neubig

"""
DyNet implementation of a sequence labeler (POS taggger).
This is a translation of this tagger in PyTorch: https://gist.github.com/hal3/8c170c4400576eb8d0a8bd94ab231232

Basic architecture:
 - take words
 - run though bidirectional GRU
 - predict labels one word at a time (left to right), using a recurrent neural network "decoder"
The decoder updates hidden state based on:
 - most recent word

rkern

#!/usr/bin/env python
""" PRNG seed sequence implementation for np.random.

The algorithms are derived from Melissa E. O'Neill's C++11 `std::seed_seq`
implementation, as it has a lot of nice properties that we want.

https://gist.github.com/imneme/540829265469e673d045
http://www.pcg-random.org/posts/developing-a-seed_seq-alternative.html

The MIT License (MIT)

ar45

PyAbstractClass
PyArgumentEqualDefault
PyArgumentList
PyAssignmentToLoopOrWithParameter
PyAttributeOutsideInit
PyAugmentAssignment
PyBroadException
PyByteLiteral
PyCallByClass
PyCallingNonCallable

ybj14

# Refer to `https://j-towns.github.io/2017/06/12/A-new-trick.html` for math details.
import torch
from torch import nn
from torch import autograd

def get_jvp(net, x, v):
  '''
  Generate jacobian vector product. Requires x.requires_grad()
  and v.requires_grad().
  Args:

tushortz

def unicodetoascii(text):

    TEXT = (text.
    		replace('\\xe2\\x80\\x99', "'").
            replace('\\xc3\\xa9', 'e').
            replace('\\xe2\\x80\\x90', '-').
            replace('\\xe2\\x80\\x91', '-').
            replace('\\xe2\\x80\\x92', '-').
            replace('\\xe2\\x80\\x93', '-').
            replace('\\xe2\\x80\\x94', '-').

0xabad1dea

Risk Assessment of GitHub Copilot

0xabad1dea, July 2021

this is a rough draft and may be updated with more examples

GitHub was kind enough to grant me swift access to the Copilot test phase despite me @'ing them several hundred times about ICE. I would like to examine it not in terms of productivity, but security. How risky is it to allow an AI to write some or all of your code?

Ultimately, a human being must take responsibility for every line of code that is committed. AI should not be used for "responsibility washing." However, Copilot is a tool, and workers need their tools to be reliable. A carpenter doesn't have to

sbarratt

def get_jacobian(net, x, noutputs):
    x = x.squeeze()
    n = x.size()[0]
    x = x.repeat(noutputs, 1)
    x.requires_grad_(True)
    y = net(x)
    y.backward(torch.eye(noutputs))
    return x.grad.data

bombs-kim

# This instruction is for Unix-like OS users.
# I refered to the following guide when I wrote it.
# https://conda.io/docs/user-guide/tasks/build-packages/recipe.html

# Activate an environment if needed
conda activate myenv

# Install conda-build if you haven't. Ironically, installing conda-build
# includes installing python. This python will be effectively replaced with
# your new python binary so don't worry.

egel

#!/usr/bin/python
# -*- coding: utf-8 -*-

import sublime_plugin
import subprocess
from time import sleep
import sys
 
cl = lambda line: subprocess.Popen(line, shell=True, stdout=subprocess.PIPE).communicate()[0].strip()
log = lambda message: sys.stderr.write("Log: %s\n" % message)

W4ngatang

''' Script for downloading all GLUE data.

Note: for legal reasons, we are unable to host MRPC.
You can either use the version hosted by the SentEval team, which is already tokenized, 
or you can download the original data from (https://download.microsoft.com/download/D/4/6/D46FF87A-F6B9-4252-AA8B-3604ED519838/MSRParaphraseCorpus.msi) and extract the data from it manually.
For Windows users, you can run the .msi file. For Mac and Linux users, consider an external library such as 'cabextract' (see below for an example).
You should then rename and place specific files in a folder (see below for an example).

mkdir MRPC
cabextract MSRParaphraseCorpus.msi -d MRPC