mrdrozdov

import random

def Merge(x, y):
    return ''.join([a for t in zip(x, y) for a in t])

def Peano(x, y):
    interim = ''.join(Merge(format(x, '08b'), format(y, '08b')))
    return int(interim, base=2)

def generate(n=100, ln=30):

mrdrozdov

var pushLeft = slate.operation("push", {
  "direction" : "left",
  "style" : "bar-resize:screenSizeX/2"
});

var pushRight = slate.operation("push", {
  "direction" : "right",
  "style" : "bar-resize:screenSizeX/2"
});

mrdrozdov

"""
Use at own risk.

I use nltk for many purposes, but I needed a faster version of nltk.Tree.fromstring.

This implementation is much faster and works for my use case, but it
has not been tested extensively on complex input.

Original fromstring implementation from nltk is here:
https://github.com/nltk/nltk/blob/develop/nltk/tree.py

mrdrozdov

import os
import hashlib

from allennlp.commands.elmo import ElmoEmbedder
from allennlp.data.token_indexers.elmo_indexer import ELMoCharacterMapper
import numpy as np


def save_elmo_cache(path, vectors):
    np.save(path, vectors)

mrdrozdov

YES Bernard's secret revealed. # His name was an anagram.
NO Man in black pays a visit to Dolores.
YES Host gets repaired. # Bernard
YES There's a flashback. # Bernard has a flashback
? Radiohead's song played.
? The maze is referenced.
NO Maeve uses her powers.
YES A character gets redemption. # Dolores gets redemption on some of the wealthy men.
? "Wake up, Dolores".
? A new host storyline is presented (perhaps by Sizemore).

mrdrozdov

#!/usr/bin/env python

import json
import os
import shutil
import subprocess
import sys
import tempfile

# env

mrdrozdov

https://docs.google.com/presentation/d/1vtsYo064-oO1ZeFuB5uOBHIM5dMt3bUh6qkgFlj_kbU/edit?usp=sharing

mrdrozdov

import os
import collections

path = os.path.expanduser('~/data/conll2003/eng.train')

# Read raw data.
with open(path) as f:
    dataset = []
    example = dict()
    for line in f:

mrdrozdov

import re
import os
import sys
import time
import collections
import json

from tqdm import tqdm

mrdrozdov

# Run once for each size.

In [1]: import torch

In [2]: a = torch.arange(100).view(10, 10)

In [3]: a
Out[3]:
tensor([[ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9],
        [10, 11, 12, 13, 14, 15, 16, 17, 18, 19],