Skip to content

Latest commit

 

History

History
33 lines (21 loc) · 1.02 KB

exercise.md

File metadata and controls

33 lines (21 loc) · 1.02 KB

Week 3 activities

Watch the Lesson 3 - http://course.fast.ai/lessons/lesson3.html from Start to 1:59:37

Participate in following kaggle competition and submit your

Quickdraw doodle recognition

  1. Create a new notebook.
  2. Download data from here https://www.kaggle.com/c/quickdraw-doodle-recognition/data

Steps to Train a world-class image classifier Quickdraw doodle recognition

  1. Convert csv format image data in png format usable by FastAI
  2. Train and fiddle with various parameters e.g. CNN architechures , Freeze - unfreeze , learning rate etc.
  3. Convert result in format asked by Kaggle
  4. Submit your result on Kaggle

How to download the kaggle data

  1. Download the kaggle cli project on the paperspace machine ''' git clone https://github.com/Kaggle/kaggle-api '''
  2. Follow the installations steps mentioned on the github project page.
  3. Download the data using following commands ''' kaggle competitions download -c -f '''