bzamecnik

"""
Title: Simple MNIST convnet
Author: [fchollet](https://twitter.com/fchollet)
Date created: 2015/06/19
Last modified: 2020/04/21
Description: A simple convnet that achieves ~99% test accuracy on MNIST.


Modified:
- data wrapped as a Sequence

bzamecnik

# In order to make a full repo code review, fork a repo and clone the fork.
# Then use this script to make a new empty base branch merge the current
# main branch into it for the sake of code review. Then open
# a pull request in GitHub: "review" <- "homework".

git checkout --orphan review
git rm -rf .
git commit --allow-empty -m "Create empty branch"
git push --set-upstream origin review

bzamecnik

Two weeks after upgrade to Evernote 10 Mac I found that all my changes are lost!

This script helps to extract the missing notes.

Forums:

• https://discussion.evernote.com/forums/topic/130736-notes-created-on-evernote-mac-1036-do-not-sync/
• https://discussion.evernote.com/forums/topic/130885-missing-notes-after-update-to-v-1037-build-2018/

Evernote was showing that changes are saved but nothing was synchronized to other devices and most importantly upon app restart (or crash) everything returned to the state

bzamecnik

for remote in $(git remote); do
  old_url=$(git remote get-url "$remote")
  new_url=${old_url//git.old.example.com/git.new.example.com}
  if [ "$old_url" != "$new_url" ]; then
    echo "$remote: $old_url -> $new_url"
    git remote set-url "$remote" "$new_url"
  else
    echo "$remote: $old_url (already OK)"
  fi
done

bzamecnik

Se stim smiř! -- Zmikund

           +
         |\|/|
    O    |. ,|
    \    |vvv|
     \  /\/\/|   /_\
 \_/ + \ | /

bzamecnik

#!/bin/bash
# Accept CD Wifi (Czech railways) on the command line (if you hate clicking).
# http://cdwifi.cz
#
# Put the script to ~/bin/cdwifi and make sure that export it's on the path: PATH=$PATH:~/bin
# Note that if you have a custom DNS set (eg. 8.8.8.8 or 1.1.1.1) they don't resolve
# cdwifi.cz. If you don't want to use their DNS set via DHCP for all your traffic
# the workaround here is to query their DNS server at the gateway just
# for this request.
#

bzamecnik

# https://en.wikipedia.org/wiki/Floyd–Steinberg_dithering
from numba import jit
import numpy as np

@jit(nopython=True)
def floyd_steinberg(image):
    # image: np.array of shape (height, width), dtype=float, 0.0-1.0
    # works in-place!
    h, w = image.shape
    for y in range(h):

bzamecnik

import numpy as np
from tensorpack import DataFlow


class CompareData(DataFlow):
    """
    Compares that two DataFlows generate equal data, raises ValueError if not.
    """
    def __init__(self, a, b):
        self.a = a

bzamecnik

My NVIDIA GTX 980 Ti being idle eats 53 W (out of 250 W max). Here's how to put to to a power saving mode at 30 W consumption and keep it prepared to run at higher performance automatically when there's a computation running.

TL;DR

sudo nvidia-smi -pm ENABLED -i 0
sudo nvidia-smi -ac 405,135 -i 0

bzamecnik

# Example of basic MNIST Keras model with tf.Dataset
# More up-to-date version of: https://github.com/keras-team/keras/blob/master/examples/mnist_dataset_api.py
"""
MNIST classification with TensorFlow's Dataset API.

Introduced in TensorFlow 1.3, the Dataset API is now the
standard method for loading data into TensorFlow models.

A Dataset is a sequence of elements, which are themselves
composed of tf.Tensor components. For more details, see: