Module 07 (optimizers):
- Fixed bug where param.grad was set before optimizer creation
- Optimizer.__init__ resets param.grad to None for all parameters
- Moved all gradient assignments to occur AFTER optimizer creation
- Fixes GitHub issue #1131
Module 10 (tokenization):
- Made BPETokenizer.train() vocab_size parameter optional with default None
- Fixes test failure when calling train() without explicit vocab_size
All 20 inline tests pass (verified via tito dev test --inline --ci)
Monorepo restructuring for better build performance and separation of concerns:
- Create kits/ as standalone Quarto project with teal accent styling
- Add _brand/ directory with shared design tokens (tokens.scss)
- Move labs from book/quarto/contents/labs/ to kits/contents/
- Remove labs sections (Parts VII-XII) from book sidebar
- Add CI/CD workflows for kits (kits-publish-live.yml, kits-publish-dev.yml)
- Update netlify.toml redirects: /labs -> /kits/
- Add Volume I/II test sidebar config for future use
Deployment URLs:
- Textbook: mlsysbook.ai/
- Kits: mlsysbook.ai/kits/
- TinyTorch: mlsysbook.ai/tinytorch/
All three projects build independently and successfully.
Update workspace Peacock theme to use a periwinkle/lavender-blue color
scheme (#A5B4E5) that blends blue and purple tones for a softer visual
appearance in the activity bar, status bar, and title bar.
Removes redundant elements from the window title for a cleaner
and more focused display.
This improves readability and reduces visual clutter, enhancing the
user experience.
- Add 12 strategic references in Model Optimizations chapter
- Add 12 system-level references in Efficient AI chapter
- Add 12 hardware foundation references in AI Acceleration chapter
- Create tightly interconnected optimization ecosystem
- Connect model-level → system-level → hardware-level optimizations
- Link optimization techniques to deployment and sustainability contexts
- Establish hardware acceleration as foundation enabling all optimizations
- Fix reference errors and duplicate labels
Optimization cluster now forms cohesive technical ecosystem with clear
connections between algorithmic, system, and hardware efficiency strategies.
- Add workspace-specific Peacock color theme (#5d4e75 purple-blue)
- Custom window title with book emoji for easy project identification
- Light theme with purple activity/status bars for visual distinction
