How bojieli/ai-infra-book works
This repository is an AI infrastructure textbook, accompanied by resource-calculation tools and chapter-organized experiments. The main reader workflow is to study the chapters, use calculations to test resource estimates, and consult experiments for measured or reproducible examples. The README documents multilingual editions and research, case-study, and reference collections. Source excerpts are sparse and mostly concern external framework snapshots, so internal links among these collections are omitted rather than inferred.
Main entry points: calc.py Calculation CLI · manuscripts Twelve-chapter manuscript · calculations Resource calculations · training_pipeline_schedule.py Training schedule model
Components 11
Book content
- Twelve-chapter manuscript
manuscripts
Companion tools
- Resource calculations
calculations - Calculation CLI
calculations/calc.py - Training schedule model
calculations/src/infra_calc/topics/training_pipeline_schedule.py
Supporting evidence
- Chapter experiments
experiments - System case studies
case-studies - Topic research
research - Reference library
references
Language editions
- English edition
book-en - Traditional Chinese edition
book-zh-tw
Other components
- Engineer or student
Connections 10
- Engineer or student → Twelve-chapter manuscript: reads chapters
- Engineer or student → Calculation CLI: runs estimates
- Calculation CLI → Resource calculations: uses calculations
- Engineer or student → Training schedule model: explores schedules
- Engineer or student → Chapter experiments: consults experiments
- Engineer or student → System case studies: studies cases
- Engineer or student → Topic research: consults research
- Engineer or student → Reference library: checks sources
- Engineer or student → English edition: reads translation
- Engineer or student → Traditional Chinese edition: reads translation