yoavg

A criticism of "On the Dangers of Stochastic Parrots: Can Languae Models be Too Big"

Yoav Goldberg, Jan 23, 2021.

The FAccT paper "On the Dangers of Stochastic Parrots: Can Languae Models be Too Big" by Bender, Gebru, McMillan-Major and Shmitchell has been the center of a controversary recently. The final version is now out, and, owing a lot to this controversary, would undoubtly become very widely read. I read an earlier draft of the paper, and I think that the new and updated final version is much improved in many ways: kudos for the authors for this upgrade. I also agree with and endorse most of the content. This is important stuff, you should read it.

However, I do find some aspects of the paper (and the resulting discourse around it and around technology) to be problematic. These weren't clear to me when initially reading the first draft several months ago, but they became very clear to me now. These points are for the most part

hrocha

MuhsinFatih

I am assuming you are here because like me, you installed a bazillion different python interpreters on mac and the whole thing is a spagetti. Today, I finally fixed my python installation. Whatever I install for python2 or python3 using pip JUST.WORKS.. My god! finally.

What the hell?

Here is what I had messed up, which you also probably did:

• I had too many different python interpreters
• Too many different symlinks which I lost track of
• almost no package I installed with pip worked without a headache
• any attempt to fix using online resources made it worse.

birdbrainiac

/*
basic structure
!wodice Alucard rolls [[7d10]] to use Dominate --sort --title:Fiendish Mind Control --7 --tens
!wodice Mina rolls [[5d10>7]] to resist Dominate --sort --title:Willpower --7 --tens
!wodice Alucard rolls [[7d10>7?{Special?|Yes,!|No,!11}]] to use Dominate --sort --title:Fiendish Mind Control
Structure is
!wodice [text to print out as description] --optional parameters
    --sort or -sort:true (makes it sort, otherwise its unsorted, could also add ascending or descending option)
    --title:some text or --some text (adds a title, otherwise a default title is used)
    --difficulty: 7 or --7 (sets a variable difficulty)

ashokpant

#!/bin/bash

# install CUDA Toolkit v9.0
# instructions from https://developer.nvidia.com/cuda-downloads (linux -> x86_64 -> Ubuntu -> 16.04 -> deb)
CUDA_REPO_PKG="cuda-repo-ubuntu1604-9-0-local_9.0.176-1_amd64-deb"
wget https://developer.nvidia.com/compute/cuda/9.0/Prod/local_installers/${CUDA_REPO_PKG}
sudo dpkg -i ${CUDA_REPO_PKG}
sudo apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub
sudo apt-get update
sudo apt-get -y install cuda-9-0

RWOverdijk

/**
 * Name model.
 *
 * This model is used to fetch names, belonging to lastNames.
 */
;
(function setupNameModel() {

  // The module
  var module = angular.module('core.models', ['core.providers']);

jvns

A lot of these are outright stolen from Edward O'Campo-Gooding's list of questions. I really like his list.

I'm having some trouble paring this down to a manageable list of questions -- I realistically want to know all of these things before starting to work at a company, but it's a lot to ask all at once. My current game plan is to pick 6 before an interview and ask those.

I'd love comments and suggestions about any of these.

I've found questions like "do you have smart people? Can I learn a lot at your company?" to be basically totally useless -- everybody will say "yeah, definitely!" and it's hard to learn anything from them. So I'm trying to make all of these questions pretty concrete -- if a team doesn't have an issue tracker, they don't have an issue tracker.

I'm also mostly not asking about principles, but the way things are -- not "do you think code review is important?", but "Does all code get reviewed?".

SinisterMinister

/**
 * This is a WIP of potential usage of the ORM in a NodeJS implementation. Let's try implementing
 * this as a module for other parts of the app to use. Less instantiation for later
 */
 
var ORM = require('awesomeorm-base'),
    repository = require('awesomeorm-repository-rest')('http://rest.example.com/schema'),
    dataParser = require('awesomeorm-data-jsonapi'),
    modelBuilder = require('awesomeorm-model-encasulated');
 

OpenSorceress

<!doctype html>
<!--[if lt IE 7 ]> <html class="no-js ie6" lang="en"> <![endif]-->
<!--[if IE 7 ]>    <html class="no-js ie7" lang="en"> <![endif]-->
<!--[if IE 8 ]>    <html class="no-js ie8" lang="en"> <![endif]-->
<!--[if (gte IE 9)|!(IE)]><!--> <html class="no-js" lang="en" xmlns:fb="http://www.facebook.com/2008/fbml" xmlns:og="http://ogp.me/ns#"> <!--<![endif]-->
  <head>
    <meta charset="utf-8">
    <meta http-equiv="X-UA-Compatible" content="IE=edge,chrome=1">
    <meta name='description'  content='{MetaDescription}'>
    <meta name="viewport" content="width=device-width, initial-scale=1.0">

SinisterMinister

Programming Test

Write a view class that can fetch data in multiple ways (file, web, etc) in different formats (JSON, CSV, etc) and render them in multiple ways (table, unordered lists). Bonus points if you use the following interfaces:

<?php

/**
 * Describes a data interface for use with a renderer to render an output
 */