RISECamp Tutorial Final Feedback and Todo

[simon-mo]

Tutorial Todos

Due

• Draft by Friday so Dan can make some edits.
• Test possible collision with Integration (Hari)

Split into 3 notebooks (Simon)

• TOC, Summary, Intro
  • Link to part1 and part2
• Part 1
• Part 2

Poll (Simon)

• waiting on Joey to edit the poll
• Add it to the end of the tutorial (part 1, 2)

Less reading (Simon)

• Abstract paragraphs into optional paragraphs
  • Each section should only have 2-4 sentences: here's what you are going to do and why are you going to do it.
  • Other, hidden optional sub-section. Drop-down
• Add a new util.py file to make code shorter.
• Dockerfile:
  • Make detail description optional
  • Add description as comments in actual docker file.

New small section (Rehan)

• pytorch: add a cell after batching: "bring your url"
• at the end, ask them to check grafana to see update

Nit (Rehan)

• Make sure the links open in new tabs, ex the Grafana link. change markdown links-> html links in display(Markdown())
• Reset fixme
• Reset collapse cell / or just use drop-down
• Add stop_all in part 1 in the beginning
• make -j16 for Darknet
• at the end, comment out stop_all
• Matplotlib will warning in plotting code for legend missing handles. We will fix it.
• Part 1 text: Change ‘word count model’ -> ‘flowercat model’
• Replica image has 4 replicas although we are only doing 3 replicas
• In the overview section, change part 1 to fix minor typos:
  Part 1 provides a detailed explanation of Clipper, its architecture, and the python API. It uses the deployment of a scikit-learn model that classifies flowers based on their Sepal Length and Sepal Width as example.

[simon-mo]

Simon

• Simon enable docker network in a fork using docker campnet_clipper