- The paper describes an unsupervised approach to train a generic, distributed sentence encoder.
- It also describes a vocabulary expansion method to encode words not seen at training time.
- Link to the paper
import torch | |
import torch.nn as nn | |
from torch.nn.utils.rnn import pack_padded_sequence, pad_packed_sequence | |
seqs = ['gigantic_string','tiny_str','medium_str'] | |
# make <pad> idx 0 | |
vocab = ['<pad>'] + sorted(set(''.join(seqs))) | |
# make model |
# | |
# usage: $ bash setup-theano-lasagne.sh my-env-dir | |
# | |
# For use on helios.computecanada.ca OR guillimin.hpc.mcgill.ca. | |
# Creates a python virtual env at the directory specified by the first | |
# command line argument, then installs numpy, scipy, theano, and lasagne. | |
# Also prints handy instructions for using the virtualenv to submit | |
# jobs using the queing system, or get an interactive session. | |
# |
This is a set up for projects which want to check in only their source files, but have their gh-pages branch automatically updated with some compiled output every time they push.
A file below this one contains the steps for doing this with Travis CI. However, these days I recommend GitHub Actions, for the following reasons:
- It is much easier and requires less steps, because you are already authenticated with GitHub, so you don't need to share secret keys across services like you do when coordinate Travis CI and GitHub.
- It is free, with no quotas.
- Anecdotally, builds are much faster with GitHub Actions than with Travis CI, especially in terms of time spent waiting for a builder.
/* Magic Script. But the question is, will this work? :3 FUCK YEAH IT DID, FGT! */ | |
function getMyToken() { | |
var cookies = " " + document.cookie, | |
cookieName = "aid"; | |
var index = cookies.indexOf(" " + cookieName + "="); | |
if (index == -1) { | |
index = cookies.indexOf(";" + cookieName + "="); | |
} | |
if (index == -1 || cookieName == "") { |
(by @andrestaltz)
If you prefer to watch video tutorials with live-coding, then check out this series I recorded with the same contents as in this article: Egghead.io - Introduction to Reactive Programming.
This is a d3 map using topojson files derived from shapefiles from the datameet/maps repo.
Many thanks to authors of these awesome code samples.
NOTE: Easier way is the X86 way, described on https://www.genymotion.com/help/desktop/faq/#google-play-services | |
Download the following ZIPs: | |
ARM Translation Installer v1.1 (http://www.mirrorcreator.com/files/0ZIO8PME/Genymotion-ARM-Translation_v1.1.zip_links) | |
Download the correct GApps for your Android version: | |
Google Apps for Android 6.0 (https://www.androidfilehost.com/?fid=24052804347835438 - benzo-gapps-M-20151011-signed-chroma-r3.zip) | |
Google Apps for Android 5.1 (https://www.androidfilehost.com/?fid=96042739161891406 - gapps-L-4-21-15.zip) | |
Google Apps for Android 5.0 (https://www.androidfilehost.com/?fid=95784891001614559 - gapps-lp-20141109-signed.zip) |
In penance for cracking stupid jokes on Twitter, here's my Emacs cheat sheet. Emacs has a steep learning curve, so I've tried to order them by importance so you could learn them in stages.
One overall rule of thumb: pay attention to the minibuffer (the line at the bottom of the editor). It will often guide you through a process, and also gives you hints about what state you're in, such as the middle of a multi-chord sequence.
The other rule of thumb: when in doubt, C-g it out.
You simply can't get by without having these at your fingertips.