Lyken17

Reading translation models from nbest file...
 Epoch 0...
 Mistake in epoch 0: normal mis : 9116 momentum mis : 9390
 Epoch 1...
 Mistake in epoch 1: normal mis : 4600 momentum mis : 4513
 Epoch 2...
 Mistake in epoch 2: normal mis : 3168 momentum mis : 3021
 Epoch 3...
 Mistake in epoch 3: normal mis : 2330 momentum mis : 2318
 Epoch 4...

Lyken17

class graph():
    def __init__(self, n=4):
        self.n = n
        self.E = [[] for _ in range(n)]
        self.count = 0
        self.color = ['white' for _ in range(self.n)]
        
    def addEdge(self, m, n):
        self.E[m].append(n)
        self.E[n].append(m)

Lyken17

[General]
loglevel = notify
skip-proxy = 127.0.0.1, 192.168.0.0/16, 10.0.0.0/8, 172.16.0.0/12, 100.64.0.0/10, localhost, *.local, ::ffff:0:0:0:0/1, ::ffff:128:0:0:0/1
bypass-tun = 192.168.0.0/16, 10.0.0.0/8, 172.16.0.0/12

# dns-server = 119.29.29.29,223.5.5.5,114.114.115.115
# external-controller-access = PASSWORD@0.0.0.0:6155
# ipv6 = true

// REMEMBER TO CHANGE THE external-controller-access' PASSWORD

Lyken17

#include <iostream>
#include <vector>
#include <chrono>
#include <thread>

using namespace std;

struct A {
    int x, y;
    A() : x(0), y(0){}

Lyken17

# coding: utf-8

# In[127]:

import numpy as np 
from scipy.signal import convolve2d as conv2d
p_label = np.array([[[1, 0, 0], [1, 0, 0], [1, 0, 0], [1, 0, 0], [1, 0, 0], [1, 0, 0]],
                    [[1, 0, 0], [1, 0, 0], [0, 1, 0], [0, 1, 0], [1, 0, 0], [1, 0, 0]],
                    [[1, 0, 0], [1, 0, 0], [0, 1, 0], [0, 1, 0], [1, 0, 0], [1, 0, 0]],

Lyken17

#include <iostream>
#include <vector>
#include <ctime>
using namespace std;


class Dog {
public:
	 void move() {cout << "dog move" << endl;}
};

Lyken17

MXNET is a great deep learning framework, and currently has the best support for distribued trainning. However, its documentation lacks maintaince and I get confused in many API details. Though MXNET community has provided a bunch of examples, I do not find useful examples for cifar 100. Since MXNET is using its own data format rec, I have to convert data to rec before training. data.mxnet.io/data only provides rec files of cifar 10. It takes some time for me to figure out how to generate my own cifar 100 rec file.

First, download python version of cifar 100

Then, dump it into images, thanks to apache/mxnet#235

import os
import sys
import pickle
import csv

Lyken17

Input devices

• Physical

  • mouse, joystick, keyboard

• Logical

  • string, locator, choice

• Event-driven

Lyken17

void del_substr(char *str, const char *substr)
{
    int index = 0, position = 0;
    while (str[index] != 0) {
        if (str[index] == substr[position]) position += 1;
        else position = 0;

        if (position == sizeof(substr) - 1) {
            str[index] = 0;
            return;

Lyken17

import numpy as np
from PIL import Image

import os.path as osp
import glob
import fnmatch, re, os, sys
import yaml
import matplotlib.mlab as mlab
import matplotlib.pyplot as plt
import cv2 as cv