moskomule

以下をuBlacklistに加える

https://gist.githubusercontent.com/moskomule/3bddfc5deb5845fccc0d7c2519c600ce/raw/uBlacklist.txt

moskomule

moskomule

raytracer

A simple Python raytracer that supports spheres with configurable "material" properties (base color and a bunch of light coefficients). To generate a raytraced image of the pre-defined scene, run: python raytracer.py and open image.ppm with a PPM-compatible viewer (eog works fine on Linux):

acknowledgements

I found the following resources extremely helpful:

moskomule

import torch.utils.data as data
import os
import re
import torch
import tarfile
from PIL import Image


IMG_EXTENSIONS = ['.png', '.jpg', '.jpeg']

moskomule

cat results/*/*.json | grep -o -E 'args": "[^"]+' | grep -o -E '\-\-loss.*' >f1.txt
cat results/*/*.json | grep -o -E 'args": "[^"]+' | grep -o -E '\-\-dataset[^\-]+' > f2.txt
cat results/*/*.json | grep -o -E 'miou_test": \[\[[^"]+' | grep -o -E '29, [^]]+' > f3.txt
paste f1.txt f2.txt f3.txt  | sed -e 's/--loss/|/g' | sed -e 's/--batch_size/|/g' | sed -e 's/29,/|/g'| sed -e 's/--dataset/|/g' | sort

#
# | ce | 1        | sbd   | 0.7214421629905701
# | ce | 1        | voc   | 0.6946691870689392
# | ce | 2        | sbd   | 0.7422053217887878
# | ce | 2        | voc   | 0.7283321619033813

moskomule

utility for downloading webvision2018 dataset

Requires gnu parallel and aria2c. Both can be downloaded via brew

moskomule

moskomule

def Rop(y, x, v):
  """Computes an Rop.
  
  Arguments:
    y (Variable): output of differentiated function
    x (Variable): differentiated input
    v (Variable): vector to be multiplied with Jacobian from the right
  """
  w = torch.ones_like(y, requires_grad=True)
  return torch.autograd.grad(torch.autograd.grad(y, x, w), w, v)

moskomule

wget -O grive https://docs.google.com/uc?id=0B3X9GlR6EmbnQ0FtZmJJUXEyRTA
chmod 755 grive # not sure if it's necessary
./gdrive upload [--recursive] NAME

moskomule

#!/bin/bash

# download and unzip dataset
#wget http://cs231n.stanford.edu/tiny-imagenet-200.zip
unzip tiny-imagenet-200.zip

current="$(pwd)/tiny-imagenet-200"

# training data
cd $current/train