Skip to content

Instantly share code, notes, and snippets.

View sbos's full-sized avatar

Sergey Bartunov sbos

56 followers · 1 following
View GitHub Profile
@sbos
sbos / gist:018955227319ead18529
Created November 10, 2014 06:31
-.13015382 .033825178 -.02594826 -.288166 .04162671 -.06678655 -.22145432 .1749719 .030906504 -.12878266 .08305055 .11668899 -.06319859 -.0777372 .040560413 -.0305699 .17826276 -.10443129 -.00669569 .014138044 -.064837754 -.24928348 .060342744 -.08906554 .080405414 .0045637907 .007919261 -.14719364 .07927967 .011611855 -.0018610754 .08388769 .09033888 .0517999 .020472864 .19127382 .022346925 .06461436 .035737775 .049969383 .006792533 .11398394 -.03975431 -.112482205 -.03677935 .023018813 .044070702 .09208278 -.021259554 .13439591 -.048604295 .058010057 -.051273342 .03115164 .100593045 -.12984459 .10762949 .18540311 -.009627994 .0026174872 -.05229366 -.23992816 -.076821424 .12318988 -.1506341 -.13093792 .0035753904 -.09583123 .067648284 -.09169266 -.09788873 -.1272856 -.1690156 .017811107 -.027340638 .07608332 .10959153 -.027422326 .08745462 -.18536375 .013429028 -.05132068 .04882859 .23418726 -.026919987 .034001417 -.010213713 .035071775 -.038683213 -.092883974 -.056313466 .111097194 -.074547 -.035446037 -.00
@sbos
sbos / cgt_simplification_bug.py
Created October 20, 2015 19:36
import cgt
import cgt.nn as nn
import numpy as np
from scipy.stats import norm
def gaussian_density(x, mu, sigma):
return cgt.exp(-cgt.square(x - mu) / 2 / cgt.square(sigma)) \
/ cgt.sqrt(2 * np.pi) / sigma
var_mu = nn.parameter(np.array(0.5))
@sbos
sbos / cgt_memory.py
Last active October 21, 2015 21:27
import cgt
import cgt.nn as nn
import numpy as np
from scipy.stats import norm
def gaussian_density(x, mu, sigma):
return cgt.exp(-cgt.square(x - mu) / 2 / cgt.square(sigma)) \
/ cgt.sqrt(2 * np.pi) / sigma
var_mu = nn.parameter(np.array(0.5))
@sbos
sbos / random_bug.py
Created October 24, 2015 05:01
import cgt
import cgt.nn as nn
import numpy as np
batch = 5000
mean = 0.
for i in xrange(batch):
mean += (cgt.randn() - mean) / (i+1)
@sbos
sbos / debug_th.py
Created November 26, 2015 13:30
import theano as th
import theano.tensor as T
import lasagne
import numpy as np
from theano.tensor.shared_randomstreams import RandomStreams
from lasagne.nonlinearities import tanh
import matplotlib.pyplot as plt
import sys
# change to True and see what happens
@sbos
sbos / baseline.ipynb
Created December 15, 2015 04:03
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
@sbos
sbos / gist:5859327
Created June 25, 2013 15:18
build-error.log
==> Using Homebrew-provided fortran compiler.
This may be changed by setting the FC environment variable.
==> Cloning https://github.com/JuliaLang/julia.git
git --git-dir /Library/Caches/Homebrew/julia--git/.git status -s
Updating /Library/Caches/Homebrew/julia--git
git config remote.origin.url https://github.com/JuliaLang/julia.git
git config remote.origin.fetch +refs/heads/master:refs/remotes/origin/master
git fetch origin
git checkout -f master
Already on 'master'
@sbos
sbos / adagram_export.jl
Last active October 22, 2016 18:04
function write_word2vec(path::AbstractString, vm::VectorModel, dict::Dictionary)
fout = open(path, "w")
sense_prob = zeros(T(vm))
write(fout, "$(V(vm)) $(T(vm)) $(M(vm))\n")
for v in 1:V(vm)
write(fout, "$(dict.id2word[v])\n")
expected_pi!(sense_prob, vm, v)
for k in 1:T(vm)
if sense_prob[k] < 1e-3 continue end
write(fout, "$k $(sense_prob[k]) ")
@sbos
sbos / HMM.jl
Created November 1, 2013 11:25
Hidden Markov Model in Julia
module HMM
using Distributions
import Distributions.rand
import Distributions.fit
immutable HiddenMarkovModel{TP, K}
theta::Vector{TP}
A::Matrix{Float64}
@sbos
sbos / Н.О.Ж. - Одноклассник
Created July 1, 2018 14:28
Снова учитель что-то
Пытается нам объяснять
Мне случать его неохота
Мне хочется погулять
Я после звонка наплевала
На следующий урок
Сломала окно и сьежала
Слушать тюменский панк-рок