mirror of
https://github.com/MLSysBook/TinyTorch.git
synced 2026-04-29 13:12:31 -05:00
391b0a8999
- Eliminate redundancy across rule files - Keep all rules under 500 lines (largest is 187 lines) - Create focused, actionable rules scoped to specific concerns - Extract 'Real Data, Real Systems' principles to dedicated file - Streamline testing patterns to core requirements only - Fix corrupted development-workflow.mdc file - Remove duplicate content between files - Add comprehensive rules overview documentation Rule organization: - Core Context: ml-systems-course-context, tinytorch-project-structure, user-preferences - Development Workflow: development-workflow, git-workflow, cli-patterns - Module Development: module-development-best-practices, nbdev-educational-pattern, real-data-principles - Testing: testing-patterns Each rule is now focused, actionable, and composable without redundancy.
description
| description |
|---|
| Overview of TinyTorch's focused, actionable rule set - each rule under 500 lines and scoped to specific concerns. |
TinyTorch Rules Overview
This directory contains focused, actionable rules for TinyTorch development. Each rule is scoped to specific concerns and kept under 500 lines for clarity.
Rule Organization
Core Context & Structure
- ml-systems-course-context.mdc (64 lines) - Course philosophy and learning objectives
- tinytorch-project-structure.mdc (62 lines) - Dual-structure architecture guide
- user-preferences.mdc (47 lines) - User preferences and conventions
Development Workflow
- development-workflow.mdc (93 lines) - Complete tito CLI workflow
- git-workflow.mdc (157 lines) - Git workflow for incremental commits
- cli-patterns.mdc (174 lines) - CLI development patterns
Module Development
- module-development-best-practices.mdc (158 lines) - Core module development principles
- nbdev-educational-pattern.mdc (187 lines) - NBDev patterns and directives
- real-data-principles.mdc (112 lines) - "Real Data, Real Systems" requirements
Testing
- testing-patterns.mdc (115 lines) - Testing standards with pytest and real data
Key Design Principles
1. Focused Scope
Each rule addresses a specific concern:
- Context: Course philosophy and project structure
- Workflow: Development and Git processes
- Implementation: Module development and NBDev patterns
- Quality: Testing and real data requirements
2. No Redundancy
- Testing content consolidated in
testing-patterns.mdc - Real data principles extracted to
real-data-principles.mdc - NBDev patterns focused on directives and structure
- Module development focused on core principles
3. Actionable Guidance
- Concrete examples with code snippets
- Clear DO/DON'T patterns
- Specific file references and commands
- Checklists for quality standards
4. Composable Rules
Rules reference each other without duplication:
development-workflow.mdcreferencesgit-workflow.mdcmodule-development-best-practices.mdcreferencesreal-data-principles.mdctesting-patterns.mdcstands alone with focused testing guidance
Rule Usage
Each rule is designed to be:
- Self-contained: Can be read independently
- Actionable: Provides specific guidance and examples
- Focused: Addresses one main concern
- Concise: Under 500 lines for easy consumption
- Referenced: Can be cited in development discussions
Quality Standards Met
✅ Focused: Each rule has a single, clear purpose ✅ Actionable: Concrete examples and guidance ✅ Scoped: No rule exceeds 500 lines ✅ Non-redundant: No duplicate content across files ✅ Composable: Rules reference each other cleanly ✅ Maintainable: Easy to update individual concerns