odashi

#!/usr/bin/python3

import math
import random
import sys
from argparse import ArgumentParser
from collections import defaultdict
from util.functions import trace

def parse_args():

lepture

#!/usr/bin/env python
# -*- coding: utf-8 -*-
#
# Copyright (c) 2012, lepture.com
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions
# are met:
#
#    * Redistributions of source code must retain the above copyright

MajorTal

from numpy.random import choice as random_choice, randint as random_randint, rand
MAX_INPUT_LEN = 40
AMOUNT_OF_NOISE = 0.2 / MAX_INPUT_LEN
CHARS = list("abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ .")

def add_noise_to_string(a_string, amount_of_noise):
    """Add some artificial spelling mistakes to the string"""
    if rand() < amount_of_noise * len(a_string):
        # Replace a character with a random character
        random_char_position = random_randint(len(a_string))

odashi

# usage (single sentence):
#   ref = ['This', 'is', 'a', 'pen', '.']
#   hyp = ['There', 'is', 'a', 'pen', '.']
#   stats = get_bleu_stats(ref, hyp)
#   bleu = calculate_bleu(stats) # => 0.668740
#
# usage (multiple sentences):
#   stats = defaultdict(int)
#   for ref, hyp in zip(refs, hyps):
#     for k, v in get_bleu_stats(ref, hyp).items():

basaundi

#!/usr/bin/env python
# Ander Martinez Sanchez

from __future__ import division, print_function
from math import exp, log
from collections import Counter


def ngram_count(words, n):
    if n <= len(words):

odashi

#!/usr/bin/python3

import datetime
import sys
import math
import numpy as np
from argparse import ArgumentParser
from collections import defaultdict

from chainer import FunctionSet, Variable, functions, optimizers

mangecoeur

A "virtualenv activate" for Anaconda environments

I've been using the Anaconda python package from continuum.io recently and found it to be a good way to get all the complex compiled libs you need for a scientific python environment. Even better, their conda tool lets you create environments much like virtualenv, but without having to re-compile stuff like numpy, which gets old very very quickly with virtualenv and can be a nightmare to get correctly set up on OSX.

The only thing missing was an easy way to switch environments - their docs suggest running python executables from the install folder, which I find a bit of a pain. Coincidentally I came across this article - Virtualenv's bin/activate is Doing It Wrong - which desribes a simple way to launch a sub-shell with certain environment variables set. Now simple was the key word for me since my bash-fu isn't very strong, but I managed to come up with the script below. Put this in a text file called conda-work

fperez

mbollmann

# Source:
# https://github.com/farizrahman4u/seq2seq/blob/master/seq2seq/layers/state_transfer_lstm.py

from keras import backend as K
from keras.layers.recurrent import LSTM

class StateTransferLSTM(LSTM):
    """LSTM with the ability to transfer its hidden state.

    This layer behaves just like an LSTM, except that it can transfer (or

entron

'''This scripts implements Kim's paper "Convolutional Neural Networks for Sentence Classification"
with a very small embedding size (20) than the commonly used values (100 - 300) as it gives better
result with much less parameters.

Run on GPU: THEANO_FLAGS=mode=FAST_RUN,device=gpu,floatX=float32 python imdb_cnn.py

Get to 0.853 test accuracy after 5 epochs. 13s/epoch on Nvidia GTX980 GPU.
'''

from __future__ import print_function