github-starred/TinyTorch
2
0
Fork 0
mirror of https://github.com/MLSysBook/TinyTorch.git synced 2026-03-11 00:17:28 -05:00
Code Issues 7 Packages Projects Releases Wiki Activity
1,285 Commits 13 Branches 0 Tags
debugging
Commit Graph

3 Commits

Author SHA1 Message Date
Vijay Janapa Reddi
5024c29ad5 Improve module implementations: code quality and functionality updates 
- Enhance tensor operations and autograd functionality
- Improve activation functions and layer implementations
- Refine optimizer and training code
- Update spatial operations and transformer components
- Clean up profiling, quantization, and compression modules
- Streamline benchmarking and acceleration code
 2025-11-13 10:42:49 -05:00
Vijay Janapa Reddi
ae33298805 Fix NBGrader metadata for Modules 15 and 16 
Module 15 (Quantization):
- Added locked=true to test_module cell (line 1523)
- Added NBGrader metadata to systems-thinking markdown cell (line 1751)
- Added schema_version: 3 to both cells

Module 16 (Compression):
- Added NBGrader metadata to 6 solution cells:
  * measure-sparsity (line 380)
  * magnitude-prune (line 511)
  * structured-prune (line 675)
  * low-rank-approx (line 843)
  * distillation (line 1013)
  * compress-model-comprehensive (line 1234)
- Added NBGrader metadata to 6 test cells:
  * test-measure-sparsity (line 427) - 5 points
  * test-magnitude-prune (line 567) - 10 points
  * test-structured-prune (line 733) - 10 points
  * test-low-rank (line 888) - 10 points
  * test-distillation (line 1133) - 15 points
  * test-compression-integration (line 1300) - 20 points
- Total: 70 points for Module 16

Result:
- Module 15: 0 P0-BLOCKER, 0 P1-IMPORTANT (was 1 P0 + 1 P1)
- Module 16: 0 P0-BLOCKER, 0 P1-IMPORTANT (was 12 P0)
- Both modules now production-ready for NBGrader deployment(https://claude.com/claude-code)
 2025-11-11 14:50:37 -05:00
Vijay Janapa Reddi
a71e0eded5 Reorder modules for better pedagogical flow 
Moved memoization (KV-cache) after compression to align with optimization tier milestones.

Changes:
- Module 15: Quantization (was 16)
- Module 16: Compression (was 17)
- Module 17: Memoization (was 15)

Pedagogical Rationale:
This creates clear alignment with the optimization milestone structure:
  - M06 (Profiling): Module 14
  - M07 (Compression): Modules 15-16 (Quantization + Compression)
  - M08 (Acceleration): Modules 17-18 (Memoization/KV-cache + Acceleration)

Before: Students learned KV-cache before understanding why models are slow
After: Students profile → compress → then optimize with KV-cache

Updated milestone reference in profile_kv_cache.py: Module 15 → Module 17
 2025-11-10 19:29:10 -05:00