{"metadata": {"kernelspec": {"language": "python", "display_name": "Python 3", "name": "python3"}, "language_info": {"nbconvert_exporter": "python", "file_extension": ".py", "version": "3.6.1", "codemirror_mode": {"version": 3, "name": "ipython"}, "pygments_lexer": "ipython3", "name": "python", "mimetype": "text/x-python"}}, "cells": [{"metadata": {"_cell_guid": "550cb39b-873f-47c7-9dc5-40115b089c0e", "_uuid": "b0aa04aa91d6bea46b6fa50e348b47dea8acc671"}, "source": ["## Intro\n", "\n", "Thanks to the Keras Kernel by Peter Giannakopoulos [here](https://www.kaggle.com/c/carvana-image-masking-challenge/discussion/37523), and Heng CherKeng Pytorch Kernel [here](https://www.kaggle.com/c/carvana-image-masking-challenge/discussion/37208). Both are wonderful approaches to the carvana challenge and I was very inspired from their code to make mine. \n", "\n", "\n", "## Why creating another kernel then?\n", "\n", "I found Heng CherKeng kind of an experimental playground and I wanted to have a clear and straightforward code written for Pytorch that I can refer to later in time.\n", "The goal of this kernel is not to burst public/private lb score but rather to understand how Unets works using Pytorch.\n", "It has a clear and well documented code and takes the learning path approach. Of course I'll add more and more improvements to it to eventually reach the top results (on 31-08-2017 it reaches 0.987 on public LB). Anyone who's willing to help is welcome to open PRs on the github repo. I engage myself to review and merge them.\n", "\n", "Don't hesitate to upvote the kernel if you found it useful. My motivation comes from it.\n", "\n", "**Here is the [link to the github repository](https://github.com/EKami/carvana-challenge)**"], "cell_type": "markdown"}], "nbformat_minor": 1, "nbformat": 4}