qna,

[Question] Tracking status of of learning

Follow Jun 05, 2020 · 2 mins read
[Question] Tracking status of of learning
Share this

Most contents of this QnA series are based on my own trouble, curiosity and sometimes for reminding.</br>

Reading list / Questions

Open Source

  • The (written) unwritten guide to pull requests July 25, 2019 - https://www.atlassian.com/blog/git/written-unwritten-guide-pull-requests (and this blog has lots of good articles)
  • Fastai contribution guide
  • Fastai test guide 1

Modern electronic circuit

  • What is CUDA? 9.0, 10.0, 10.1 what differs? why torch conflicts when using not corresponded version?

  • How to evaluate TPU spec with GPU? what criterion?

  • How colab is limiting the GPU for colab pro user - here / so suppose I’m implementing a model and should I wait until virtually use GPU? / How awful the resource is if I use GPU all the time? 2

  • High Performance Computing (Moore’s Law / Dennard Scaling)

Container (a.k.a. kubernetes/docker .. )

  • Julia Evan’s article of container - https://jvns.ca/blog/2016/09/15/whats-up-with-containers-docker-and-rkt/

Github

  • why clone using https (rather than ssh) is recommended?
  • GIt Pro book
    • Why there is no changing remote from https to https / ssh <-> ssh? reference git pro document

Distributed / Parallel Computing

As a Researcher

Pytorch

  • what’s difference between DataLoader / DataLoaders?

    Data format

  • Pickle: Usually saved with numpy format. is it required?
  • ZIP, GZIP, etc. many formats compress data

—-

  1. wanted to run test to see specific example, reading document but couldn’t (crying) 

  2. BTW, are GPUs at Casablanca? Why a timezone matches it?