Skip to content

Instantly share code, notes, and snippets.

View faroit's full-sized avatar
🚀
Rocket Science

Fabian-Robert Stöter faroit

🚀
Rocket Science
View GitHub Profile
@faroit
faroit / gruln.py
Created July 23, 2016 15:28 — forked from udibr/gruln.py
Keras GRU with Layer Normalization
from keras.layers import GRU, initializations, K
from collections import OrderedDict
class GRULN(GRU):
'''Gated Recurrent Unit with Layer Normalization
Current impelemtation only works with consume_less = 'gpu' which is already
set.
# Arguments
output_dim: dimension of the internal projections and the final output.
# $ brew install https://raw.github.com/gist/1121425/rubberband.rb
require 'formula'
class Rubberband < Formula
homepage ''
url 'http://code.breakfastquay.com/attachments/download/35/rubberband-1.8.1-gpl-executable-osx-universal.tar.bz2'
version '1.8.1'
sha1 'bceba7dba8cdd773a318ca973ed9c5806be03262'
# -*- coding: utf-8 -*-
"""
Bandlimited versions of scipy.signal.waveforms.
Intent is mathematical perfection over performance;
these use additive synthesis, so they are slow, but exact.
Less ideal methods using BLIT:
Sawtooth can be made by integrating BLIT minus a DC value to prevent integrator wandering off
Square can be made by integrating bipolar BLIT