mirror of
https://github.com/MLSysBook/TinyTorch.git
synced 2026-03-25 20:59:40 -05:00
11f1771f17
Addressed 3 critical issues identified by education reviewer: 1. Standardized Module 07 terminology: - M03: Changed 'training loop' to 'end-to-end training loop' - Now consistent across all milestones (M01/M02/M03/M04) 2. Added quantitative loss criteria to M03: - TinyDigits: Loss < 0.5 (gives students measurable target) - MNIST: Loss < 0.2 (realistic threshold for convergence) - Fixed parameter count: ~2K → ~2.4K (accurate calculation) 3. Clarified M06 foundational dependencies: - Added note explaining Modules 01-13 are prerequisites - Makes clear the table shows ADDITIONAL optimization modules - Prevents confusion about complete dependency chain These fixes bring milestone READMEs to production-ready quality. Education reviewer grade: A- → A (after these fixes).