github-starred/cs249r_book
2
0
Fork 0
mirror of https://github.com/harvard-edge/cs249r_book.git synced 2026-05-06 17:49:07 -05:00
Code Issues 85 Packages Projects Releases 24 Wiki Activity
14,103 Commits 18 Branches 31 Tags
dev
Commit Graph

6 Commits

Author SHA1 Message Date
Vijay Janapa Reddi
1eb30f5f86 fix(mlsysim): harden release QA and paper artifacts 
Align the MLSys·im code, docs, paper, website, workflows, and lab wheel for the 0.1.1 release. This also fixes runtime/API issues found during release review and prepares the paper PDF plus archive package.
 2026-04-25 10:06:01 -04:00
Vijay Janapa Reddi
1b66680aca docs(tutorial): simulation round 2 — NPS improved from -10/+5 to +10/+20 
Eta explanation, AllReduce numbers-first, and compression fleet economics
all validated by re-simulated personas. 2 of 5 issues fully resolved,
2 partially resolved (timing still tight), 1 unaddressed (TinyML depth).
Multi-vendor Roofline is the single biggest NPS driver.
 2026-04-02 07:34:25 -04:00
Vijay Janapa Reddi
b7bf7a4ce5 docs(tutorial): simulated Q&A — 8 tough questions + 3 hallway conversations 
Answer quality ranges 5-9/10. Weakest: MoE support (6/10), diffusion
models (5/10), TinyML depth (6/10) — all honest v0.2.0 gaps.
Strongest: spreadsheet comparison (7/10), inverse Roofline value (9/10),
CPI analogy for efficiency (9/10).
3 hallway conversations simulate real adoption decision dynamics.
 2026-04-02 07:30:04 -04:00
Vijay Janapa Reddi
5c7e8e8fc5 docs(tutorial): simulation round 1 — 5 attendee personas, NPS -10 to +5 
Key findings: efficiency never explained (fix needed), compression too
rushed (15 min), AllReduce needs numbers-before-formulas, TinyML feels
tangential. Inverse Roofline is the surprise hit. AMD engineer caught
a wrong MI300X bandwidth number.
 2026-04-02 07:28:17 -04:00
Vijay Janapa Reddi
e24a5a2d9e feat(tutorial): pilot study protocol for pre/post quiz research data 
Within-subjects design (N≥30), paired t-test analysis plan, IRB
considerations, expected effect size (d=0.8), and timeline for
running at ISCA 2026. Produces publishable data for SIGCSE/L@S.
 2026-04-01 19:14:49 -04:00
Vijay Janapa Reddi
b61f632b71 feat(tutorial): pre/post assessment quiz — 10 questions mapping to 6 understanding goals 
Designed for dual use: tutorial engagement + publishable research data.
Tests transfer (apply framework to unfamiliar systems), not recall.
Includes distractor analysis and scoring rubric.
 2026-04-01 19:13:27 -04:00