06_PixelRNN.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Welcome to assignment #6!\n",
    "\n",
    "Please submit your solution of this notebook in the Whiteboard at the corresponding Assignment entry. We need you to upload the .ipynb-file and the exported .pdf of this notebook.\n",
    "\n",
    "If you have any questions, ask them in either in the tutorials or in the \"Mattermost\" channel. The channel is called SSL_WS_2324, you can join the server using this [Link](https://mattermost.imp.fu-berlin.de/signup_user_complete/?id=h5ssupqokprtpyf4dr7xabiwpc&md=link&sbr=su) and can search for the public channel.\n",
    "\n",
    "\n",
    "This week we want to to implement a PixelRNN on MNIST."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# PapagAI\n",
    "\n",
    "From the second week onwards we started the reflective study.\n",
    "Register on the [PapagAI website](https://www.papag.ai) and write your first reflection about your impressions and challenges in the context of the lectures and tutorials you had this and previous week. The size of reflection can be anywhere bigger than 100 words. You can check out this [YouTube video](https://www.youtube.com/watch?v=QdmZHocZQBk&ab_channel=FernandoRamosL%C3%B3pez) with instructions on how to register, create a reflection and get an ai feedback.\n",
    "\n",
    "Please note, that this task is an obligatory one for this course and make sure each of you does the reflection, not only one person per group."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Please state both names of your group members here:\n",
    "Authors: Jane and John Doe"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Assignment 6: PixelRNN\n",
    "\n",
    "Paper: [https://arxiv.org/pdf/1601.06759.pdf](https://arxiv.org/pdf/1601.06759.pdf) <br>\n",
    "(Probably) Useful Blogpost: [https://towardsdatascience.com/auto-regressive-generative-models-pixelrnn-pixelcnn-32d192911173](https://towardsdatascience.com/auto-regressive-generative-models-pixelrnn-pixelcnn-32d192911173)\n",
    "\n",
    "## Ex. 6.1 Architecture\n",
    "\n",
    "Pixel Recurrent Neural Networks perform a generative task using a recurrent network architecture. Working on images, we usually work with 2 spatial dimensions (+ color channel dimension in case of colored images). The paper presents 2 different versions of LSTM layers to make it work, but you don't have to. You are free to use any other recurrent layer and any library you want. Whatever works best for you.\n",
    "\n",
    "Implement your model architecture here. **(RESULT)**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Code here\n",
    "\n",
    "class yourPixelRNN():\n",
    "    pass"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Ex. 6.2 Training on MNIST\n",
    "\n",
    "Train your model on the [MNIST](https://pytorch.org/vision/main/generated/torchvision.datasets.MNIST.html) dataset. **(RESULT)**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Code here"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Ex. 6.3 Complete some samples\n",
    "\n",
    "Using your trained model, are you able to complete MNIST samples that are partially masked? Show your best results and compare them to the original. **(RESULT)**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Code here"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "uptodate",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}