,120 total · +16 stars measured 2026-10-29, 00:07–05:45 UTC, ≈ 55 by evening
- Star trust: star growth looks organic
- Category: Learning & lists · Language: Python · License: Apache-2.0 · Created: 2026-08-22 · Last push: 2026-10-01
- GitHub: https://github.com/bojieli/ai-infra-book · Homepage: https://bojieli.github.io/ai-infra-book/ · Page: https://gitnova.dev/en/r/bojieli/ai-infra-book

## Useful for

- Estimate Qwen3-8B resources at 8192-token prefill via calculations/calc.py
- Study chapter by chapter how batching and KV cache affect LLM inference
- Reproduce experiments from experiments/ for a chosen chapter

## Why it’s here

- Star-counter measurements on 2026-10-29 (UTC), 00:07–05:45: 7104 → 7120 stars (+16). This is the change over that interval.
- Estimated end-of-day forecast: about +55 stars, using observed gains and the previous day.
- Top new repositories in the last 4 months: #98.
- About 5 forks a day — people are taking the code.

## Star trust

Star growth looks organic.

Star-trust labels are heuristics based on the repository’s behavior, not a check of every stargazer.

## Numbers

- Forks: 567
- Issues and pull requests: 40
- Watchers: 20
- Average over the last week: 42 per day
- Usual pace: 65 per day
- Stars in the last hour (measured): 1
- Latest release: build-20261001-161920 (2026-10-01)

## Stars per day, last 30 days (oldest → newest, today is partial)

2026-09-30 … 2026-10-29: 43, 60, 45, 40, 39, 33, 41, 75, 81, 64, 32, 51, 80, 127, 64, 68, 47, 34, 49, 46, 56, 35, 43, 38, 27, 37, 47, 42, 47, 16

## Magnitude by day, last 14 days

10-16 2.9, 10-17 2.5, 10-18 2.2, 10-19 2.5, 10-20 2.6, 10-21 2.4, 10-22 2.5, 10-23 2.5, 10-24 2.3, 10-25 2.0, 10-26 2.4, 10-27 2.5, 10-28 2.6, 10-29 2.5

## Spotted in now

- Top new repositories in the last 4 months: #98

## More in this category

1. **smithjohn78390/tongjiang-lab** — 7.9 · Early signal · Learning & lists · HTML · +155 stars measured 2026-10-29, 00:02–05:41 UTC, ≈ 630 by evening
   Free interactive 3D science lessons for high school physics, chemistry, biology and geography: 55 lessons, each a single self-contained HTML file that works offline.
   Full card: https://gitnova.dev/en/r/smithjohn78390/tongjiang-lab.md
2. **amosblomqvist/learnharness** — 6.7 · Breakout · Learning & lists · TypeScript · +69 stars measured 2026-10-29, 00:02–05:41 UTC, ≈ 374 by evening
   A desktop app that uses AI agents to plan and teach lessons from your current understanding to a goal, tracking progress in a persistent knowledge graph.
   Full card: https://gitnova.dev/en/r/amosblomqvist/learnharness.md
3. **xiaolai/the-craft-of-selfteaching** — 6.0 · Breakout · Learning & lists · Jupyter Notebook · +164 stars measured 2026-10-29, 00:03–05:42 UTC, ≈ 548 by evening
   Li Xiaolai's book on self-teaching, using Python as a practical entry point to learning on your own. Written as Jupyter Notebooks with a Markdown version available.
   Full card: https://gitnova.dev/en/r/xiaolai/the-craft-of-selfteaching.md
4. **designfuchen/word-atlas** — 5.5 · Cooling · Learning & lists · TypeScript · +67 stars measured 2026-10-29, 00:02–05:42 UTC, ≈ 307 by evening
   A free, open-source PWA for memorizing 7,298 IELTS words, each with an original illustration, UK/US audio and example sentences, using FSRS spaced repetition with no login. The dataset is CC BY 4.0.
   Full card: https://gitnova.dev/en/r/designfuchen/word-atlas.md
5. **xiaoyang111222/cn-math-taxonomy** — 4.8 · Peaking · Learning & lists · Python · +163 stars measured 2026-10-29, 00:03–05:42 UTC, ≈ 494 by evening
   Knowledge graph for Chinese middle-school math (grades 7–9): 315 topic nodes with 616 prerequisite edges, mapping to the 2022 curriculum standards, plus a 1407-question diagnostic bank. An AI-generated draft not yet reviewed by teachers.
   Full card: https://gitnova.dev/en/r/xiaoyang111222/cn-math-taxonomy.md

---

Magnitude (0–10) measures how fast and how unusually interest in a repository is growing right now. It is not a quality score.
Days are UTC. “So far today” is a fact; “expected by the end of the day” is a forecast.
Summaries and use cases are written by an LLM (DeepSeek V4.1 Flash) from the README and may be inaccurate: verify specific claims (benchmarks, speed, hardware) in the repository itself.
Data as of 2026-10-29 05:52 UTC, updated every 30 minutes.